1 /*
2  * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
3  *
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.
7  *
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.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/stat.h>
21 #include <sys/wait.h>
22 
23 #include <limits.h>
24 #include <stdlib.h>
25 #include <stdio.h>
26 #include <sha1.h>
27 #include <string.h>
28 #include <zlib.h>
29 #include <errno.h>
30 #include <libgen.h>
31 #include <stdint.h>
32 #include <imsg.h>
33 
34 #include "got_error.h"
35 #include "got_reference.h"
36 #include "got_repository.h"
37 #include "got_worktree.h"
38 #include "got_object.h"
39 
40 #include "got_lib_path.h"
41 #include "got_lib_delta.h"
42 #include "got_lib_inflate.h"
43 #include "got_lib_object.h"
44 #include "got_lib_pack.h"
45 #include "got_lib_repository.h"
46 #include "got_lib_worktree.h"
47 #include "got_lib_object_idcache.h"
48 #include "got_lib_privsep.h"
49 
50 #ifndef nitems
51 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
52 #endif
53 
54 #define GOT_GIT_DIR	".git"
55 
56 /* Mandatory files and directories inside the git directory. */
57 #define GOT_OBJECTS_DIR		"objects"
58 #define GOT_REFS_DIR		"refs"
59 #define GOT_HEAD_FILE		"HEAD"
60 
61 /* Other files and directories inside the git directory. */
62 #define GOT_FETCH_HEAD_FILE	"FETCH_HEAD"
63 #define GOT_ORIG_HEAD_FILE	"ORIG_HEAD"
64 #define GOT_OBJECTS_PACK_DIR	"objects/pack"
65 
66 char *
67 got_repo_get_path(struct got_repository *repo)
68 {
69 	return strdup(repo->path);
70 }
71 
72 char *
73 got_repo_get_path_git_dir(struct got_repository *repo)
74 {
75 	return strdup(repo->path_git_dir);
76 }
77 
78 int
79 got_repo_is_bare(struct got_repository *repo)
80 {
81 	return (strcmp(repo->path, repo->path_git_dir) == 0);
82 }
83 
84 static char *
85 get_path_git_child(struct got_repository *repo, const char *basename)
86 {
87 	char *path_child;
88 	
89 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
90 	    basename) == -1)
91 		return NULL;
92 
93 	return path_child;
94 }
95 
96 char *
97 got_repo_get_path_objects(struct got_repository *repo)
98 {
99 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
100 }
101 
102 char *
103 got_repo_get_path_objects_pack(struct got_repository *repo)
104 {
105 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
106 }
107 
108 char *
109 got_repo_get_path_refs(struct got_repository *repo)
110 {
111 	return get_path_git_child(repo, GOT_REFS_DIR);
112 }
113 
114 static char *
115 get_path_head(struct got_repository *repo)
116 {
117 	return get_path_git_child(repo, GOT_HEAD_FILE);
118 }
119 
120 static int
121 is_git_repo(struct got_repository *repo)
122 {
123 	char *path_git = got_repo_get_path_git_dir(repo);
124 	char *path_objects = got_repo_get_path_objects(repo);
125 	char *path_refs = got_repo_get_path_refs(repo);
126 	char *path_head = get_path_head(repo);
127 	int ret = 0;
128 	struct stat sb;
129 	struct got_reference *head_ref;
130 
131 	if (lstat(path_git, &sb) == -1)
132 		goto done;
133 	if (!S_ISDIR(sb.st_mode))
134 		goto done;
135 
136 	if (lstat(path_objects, &sb) == -1)
137 		goto done;
138 	if (!S_ISDIR(sb.st_mode))
139 		goto done;
140 
141 	if (lstat(path_refs, &sb) == -1)
142 		goto done;
143 	if (!S_ISDIR(sb.st_mode))
144 		goto done;
145 
146 	if (lstat(path_head, &sb) == -1)
147 		goto done;
148 	if (!S_ISREG(sb.st_mode))
149 		goto done;
150 
151 	/* Check if the HEAD reference can be opened. */
152 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD) != NULL)
153 		goto done;
154 	got_ref_close(head_ref);
155 
156 	ret = 1;
157 done:
158 	free(path_git);
159 	free(path_objects);
160 	free(path_refs);
161 	free(path_head);
162 	return ret;
163 
164 }
165 
166 #ifndef GOT_NO_OBJ_CACHE
167 static const struct got_error *
168 cache_add(struct got_object_cache *cache, struct got_object_id *id, void *item)
169 {
170 	const struct got_error *err = NULL;
171 	struct got_object_cache_entry *ce;
172 	int nelem;
173 
174 	nelem = got_object_idcache_num_elements(cache->idcache);
175 	if (nelem >= cache->size) {
176 		err = got_object_idcache_remove_least_used((void **)&ce,
177 		    cache->idcache);
178 		if (err)
179 			return err;
180 		switch (cache->type) {
181 		case GOT_OBJECT_CACHE_TYPE_OBJ:
182 			got_object_close(ce->data.obj);
183 			break;
184 		case GOT_OBJECT_CACHE_TYPE_TREE:
185 			got_object_tree_close(ce->data.tree);
186 			break;
187 		case GOT_OBJECT_CACHE_TYPE_COMMIT:
188 			got_object_commit_close(ce->data.commit);
189 			break;
190 		}
191 		free(ce);
192 	}
193 
194 	ce = calloc(1, sizeof(*ce));
195 	if (ce == NULL)
196 		return got_error_from_errno();
197 	memcpy(&ce->id, id, sizeof(ce->id));
198 	switch (cache->type) {
199 	case GOT_OBJECT_CACHE_TYPE_OBJ:
200 		ce->data.obj = (struct got_object *)item;
201 		break;
202 	case GOT_OBJECT_CACHE_TYPE_TREE:
203 		ce->data.tree = (struct got_tree_object *)item;
204 		break;
205 	case GOT_OBJECT_CACHE_TYPE_COMMIT:
206 		ce->data.commit = (struct got_commit_object *)item;
207 		break;
208 	}
209 
210 	err = got_object_idcache_add(cache->idcache, id, ce);
211 	if (err) {
212 		if (err->code == GOT_ERR_OBJ_EXISTS) {
213 			free(ce);
214 			err = NULL;
215 		}
216 	}
217 	return err;
218 }
219 #endif
220 
221 const struct got_error *
222 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
223     struct got_object *obj)
224 {
225 #ifndef GOT_NO_OBJ_CACHE
226 	const struct got_error *err = NULL;
227 	err = cache_add(&repo->objcache, id, obj);
228 	if (err)
229 		return err;
230 	obj->refcnt++;
231 #endif
232 	return NULL;
233 }
234 
235 struct got_object *
236 got_repo_get_cached_object(struct got_repository *repo,
237     struct got_object_id *id)
238 {
239 	struct got_object_cache_entry *ce;
240 
241 	ce = got_object_idcache_get(repo->objcache.idcache, id);
242 	if (ce) {
243 		repo->objcache.cache_hit++;
244 		return ce->data.obj;
245 	}
246 
247 	repo->objcache.cache_miss++;
248 	return NULL;
249 }
250 
251 const struct got_error *
252 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
253     struct got_tree_object *tree)
254 {
255 #ifndef GOT_NO_OBJ_CACHE
256 	const struct got_error *err = NULL;
257 	err = cache_add(&repo->treecache, id, tree);
258 	if (err)
259 		return err;
260 	tree->refcnt++;
261 #endif
262 	return NULL;
263 }
264 
265 struct got_tree_object *
266 got_repo_get_cached_tree(struct got_repository *repo,
267     struct got_object_id *id)
268 {
269 	struct got_object_cache_entry *ce;
270 
271 	ce = got_object_idcache_get(repo->treecache.idcache, id);
272 	if (ce) {
273 		repo->treecache.cache_hit++;
274 		return ce->data.tree;
275 	}
276 
277 	repo->treecache.cache_miss++;
278 	return NULL;
279 }
280 
281 const struct got_error *
282 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
283     struct got_commit_object *commit)
284 {
285 #ifndef GOT_NO_OBJ_CACHE
286 	const struct got_error *err = NULL;
287 	err = cache_add(&repo->commitcache, id, commit);
288 	if (err)
289 		return err;
290 
291 	commit->refcnt++;
292 #endif
293 	return NULL;
294 }
295 
296 struct got_commit_object *
297 got_repo_get_cached_commit(struct got_repository *repo,
298     struct got_object_id *id)
299 {
300 	struct got_object_cache_entry *ce;
301 
302 	ce = got_object_idcache_get(repo->commitcache.idcache, id);
303 	if (ce) {
304 		repo->commitcache.cache_hit++;
305 		return ce->data.commit;
306 	}
307 
308 	repo->commitcache.cache_miss++;
309 	return NULL;
310 }
311 
312 const struct got_error *
313 open_repo(struct got_repository *repo, const char *path)
314 {
315 	const struct got_error *err = NULL;
316 	struct got_worktree *worktree = NULL;
317 
318 	/* bare git repository? */
319 	repo->path_git_dir = strdup(path);
320 	if (repo->path_git_dir == NULL) {
321 		err = got_error_from_errno();
322 		goto done;
323 	}
324 	if (is_git_repo(repo)) {
325 		repo->path = strdup(repo->path_git_dir);
326 		if (repo->path == NULL) {
327 			err = got_error_from_errno();
328 			goto done;
329 		}
330 		return NULL;
331 	}
332 
333 	/* git repository with working tree? */
334 	free(repo->path_git_dir);
335 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
336 		err = got_error_from_errno();
337 		goto done;
338 	}
339 	if (is_git_repo(repo)) {
340 		repo->path = strdup(path);
341 		if (repo->path == NULL) {
342 			err = got_error_from_errno();
343 			goto done;
344 		}
345 		return NULL;
346 	}
347 
348 	/* got work tree checked out from bare git repository? */
349 	free(repo->path_git_dir);
350 	repo->path_git_dir = NULL;
351 	err = got_worktree_open(&worktree, path);
352 	if (err) {
353 		if (err->code == GOT_ERR_ERRNO && errno == ENOENT)
354 			err = got_error(GOT_ERR_NOT_GIT_REPO);
355 		goto done;
356 	}
357 	repo->path_git_dir = strdup(worktree->repo_path);
358 	if (repo->path_git_dir == NULL) {
359 		err = got_error_from_errno();
360 		goto done;
361 	}
362 
363 	/* got work tree checked out from git repository with working tree? */
364 	if (!is_git_repo(repo)) {
365 		free(repo->path_git_dir);
366 		if (asprintf(&repo->path_git_dir, "%s/%s", worktree->repo_path,
367 		    GOT_GIT_DIR) == -1) {
368 			err = got_error_from_errno();
369 			repo->path_git_dir = NULL;
370 			goto done;
371 		}
372 		if (!is_git_repo(repo)) {
373 			err = got_error(GOT_ERR_NOT_GIT_REPO);
374 			goto done;
375 		}
376 		repo->path = strdup(worktree->repo_path);
377 		if (repo->path == NULL) {
378 			err = got_error_from_errno();
379 			goto done;
380 		}
381 	} else {
382 		repo->path = strdup(repo->path_git_dir);
383 		if (repo->path == NULL) {
384 			err = got_error_from_errno();
385 			goto done;
386 		}
387 	}
388 done:
389 	if (worktree)
390 		got_worktree_close(worktree);
391 	return err;
392 }
393 
394 const struct got_error *
395 got_repo_open(struct got_repository **repop, const char *path)
396 {
397 	struct got_repository *repo = NULL;
398 	const struct got_error *err = NULL;
399 	char *abspath, *normpath = NULL;
400 	int i, tried_root = 0;
401 
402 	*repop = NULL;
403 
404 	if (got_path_is_absolute(path))
405 		abspath = strdup(path);
406 	else
407 		abspath = got_path_get_absolute(path);
408 	if (abspath == NULL)
409 		return got_error(GOT_ERR_BAD_PATH);
410 
411 	repo = calloc(1, sizeof(*repo));
412 	if (repo == NULL) {
413 		err = got_error_from_errno();
414 		goto done;
415 	}
416 
417 	for (i = 0; i < nitems(repo->privsep_children); i++) {
418 		memset(&repo->privsep_children[i], 0,
419 		    sizeof(repo->privsep_children[0]));
420 		repo->privsep_children[i].imsg_fd = -1;
421 	}
422 
423 	repo->objcache.type = GOT_OBJECT_CACHE_TYPE_OBJ;
424 	repo->objcache.size = GOT_OBJECT_CACHE_SIZE_OBJ;
425 	repo->objcache.idcache = got_object_idcache_alloc(repo->objcache.size);
426 	if (repo->objcache.idcache == NULL) {
427 		err = got_error_from_errno();
428 		goto done;
429 	}
430 
431 	repo->treecache.type = GOT_OBJECT_CACHE_TYPE_TREE;
432 	repo->treecache.size = GOT_OBJECT_CACHE_SIZE_TREE;
433 	repo->treecache.idcache =
434 	    got_object_idcache_alloc(repo->treecache.size);
435 	if (repo->treecache.idcache == NULL) {
436 		err = got_error_from_errno();
437 		goto done;
438 	}
439 
440 	repo->commitcache.type = GOT_OBJECT_CACHE_TYPE_COMMIT;
441 	repo->commitcache.size = GOT_OBJECT_CACHE_SIZE_COMMIT;
442 	repo->commitcache.idcache =
443 	    got_object_idcache_alloc(repo->commitcache.size);
444 	if (repo->commitcache.idcache == NULL) {
445 		err = got_error_from_errno();
446 		goto done;
447 	}
448 
449 	normpath = got_path_normalize(abspath);
450 	if (normpath == NULL) {
451 		err = got_error(GOT_ERR_BAD_PATH);
452 		goto done;
453 	}
454 
455 	path = normpath;
456 	do {
457 		err = open_repo(repo, path);
458 		if (err == NULL)
459 			break;
460 		if (err->code != GOT_ERR_NOT_GIT_REPO)
461 			break;
462 		if (path[0] == '/' && path[1] == '\0') {
463 			if (tried_root) {
464 				err = got_error(GOT_ERR_NOT_GIT_REPO);
465 				break;
466 			}
467 			tried_root = 1;
468 		}
469 		path = dirname(path);
470 		if (path == NULL)
471 			err = got_error_from_errno();
472 	} while (path);
473 done:
474 	if (err)
475 		err = got_repo_close(repo);
476 	else
477 		*repop = repo;
478 	free(abspath);
479 	free(normpath);
480 	return err;
481 }
482 
483 #if 0
484 static void
485 print_cache_stats(struct got_object_cache *cache, const char *name)
486 {
487 	fprintf(stderr, "%s cache: %d elements, %d hits, %d missed\n",
488 	    name, got_object_idcache_num_elements(cache->idcache),
489 	    cache->cache_hit, cache->cache_miss);
490 }
491 
492 void check_refcount(struct got_object_id *id, void *data, void *arg)
493 {
494 	struct got_object_cache *cache = arg;
495 	struct got_object_cache_entry *ce = data;
496 	struct got_object *obj;
497 	struct got_tree_object *tree;
498 	struct got_commit_object *commit;
499 	char *id_str;
500 
501 	if (got_object_id_str(&id_str, id) != NULL)
502 		return;
503 
504 	switch (cache->type) {
505 	case GOT_OBJECT_CACHE_TYPE_OBJ:
506 		obj = ce->data.obj;
507 		if (obj->refcnt == 1)
508 			break;
509 		fprintf(stderr, "object %s has %d unclaimed references\n",
510 		    id_str, obj->refcnt - 1);
511 		break;
512 	case GOT_OBJECT_CACHE_TYPE_TREE:
513 		tree = ce->data.tree;
514 		if (tree->refcnt == 1)
515 			break;
516 		fprintf(stderr, "tree %s has %d unclaimed references\n",
517 		    id_str, tree->refcnt - 1);
518 		break;
519 	case GOT_OBJECT_CACHE_TYPE_COMMIT:
520 		commit = ce->data.commit;
521 		if (commit->refcnt == 1)
522 			break;
523 		fprintf(stderr, "commit %s has %d unclaimed references\n",
524 		    id_str, commit->refcnt);
525 		break;
526 	}
527 	free(id_str);
528 }
529 #endif
530 
531 static const struct got_error *
532 wait_for_child(pid_t pid)
533 {
534 	int child_status;
535 
536 	waitpid(pid, &child_status, 0);
537 
538 	if (!WIFEXITED(child_status))
539 		return got_error(GOT_ERR_PRIVSEP_DIED);
540 
541 	if (WEXITSTATUS(child_status) != 0)
542 		return got_error(GOT_ERR_PRIVSEP_EXIT);
543 
544 	return NULL;
545 }
546 
547 const struct got_error *
548 got_repo_close(struct got_repository *repo)
549 {
550 	const struct got_error *err = NULL, *child_err;
551 	int i;
552 
553 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
554 		if (repo->packidx_cache[i] == NULL)
555 			break;
556 		got_packidx_close(repo->packidx_cache[i]);
557 	}
558 
559 	for (i = 0; i < nitems(repo->packs); i++) {
560 		if (repo->packs[i].path_packfile == NULL)
561 			break;
562 		got_pack_close(&repo->packs[i]);
563 	}
564 
565 	free(repo->path);
566 	free(repo->path_git_dir);
567 
568 #if 0
569 	print_cache_stats(&repo->objcache, "object");
570 	print_cache_stats(&repo->treecache, "tree");
571 	print_cache_stats(&repo->commitcache, "commit");
572 	got_object_idcache_for_each(repo->objcache.idcache, check_refcount,
573 	    &repo->objcache);
574 	got_object_idcache_for_each(repo->treecache.idcache, check_refcount,
575 	    &repo->treecache);
576 	got_object_idcache_for_each(repo->commitcache.idcache, check_refcount,
577 	    &repo->commitcache);
578 #endif
579 
580 	if (repo->objcache.idcache)
581 		got_object_idcache_free(repo->objcache.idcache);
582 	if (repo->treecache.idcache)
583 		got_object_idcache_free(repo->treecache.idcache);
584 	if (repo->commitcache.idcache)
585 		got_object_idcache_free(repo->commitcache.idcache);
586 
587 	for (i = 0; i < nitems(repo->privsep_children); i++) {
588 		if (repo->privsep_children[i].imsg_fd == -1)
589 			continue;
590 		imsg_clear(repo->privsep_children[i].ibuf);
591 		free(repo->privsep_children[i].ibuf);
592 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
593 		child_err = wait_for_child(repo->privsep_children[i].pid);
594 		if (child_err && err == NULL)
595 			err = child_err;
596 		close(repo->privsep_children[i].imsg_fd);
597 	}
598 	free(repo);
599 
600 	return err;
601 }
602 
603 const struct got_error *
604 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
605     const char *input_path)
606 {
607 	const struct got_error *err = NULL;
608 	char *repo_abspath = NULL, *cwd = NULL;
609 	struct stat sb;
610 	size_t repolen, cwdlen, len;
611 	char *canonpath, *path;
612 
613 	*in_repo_path = NULL;
614 
615 	cwd = getcwd(NULL, 0);
616 	if (cwd == NULL)
617 		return got_error_from_errno();
618 
619 	canonpath = strdup(input_path);
620 	if (canonpath == NULL) {
621 		err = got_error_from_errno();
622 		goto done;
623 	}
624 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
625 	if (err)
626 		goto done;
627 
628 	repo_abspath = got_repo_get_path(repo);
629 	if (repo_abspath == NULL) {
630 		err = got_error_from_errno();
631 		goto done;
632 	}
633 
634 	/* TODO: Call "get in-repository path of work-tree node" API. */
635 
636 	if (lstat(canonpath, &sb) != 0) {
637 		if (errno != ENOENT) {
638 			err = got_error_from_errno();
639 			goto done;
640 		}
641 		/*
642 		 * Path is not on disk.
643 		 * Assume it is already relative to repository root.
644 		 */
645 		path = strdup(canonpath);
646 	} else {
647 		int is_repo_child = 0, is_cwd_child = 0;
648 
649 		path = realpath(canonpath, NULL);
650 		if (path == NULL) {
651 			err = got_error_from_errno();
652 			goto done;
653 		}
654 
655 		repolen = strlen(repo_abspath);
656 		cwdlen = strlen(cwd);
657 		len = strlen(path);
658 
659 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
660 			is_repo_child = 1;
661 		if (len > cwdlen && strncmp(path, cwd, cwdlen) == 0)
662 			is_cwd_child = 1;
663 
664 		if (strcmp(path, repo_abspath) == 0) {
665 			free(path);
666 			path = strdup("");
667 			if (path == NULL) {
668 				err = got_error_from_errno();
669 				goto done;
670 			}
671 		} else if (is_repo_child && is_cwd_child) {
672 			char *child;
673 			/* TODO: Is path inside a got worktree? */
674 			/* Strip common prefix with repository path. */
675 			err = got_path_skip_common_ancestor(&child,
676 			    repo_abspath, path);
677 			if (err)
678 				goto done;
679 			free(path);
680 			path = child;
681 		} else if (is_repo_child) {
682 			/* Matched an on-disk path inside repository. */
683 			if (got_repo_is_bare(repo)) {
684 				/*
685 				 * Matched an on-disk path inside repository
686 				 * database. Treat as repository-relative.
687 				 */
688 			} else {
689 				char *child;
690 				/* Strip common prefix with repository path. */
691 				err = got_path_skip_common_ancestor(&child,
692 				    repo_abspath, path);
693 				if (err)
694 					goto done;
695 				free(path);
696 				path = child;
697 			}
698 		} else if (is_cwd_child) {
699 			char *child;
700 			/* TODO: Is path inside a got worktree? */
701 			/* Strip common prefix with cwd. */
702 			err = got_path_skip_common_ancestor(&child, cwd,
703 			    path);
704 			if (err)
705 				goto done;
706 			free(path);
707 			path = child;
708 		} else {
709 			/*
710 			 * Matched unrelated on-disk path.
711 			 * Treat it as repository-relative.
712 			 */
713 		}
714 	}
715 
716 	/* Make in-repository path absolute */
717 	if (path[0] != '/') {
718 		char *abspath;
719 		if (asprintf(&abspath, "/%s", path) == -1) {
720 			err = got_error_from_errno();
721 			goto done;
722 		}
723 		free(path);
724 		path = abspath;
725 	}
726 
727 done:
728 	free(repo_abspath);
729 	free(cwd);
730 	free(canonpath);
731 	if (err)
732 		free(path);
733 	else
734 		*in_repo_path = path;
735 	return err;
736 }