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 		repo->privsep_children[i].imsg_fd = -1;
419 		repo->privsep_children[i].pid = 0;
420 	}
421 
422 	repo->objcache.type = GOT_OBJECT_CACHE_TYPE_OBJ;
423 	repo->objcache.size = GOT_OBJECT_CACHE_SIZE_OBJ;
424 	repo->objcache.idcache = got_object_idcache_alloc(repo->objcache.size);
425 	if (repo->objcache.idcache == NULL) {
426 		err = got_error_from_errno();
427 		goto done;
428 	}
429 
430 	repo->treecache.type = GOT_OBJECT_CACHE_TYPE_TREE;
431 	repo->treecache.size = GOT_OBJECT_CACHE_SIZE_TREE;
432 	repo->treecache.idcache =
433 	    got_object_idcache_alloc(repo->treecache.size);
434 	if (repo->treecache.idcache == NULL) {
435 		err = got_error_from_errno();
436 		goto done;
437 	}
438 
439 	repo->commitcache.type = GOT_OBJECT_CACHE_TYPE_COMMIT;
440 	repo->commitcache.size = GOT_OBJECT_CACHE_SIZE_COMMIT;
441 	repo->commitcache.idcache =
442 	    got_object_idcache_alloc(repo->commitcache.size);
443 	if (repo->commitcache.idcache == NULL) {
444 		err = got_error_from_errno();
445 		goto done;
446 	}
447 
448 	normpath = got_path_normalize(abspath);
449 	if (normpath == NULL) {
450 		err = got_error(GOT_ERR_BAD_PATH);
451 		goto done;
452 	}
453 
454 	path = normpath;
455 	do {
456 		err = open_repo(repo, path);
457 		if (err == NULL)
458 			break;
459 		if (err->code != GOT_ERR_NOT_GIT_REPO)
460 			break;
461 		if (path[0] == '/' && path[1] == '\0') {
462 			if (tried_root) {
463 				err = got_error(GOT_ERR_NOT_GIT_REPO);
464 				break;
465 			}
466 			tried_root = 1;
467 		}
468 		path = dirname(path);
469 		if (path == NULL)
470 			err = got_error_from_errno();
471 	} while (path);
472 done:
473 	if (err)
474 		err = got_repo_close(repo);
475 	else
476 		*repop = repo;
477 	free(abspath);
478 	free(normpath);
479 	return err;
480 }
481 
482 #if 0
483 static void
484 print_cache_stats(struct got_object_cache *cache, const char *name)
485 {
486 	fprintf(stderr, "%s cache: %d elements, %d hits, %d missed\n",
487 	    name, got_object_idcache_num_elements(cache->idcache),
488 	    cache->cache_hit, cache->cache_miss);
489 }
490 
491 void check_refcount(struct got_object_id *id, void *data, void *arg)
492 {
493 	struct got_object_cache *cache = arg;
494 	struct got_object_cache_entry *ce = data;
495 	struct got_object *obj;
496 	struct got_tree_object *tree;
497 	struct got_commit_object *commit;
498 	char *id_str;
499 
500 	if (got_object_id_str(&id_str, id) != NULL)
501 		return;
502 
503 	switch (cache->type) {
504 	case GOT_OBJECT_CACHE_TYPE_OBJ:
505 		obj = ce->data.obj;
506 		if (obj->refcnt == 1)
507 			break;
508 		fprintf(stderr, "object %s has %d unclaimed references\n",
509 		    id_str, obj->refcnt - 1);
510 		break;
511 	case GOT_OBJECT_CACHE_TYPE_TREE:
512 		tree = ce->data.tree;
513 		if (tree->refcnt == 1)
514 			break;
515 		fprintf(stderr, "tree %s has %d unclaimed references\n",
516 		    id_str, tree->refcnt - 1);
517 		break;
518 	case GOT_OBJECT_CACHE_TYPE_COMMIT:
519 		commit = ce->data.commit;
520 		if (commit->refcnt == 1)
521 			break;
522 		fprintf(stderr, "commit %s has %d unclaimed references\n",
523 		    id_str, commit->refcnt);
524 		break;
525 	}
526 	free(id_str);
527 }
528 #endif
529 
530 static const struct got_error *
531 wait_for_child(pid_t pid)
532 {
533 	int child_status;
534 
535 	waitpid(pid, &child_status, 0);
536 
537 	if (!WIFEXITED(child_status))
538 		return got_error(GOT_ERR_PRIVSEP_DIED);
539 
540 	if (WEXITSTATUS(child_status) != 0)
541 		return got_error(GOT_ERR_PRIVSEP_EXIT);
542 
543 	return NULL;
544 }
545 
546 const struct got_error *
547 got_repo_close(struct got_repository *repo)
548 {
549 	const struct got_error *err = NULL, *child_err;
550 	int i;
551 
552 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
553 		if (repo->packidx_cache[i] == NULL)
554 			break;
555 		got_packidx_close(repo->packidx_cache[i]);
556 	}
557 
558 	for (i = 0; i < nitems(repo->packs); i++) {
559 		if (repo->packs[i].path_packfile == NULL)
560 			break;
561 		got_pack_close(&repo->packs[i]);
562 	}
563 
564 	free(repo->path);
565 	free(repo->path_git_dir);
566 
567 #if 0
568 	print_cache_stats(&repo->objcache, "object");
569 	print_cache_stats(&repo->treecache, "tree");
570 	print_cache_stats(&repo->commitcache, "commit");
571 	got_object_idcache_for_each(repo->objcache.idcache, check_refcount,
572 	    &repo->objcache);
573 	got_object_idcache_for_each(repo->treecache.idcache, check_refcount,
574 	    &repo->treecache);
575 	got_object_idcache_for_each(repo->commitcache.idcache, check_refcount,
576 	    &repo->commitcache);
577 #endif
578 
579 	if (repo->objcache.idcache)
580 		got_object_idcache_free(repo->objcache.idcache);
581 	if (repo->treecache.idcache)
582 		got_object_idcache_free(repo->treecache.idcache);
583 	if (repo->commitcache.idcache)
584 		got_object_idcache_free(repo->commitcache.idcache);
585 
586 	for (i = 0; i < nitems(repo->privsep_children); i++) {
587 		if (repo->privsep_children[i].imsg_fd == -1)
588 			continue;
589 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
590 		child_err = wait_for_child(repo->privsep_children[i].pid);
591 		if (child_err && err == NULL)
592 			err = child_err;
593 		close(repo->privsep_children[i].imsg_fd);
594 	}
595 	free(repo);
596 
597 	return err;
598 }
599 
600 const struct got_error *
601 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
602     const char *input_path)
603 {
604 	const struct got_error *err = NULL;
605 	char *repo_abspath = NULL, *cwd = NULL;
606 	struct stat sb;
607 	size_t repolen, cwdlen, len;
608 	char *canonpath, *path;
609 
610 	*in_repo_path = NULL;
611 
612 	cwd = getcwd(NULL, 0);
613 	if (cwd == NULL)
614 		return got_error_from_errno();
615 
616 	canonpath = strdup(input_path);
617 	if (canonpath == NULL) {
618 		err = got_error_from_errno();
619 		goto done;
620 	}
621 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
622 	if (err)
623 		goto done;
624 
625 	repo_abspath = got_repo_get_path(repo);
626 	if (repo_abspath == NULL) {
627 		err = got_error_from_errno();
628 		goto done;
629 	}
630 
631 	/* TODO: Call "get in-repository path of work-tree node" API. */
632 
633 	if (lstat(canonpath, &sb) != 0) {
634 		if (errno != ENOENT) {
635 			err = got_error_from_errno();
636 			goto done;
637 		}
638 		/*
639 		 * Path is not on disk.
640 		 * Assume it is already relative to repository root.
641 		 */
642 		path = strdup(canonpath);
643 	} else {
644 		int is_repo_child = 0, is_cwd_child = 0;
645 
646 		path = realpath(canonpath, NULL);
647 		if (path == NULL) {
648 			err = got_error_from_errno();
649 			goto done;
650 		}
651 
652 		repolen = strlen(repo_abspath);
653 		cwdlen = strlen(cwd);
654 		len = strlen(path);
655 
656 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
657 			is_repo_child = 1;
658 		if (len > cwdlen && strncmp(path, cwd, cwdlen) == 0)
659 			is_cwd_child = 1;
660 
661 		if (strcmp(path, repo_abspath) == 0) {
662 			free(path);
663 			path = strdup("");
664 			if (path == NULL) {
665 				err = got_error_from_errno();
666 				goto done;
667 			}
668 		} else if (is_repo_child && is_cwd_child) {
669 			char *child;
670 			/* TODO: Is path inside a got worktree? */
671 			/* Strip common prefix with repository path. */
672 			err = got_path_skip_common_ancestor(&child,
673 			    repo_abspath, path);
674 			if (err)
675 				goto done;
676 			free(path);
677 			path = child;
678 		} else if (is_repo_child) {
679 			/* Matched an on-disk path inside repository. */
680 			if (got_repo_is_bare(repo)) {
681 				/*
682 				 * Matched an on-disk path inside repository
683 				 * database. Treat as repository-relative.
684 				 */
685 			} else {
686 				char *child;
687 				/* Strip common prefix with repository path. */
688 				err = got_path_skip_common_ancestor(&child,
689 				    repo_abspath, path);
690 				if (err)
691 					goto done;
692 				free(path);
693 				path = child;
694 			}
695 		} else if (is_cwd_child) {
696 			char *child;
697 			/* TODO: Is path inside a got worktree? */
698 			/* Strip common prefix with cwd. */
699 			err = got_path_skip_common_ancestor(&child, cwd,
700 			    path);
701 			if (err)
702 				goto done;
703 			free(path);
704 			path = child;
705 		} else {
706 			/*
707 			 * Matched unrelated on-disk path.
708 			 * Treat it as repository-relative.
709 			 */
710 		}
711 	}
712 
713 	/* Make in-repository path absolute */
714 	if (path[0] != '/') {
715 		char *abspath;
716 		if (asprintf(&abspath, "/%s", path) == -1) {
717 			err = got_error_from_errno();
718 			goto done;
719 		}
720 		free(path);
721 		path = abspath;
722 	}
723 
724 done:
725 	free(repo_abspath);
726 	free(cwd);
727 	free(canonpath);
728 	if (err)
729 		free(path);
730 	else
731 		*in_repo_path = path;
732 	return err;
733 }