Gotweb

Blob

Date:: Tue Feb 5 23:04:59 2019 UTC
Message:: make 'got diff' show local changes in work tree
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018, 2019 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/stat.h>
18 #include <sys/limits.h>
19 #include <sys/queue.h>
20 #include <sys/tree.h>
21 
22 #include <dirent.h>
23 #include <stddef.h>
24 #include <string.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <fcntl.h>
28 #include <errno.h>
29 #include <unistd.h>
30 #include <sha1.h>
31 #include <zlib.h>
32 #include <fnmatch.h>
33 #include <libgen.h>
34 
35 #include "got_error.h"
36 #include "got_repository.h"
37 #include "got_reference.h"
38 #include "got_object.h"
39 #include "got_worktree.h"
40 #include "got_opentemp.h"
41 
42 #include "got_lib_worktree.h"
43 #include "got_lib_path.h"
44 #include "got_lib_sha1.h"
45 #include "got_lib_fileindex.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_delta.h"
48 #include "got_lib_object.h"
49 
50 #ifndef MIN
51 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
52 #endif
53 
54 static const struct got_error *
55 create_meta_file(const char *path_got, const char *name, const char *content)
56 {
57 	const struct got_error *err = NULL;
58 	char *path;
59 	int fd = -1;
60 
61 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
62 		err = got_error_from_errno();
63 		path = NULL;
64 		goto done;
65 	}
66 
67 	fd = open(path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
68 	    GOT_DEFAULT_FILE_MODE);
69 	if (fd == -1) {
70 		err = got_error_from_errno();
71 		goto done;
72 	}
73 
74 	if (content) {
75 		int len = dprintf(fd, "%s\n", content);
76 		if (len != strlen(content) + 1) {
77 			err = got_error_from_errno();
78 			goto done;
79 		}
80 	}
81 
82 done:
83 	if (fd != -1 && close(fd) == -1 && err == NULL)
84 		err = got_error_from_errno();
85 	free(path);
86 	return err;
87 }
88 
89 static const struct got_error *
90 update_meta_file(const char *path_got, const char *name, const char *content)
91 {
92 	const struct got_error *err = NULL;
93 	FILE *tmpfile = NULL;
94 	char *tmppath = NULL;
95 	char *path = NULL;
96 
97 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
98 		err = got_error_from_errno();
99 		path = NULL;
100 		goto done;
101 	}
102 
103 	err = got_opentemp_named(&tmppath, &tmpfile, path);
104 	if (err)
105 		goto done;
106 
107 	if (content) {
108 		int len = fprintf(tmpfile, "%s\n", content);
109 		if (len != strlen(content) + 1) {
110 			err = got_error_from_errno();
111 			goto done;
112 		}
113 	}
114 
115 	if (rename(tmppath, path) != 0) {
116 		err = got_error_from_errno();
117 		goto done;
118 	}
119 
120 done:
121 	free(tmppath);
122 	fclose(tmpfile);
123 	return err;
124 }
125 
126 static const struct got_error *
127 read_meta_file(char **content, const char *path_got, const char *name)
128 {
129 	const struct got_error *err = NULL;
130 	char *path;
131 	int fd = -1;
132 	ssize_t n;
133 	struct stat sb;
134 
135 	*content = NULL;
136 
137 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
138 		err = got_error_from_errno();
139 		path = NULL;
140 		goto done;
141 	}
142 
143 	fd = open(path, O_RDONLY | O_NOFOLLOW);
144 	if (fd == -1) {
145 		err = got_error_from_errno();
146 		goto done;
147 	}
148 	if (flock(fd, LOCK_SH | LOCK_NB) == -1) {
149 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
150 		    : got_error_from_errno());
151 		goto done;
152 	}
153 
154 	stat(path, &sb);
155 	*content = calloc(1, sb.st_size);
156 	if (*content == NULL) {
157 		err = got_error_from_errno();
158 		goto done;
159 	}
160 
161 	n = read(fd, *content, sb.st_size);
162 	if (n != sb.st_size) {
163 		err = (n == -1 ? got_error_from_errno() :
164 		    got_error(GOT_ERR_WORKTREE_META));
165 		goto done;
166 	}
167 	if ((*content)[sb.st_size - 1] != '\n') {
168 		err = got_error(GOT_ERR_WORKTREE_META);
169 		goto done;
170 	}
171 	(*content)[sb.st_size - 1] = '\0';
172 
173 done:
174 	if (fd != -1 && close(fd) == -1 && err == NULL)
175 		err = got_error_from_errno();
176 	free(path);
177 	if (err) {
178 		free(*content);
179 		*content = NULL;
180 	}
181 	return err;
182 }
183 
184 const struct got_error *
185 got_worktree_init(const char *path, struct got_reference *head_ref,
186     const char *prefix, struct got_repository *repo)
187 {
188 	const struct got_error *err = NULL;
189 	struct got_object_id *commit_id = NULL;
190 	int obj_type;
191 	char *path_got = NULL;
192 	char *refstr = NULL;
193 	char *formatstr = NULL;
194 	char *absprefix = NULL;
195 	char *basestr = NULL;
196 
197 	err = got_ref_resolve(&commit_id, repo, head_ref);
198 	if (err)
199 		return err;
200 	err = got_object_get_type(&obj_type, repo, commit_id);
201 	if (err)
202 		return err;
203 	if (obj_type != GOT_OBJ_TYPE_COMMIT)
204 		return got_error(GOT_ERR_OBJ_TYPE);
205 
206 	if (!got_path_is_absolute(prefix)) {
207 		if (asprintf(&absprefix, "/%s", prefix) == -1)
208 			return got_error_from_errno();
209 	}
210 
211 	/* Create top-level directory (may already exist). */
212 	if (mkdir(path, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
213 		err = got_error_from_errno();
214 		goto done;
215 	}
216 
217 	/* Create .got directory (may already exist). */
218 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
219 		err = got_error_from_errno();
220 		goto done;
221 	}
222 	if (mkdir(path_got, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
223 		err = got_error_from_errno();
224 		goto done;
225 	}
226 
227 	/* Create an empty lock file. */
228 	err = create_meta_file(path_got, GOT_WORKTREE_LOCK, NULL);
229 	if (err)
230 		goto done;
231 
232 	/* Create an empty file index. */
233 	err = create_meta_file(path_got, GOT_WORKTREE_FILE_INDEX, NULL);
234 	if (err)
235 		goto done;
236 
237 	/* Write the HEAD reference. */
238 	refstr = got_ref_to_str(head_ref);
239 	if (refstr == NULL) {
240 		err = got_error_from_errno();
241 		goto done;
242 	}
243 	err = create_meta_file(path_got, GOT_WORKTREE_HEAD_REF, refstr);
244 	if (err)
245 		goto done;
246 
247 	/* Record our base commit. */
248 	err = got_object_id_str(&basestr, commit_id);
249 	if (err)
250 		goto done;
251 	err = create_meta_file(path_got, GOT_WORKTREE_BASE_COMMIT, basestr);
252 	if (err)
253 		goto done;
254 
255 	/* Store path to repository. */
256 	err = create_meta_file(path_got, GOT_WORKTREE_REPOSITORY,
257 	    got_repo_get_path(repo));
258 	if (err)
259 		goto done;
260 
261 	/* Store in-repository path prefix. */
262 	err = create_meta_file(path_got, GOT_WORKTREE_PATH_PREFIX,
263 	    absprefix ? absprefix : prefix);
264 	if (err)
265 		goto done;
266 
267 	/* Stamp work tree with format file. */
268 	if (asprintf(&formatstr, "%d", GOT_WORKTREE_FORMAT_VERSION) == -1) {
269 		err = got_error_from_errno();
270 		goto done;
271 	}
272 	err = create_meta_file(path_got, GOT_WORKTREE_FORMAT, formatstr);
273 	if (err)
274 		goto done;
275 
276 done:
277 	free(commit_id);
278 	free(path_got);
279 	free(formatstr);
280 	free(refstr);
281 	free(absprefix);
282 	free(basestr);
283 	return err;
284 }
285 
286 static const struct got_error *
287 open_worktree(struct got_worktree **worktree, const char *path)
288 {
289 	const struct got_error *err = NULL;
290 	char *path_got;
291 	char *formatstr = NULL;
292 	char *path_lock = NULL;
293 	char *base_commit_id_str = NULL;
294 	char *head_ref_str = NULL;
295 	int version, fd = -1;
296 	const char *errstr;
297 	struct got_repository *repo = NULL;
298 
299 	*worktree = NULL;
300 
301 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
302 		err = got_error_from_errno();
303 		path_got = NULL;
304 		goto done;
305 	}
306 
307 	if (asprintf(&path_lock, "%s/%s", path_got, GOT_WORKTREE_LOCK) == -1) {
308 		err = got_error_from_errno();
309 		path_lock = NULL;
310 		goto done;
311 	}
312 
313 	fd = open(path_lock, O_RDWR | O_EXLOCK | O_NONBLOCK);
314 	if (fd == -1) {
315 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
316 		    : got_error_from_errno());
317 		goto done;
318 	}
319 
320 	err = read_meta_file(&formatstr, path_got, GOT_WORKTREE_FORMAT);
321 	if (err)
322 		goto done;
323 
324 	version = strtonum(formatstr, 1, INT_MAX, &errstr);
325 	if (errstr) {
326 		err = got_error(GOT_ERR_WORKTREE_META);
327 		goto done;
328 	}
329 	if (version != GOT_WORKTREE_FORMAT_VERSION) {
330 		err = got_error(GOT_ERR_WORKTREE_VERS);
331 		goto done;
332 	}
333 
334 	*worktree = calloc(1, sizeof(**worktree));
335 	if (*worktree == NULL) {
336 		err = got_error_from_errno();
337 		goto done;
338 	}
339 	(*worktree)->lockfd = -1;
340 
341 	(*worktree)->root_path = strdup(path);
342 	if ((*worktree)->root_path == NULL) {
343 		err = got_error_from_errno();
344 		goto done;
345 	}
346 	err = read_meta_file(&(*worktree)->repo_path, path_got,
347 	    GOT_WORKTREE_REPOSITORY);
348 	if (err)
349 		goto done;
350 
351 	err = read_meta_file(&(*worktree)->path_prefix, path_got,
352 	    GOT_WORKTREE_PATH_PREFIX);
353 	if (err)
354 		goto done;
355 
356 	err = read_meta_file(&base_commit_id_str, path_got,
357 	    GOT_WORKTREE_BASE_COMMIT);
358 	if (err)
359 		goto done;
360 
361 	err = got_repo_open(&repo, (*worktree)->repo_path);
362 	if (err)
363 		goto done;
364 
365 	err = got_object_resolve_id_str(&(*worktree)->base_commit_id, repo,
366 	    base_commit_id_str);
367 	if (err)
368 		goto done;
369 
370 	err = read_meta_file(&head_ref_str, path_got, GOT_WORKTREE_HEAD_REF);
371 	if (err)
372 		goto done;
373 
374 	err = got_ref_open(&(*worktree)->head_ref, repo, head_ref_str);
375 done:
376 	if (repo)
377 		got_repo_close(repo);
378 	free(path_got);
379 	free(path_lock);
380 	free(head_ref_str);
381 	free(base_commit_id_str);
382 	if (err) {
383 		if (fd != -1)
384 			close(fd);
385 		if (*worktree != NULL)
386 			got_worktree_close(*worktree);
387 		*worktree = NULL;
388 	} else
389 		(*worktree)->lockfd = fd;
390 
391 	return err;
392 }
393 
394 const struct got_error *
395 got_worktree_open(struct got_worktree **worktree, const char *path)
396 {
397 	const struct got_error *err = NULL;
398 
399 	do {
400 		err = open_worktree(worktree, path);
401 		if (err && (err->code != GOT_ERR_ERRNO && errno != ENOENT))
402 			return err;
403 		if (*worktree)
404 			return NULL;
405 		path = dirname(path);
406 		if (path == NULL)
407 			return got_error_from_errno();
408 	} while (!((path[0] == '.' || path[0] == '/') && path[1] == '\0'));
409 
410 	return got_error(GOT_ERR_NOT_WORKTREE);
411 }
412 
413 void
414 got_worktree_close(struct got_worktree *worktree)
415 {
416 	free(worktree->root_path);
417 	free(worktree->repo_path);
418 	free(worktree->path_prefix);
419 	free(worktree->base_commit_id);
420 	if (worktree->head_ref)
421 		got_ref_close(worktree->head_ref);
422 	if (worktree->lockfd != -1)
423 		close(worktree->lockfd);
424 	free(worktree);
425 }
426 
427 const char *
428 got_worktree_get_root_path(struct got_worktree *worktree)
429 {
430 	return worktree->root_path;
431 }
432 
433 const char *
434 got_worktree_get_repo_path(struct got_worktree *worktree)
435 {
436 	return worktree->repo_path;
437 }
438 
439 const char *
440 got_worktree_get_path_prefix(struct got_worktree *worktree)
441 {
442 	return worktree->path_prefix;
443 }
444 
445 const struct got_error *
446 got_worktree_match_path_prefix(int *match, struct got_worktree *worktree,
447     const char *path_prefix)
448 {
449 	char *absprefix = NULL;
450 
451 	if (!got_path_is_absolute(path_prefix)) {
452 		if (asprintf(&absprefix, "/%s", path_prefix) == -1)
453 			return got_error_from_errno();
454 	}
455 	*match = (strcmp(absprefix ? absprefix : path_prefix,
456 	    worktree->path_prefix) == 0);
457 	free(absprefix);
458 	return NULL;
459 }
460 
461 char *
462 got_worktree_get_head_ref_name(struct got_worktree *worktree)
463 {
464 	return got_ref_to_str(worktree->head_ref);
465 }
466 
467 struct got_reference *
468 got_worktree_get_head_ref(struct got_worktree *worktree)
469 {
470 	return got_ref_dup(worktree->head_ref);
471 }
472 
473 struct got_object_id *
474 got_worktree_get_base_commit_id(struct got_worktree *worktree)
475 {
476 	return worktree->base_commit_id;
477 }
478 
479 const struct got_error *
480 got_worktree_set_base_commit_id(struct got_worktree *worktree,
481     struct got_repository *repo, struct got_object_id *commit_id)
482 {
483 	const struct got_error *err;
484 	struct got_object *obj = NULL;
485 	char *id_str = NULL;
486 	char *path_got = NULL;
487 
488 	if (asprintf(&path_got, "%s/%s", worktree->root_path,
489 	    GOT_WORKTREE_GOT_DIR) == -1) {
490 		err = got_error_from_errno();
491 		path_got = NULL;
492 		goto done;
493 	}
494 
495 	err = got_object_open(&obj, repo, commit_id);
496 	if (err)
497 		return err;
498 
499 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
500 		err = got_error(GOT_ERR_OBJ_TYPE);
501 		goto done;
502 	}
503 
504 	/* Record our base commit. */
505 	err = got_object_id_str(&id_str, commit_id);
506 	if (err)
507 		goto done;
508 	err = update_meta_file(path_got, GOT_WORKTREE_BASE_COMMIT, id_str);
509 	if (err)
510 		goto done;
511 
512 	free(worktree->base_commit_id);
513 	worktree->base_commit_id = got_object_id_dup(commit_id);
514 	if (worktree->base_commit_id == NULL) {
515 		err = got_error_from_errno();
516 		goto done;
517 	}
518 done:
519 	if (obj)
520 		got_object_close(obj);
521 	free(id_str);
522 	free(path_got);
523 	return err;
524 }
525 
526 static const struct got_error *
527 lock_worktree(struct got_worktree *worktree, int operation)
528 {
529 	if (flock(worktree->lockfd, operation | LOCK_NB) == -1)
530 		return (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
531 		    : got_error_from_errno());
532 	return NULL;
533 }
534 
535 static const struct got_error *
536 make_parent_dirs(const char *abspath)
537 {
538 	const struct got_error *err = NULL;
539 
540 	char *parent = dirname(abspath);
541 	if (parent == NULL)
542 		return NULL;
543 
544 	if (mkdir(parent, GOT_DEFAULT_DIR_MODE) == -1) {
545 		if (errno == ENOENT) {
546 			err = make_parent_dirs(parent);
547 			if (err)
548 				return err;
549 			if (mkdir(parent, GOT_DEFAULT_DIR_MODE) == -1)
550 				return got_error_from_errno();
551 		} else
552 			err = got_error_from_errno();
553 	}
554 
555 	return err;
556 }
557 
558 static const struct got_error *
559 add_dir_on_disk(struct got_worktree *worktree, const char *path)
560 {
561 	const struct got_error *err = NULL;
562 	char *abspath;
563 
564 	if (asprintf(&abspath, "%s/%s", worktree->root_path, path) == -1)
565 		return got_error_from_errno();
566 
567 	/* XXX queue work rather than editing disk directly? */
568 	if (mkdir(abspath, GOT_DEFAULT_DIR_MODE) == -1) {
569 		struct stat sb;
570 
571 		if (errno == EEXIST) {
572 			if (lstat(abspath, &sb) == -1) {
573 				err = got_error_from_errno();
574 				goto done;
575 			}
576 
577 			if (!S_ISDIR(sb.st_mode)) {
578 				/* TODO directory is obstructed; do something */
579 				err = got_error(GOT_ERR_FILE_OBSTRUCTED);
580 				goto done;
581 			}
582 
583 			return NULL;
584 		} else if (errno == ENOENT) {
585 			err = make_parent_dirs(abspath);
586 			if (err)
587 				goto done;
588 			if (mkdir(abspath, GOT_DEFAULT_DIR_MODE) == -1)
589 				err = got_error_from_errno();
590 		} else
591 			err = got_error_from_errno();
592 	}
593 
594 done:
595 	free(abspath);
596 	return err;
597 }
598 
599 static const struct got_error *
600 install_blob(struct got_worktree *worktree, struct got_fileindex *fileindex,
601    struct got_fileindex_entry *entry, const char *path,
602    struct got_blob_object *blob,
603    struct got_repository *repo, got_worktree_checkout_cb progress_cb,
604    void *progress_arg)
605 {
606 	const struct got_error *err = NULL;
607 	char *ondisk_path;
608 	int fd = -1;
609 	size_t len, hdrlen;
610 	int update = 0;
611 	char *tmppath = NULL;
612 
613 	if (asprintf(&ondisk_path, "%s/%s", worktree->root_path, path) == -1)
614 		return got_error_from_errno();
615 
616 	fd = open(ondisk_path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
617 	    GOT_DEFAULT_FILE_MODE);
618 	if (fd == -1) {
619 		if (errno == ENOENT) {
620 			char *parent = dirname(path);
621 			if (parent == NULL)
622 				return got_error_from_errno();
623 			err = add_dir_on_disk(worktree, parent);
624 			if (err)
625 				return err;
626 			fd = open(ondisk_path,
627 			    O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
628 			    GOT_DEFAULT_FILE_MODE);
629 			if (fd == -1)
630 				return got_error_from_errno();
631 		} else if (errno == EEXIST) {
632 			struct stat sb;
633 			if (lstat(ondisk_path, &sb) == -1) {
634 				err = got_error_from_errno();
635 				goto done;
636 			} else if (!S_ISREG(sb.st_mode)) {
637 				/* TODO file is obstructed; do something */
638 				err = got_error(GOT_ERR_FILE_OBSTRUCTED);
639 				goto done;
640 			} else {
641 				err = got_opentemp_named_fd(&tmppath, &fd,
642 				    ondisk_path);
643 				if (err)
644 					goto done;
645 				update = 1;
646 			}
647 		} else
648 			return got_error_from_errno();
649 	}
650 
651 	(*progress_cb)(progress_arg,
652 	    update ? GOT_STATUS_UPDATE : GOT_STATUS_ADD, path);
653 
654 	hdrlen = got_object_blob_get_hdrlen(blob);
655 	do {
656 		const uint8_t *buf = got_object_blob_get_read_buf(blob);
657 		err = got_object_blob_read_block(&len, blob);
658 		if (err)
659 			break;
660 		if (len > 0) {
661 			/* Skip blob object header first time around. */
662 			ssize_t outlen = write(fd, buf + hdrlen, len - hdrlen);
663 			if (outlen == -1) {
664 				err = got_error_from_errno();
665 				goto done;
666 			} else if (outlen != len - hdrlen) {
667 				err = got_error(GOT_ERR_IO);
668 				goto done;
669 			}
670 			hdrlen = 0;
671 		}
672 	} while (len != 0);
673 
674 	fsync(fd);
675 
676 	if (update) {
677 		if (rename(tmppath, ondisk_path) != 0) {
678 			err = got_error_from_errno();
679 			goto done;
680 		}
681 	}
682 
683 	if (entry == NULL)
684 		entry = got_fileindex_entry_get(fileindex, path);
685 	if (entry)
686 		err = got_fileindex_entry_update(entry, ondisk_path,
687 		    blob->id.sha1, worktree->base_commit_id->sha1);
688 	else {
689 		err = got_fileindex_entry_alloc(&entry, ondisk_path,
690 		    path, blob->id.sha1, worktree->base_commit_id->sha1);
691 		if (err)
692 			goto done;
693 		err = got_fileindex_entry_add(fileindex, entry);
694 	}
695 done:
696 	if (fd != -1)
697 		close(fd);
698 	free(ondisk_path);
699 	free(tmppath);
700 	return err;
701 }
702 
703 static const struct got_error *
704 update_blob(struct got_worktree *worktree,
705     struct got_fileindex *fileindex, struct got_fileindex_entry *ie,
706     struct got_tree_entry *te, const char *path,
707     struct got_repository *repo, got_worktree_checkout_cb progress_cb,
708     void *progress_arg, got_worktree_cancel_cb cancel_cb, void *cancel_arg)
709 {
710 	const struct got_error *err = NULL;
711 	struct got_blob_object *blob = NULL;
712 
713 	if (ie) {
714 		if (memcmp(ie->commit_sha1, worktree->base_commit_id->sha1,
715 		    SHA1_DIGEST_LENGTH) == 0) {
716 			(*progress_cb)(progress_arg, GOT_STATUS_EXISTS,
717 			    path);
718 			return NULL;
719 		}
720 		if (memcmp(ie->blob_sha1,
721 		    te->id->sha1, SHA1_DIGEST_LENGTH) == 0)
722 			return NULL;
723 	}
724 
725 	err = got_object_open_as_blob(&blob, repo, te->id, 8192);
726 	if (err)
727 		return err;
728 
729 	err = install_blob(worktree, fileindex, ie, path, blob, repo,
730 	    progress_cb, progress_arg);
731 	got_object_blob_close(blob);
732 	return err;
733 }
734 
735 static const struct got_error *
736 remove_ondisk_file(const char *root_path, const char *path)
737 {
738 	const struct got_error *err = NULL;
739 	char *ondisk_path = NULL;
740 
741 	if (asprintf(&ondisk_path, "%s/%s", root_path, path) == -1)
742 		return got_error_from_errno();
743 
744 	if (unlink(ondisk_path) == -1) {
745 		if (errno != ENOENT)
746 			err = got_error_from_errno();
747 	} else {
748 		char *parent = dirname(ondisk_path);
749 		while (parent && strcmp(parent, root_path) != 0) {
750 			if (rmdir(parent) == -1) {
751 				if (errno != ENOTEMPTY)
752 					err = got_error_from_errno();
753 				break;
754 			}
755 			parent = dirname(parent);
756 		}
757 	}
758 	free(ondisk_path);
759 	return err;
760 }
761 
762 struct diff_cb_arg {
763     struct got_fileindex *fileindex;
764     struct got_worktree *worktree;
765     struct got_repository *repo;
766     got_worktree_checkout_cb progress_cb;
767     void *progress_arg;
768     got_worktree_cancel_cb cancel_cb;
769     void *cancel_arg;
770 };
771 
772 static const struct got_error *
773 diff_old_new(void *arg, struct got_fileindex_entry *ie,
774     struct got_tree_entry *te, const char *parent_path)
775 {
776 	struct diff_cb_arg *a = arg;
777 
778 	return update_blob(a->worktree, a->fileindex, ie, te,
779 	    ie->path, a->repo, a->progress_cb, a->progress_arg,
780 	    a->cancel_cb, a->cancel_arg);
781 }
782 
783 static const struct got_error *
784 diff_old(void *arg, struct got_fileindex_entry *ie, const char *parent_path)
785 {
786 	const struct got_error *err;
787 	struct diff_cb_arg *a = arg;
788 
789 	(*a->progress_cb)(a->progress_arg, GOT_STATUS_DELETE, ie->path);
790 
791 	err = remove_ondisk_file(a->worktree->root_path, ie->path);
792 	if (err)
793 		return err;
794 	got_fileindex_entry_remove(a->fileindex, ie);
795 	return NULL;
796 }
797 
798 static const struct got_error *
799 diff_new(void *arg, struct got_tree_entry *te, const char *parent_path)
800 {
801 	struct diff_cb_arg *a = arg;
802 	const struct got_error *err;
803 	char *path;
804 
805 	if (asprintf(&path, "%s%s%s", parent_path,
806 	    parent_path[0] ? "/" : "", te->name)
807 	    == -1)
808 		return got_error_from_errno();
809 
810 	if (S_ISDIR(te->mode))
811 		err = add_dir_on_disk(a->worktree, path);
812 	else
813 		err = update_blob(a->worktree, a->fileindex, NULL, te, path,
814 		    a->repo, a->progress_cb, a->progress_arg,
815 		    a->cancel_cb, a->cancel_arg);
816 
817 	free(path);
818 	return err;
819 }
820 
821 const struct got_error *
822 got_worktree_checkout_files(struct got_worktree *worktree,
823     struct got_repository *repo, got_worktree_checkout_cb progress_cb,
824     void *progress_arg, got_worktree_cancel_cb cancel_cb, void *cancel_arg)
825 {
826 	const struct got_error *err = NULL, *unlockerr, *checkout_err = NULL;
827 	struct got_commit_object *commit = NULL;
828 	struct got_object_id *tree_id = NULL;
829 	struct got_tree_object *tree = NULL;
830 	char *fileindex_path = NULL, *new_fileindex_path = NULL;
831 	struct got_fileindex *fileindex = NULL;
832 	FILE *index = NULL, *new_index = NULL;
833 	struct got_fileindex_diff_tree_cb diff_cb;
834 	struct diff_cb_arg arg;
835 
836 	err = lock_worktree(worktree, LOCK_EX);
837 	if (err)
838 		return err;
839 
840 	fileindex = got_fileindex_alloc();
841 	if (fileindex == NULL) {
842 		err = got_error_from_errno();
843 		goto done;
844 	}
845 
846 	if (asprintf(&fileindex_path, "%s/%s/%s", worktree->root_path,
847 	    GOT_WORKTREE_GOT_DIR, GOT_WORKTREE_FILE_INDEX) == -1) {
848 		err = got_error_from_errno();
849 		fileindex_path = NULL;
850 		goto done;
851 	}
852 
853 	/*
854 	 * Read the file index.
855 	 * Checking out files is supposed to be an idempotent operation.
856 	 * If the on-disk file index is incomplete we will try to complete it.
857 	 */
858 	index = fopen(fileindex_path, "rb");
859 	if (index == NULL) {
860 		if (errno != ENOENT) {
861 			err = got_error_from_errno();
862 			goto done;
863 		}
864 	} else {
865 		err = got_fileindex_read(fileindex, index);
866 		fclose(index);
867 		if (err)
868 			goto done;
869 	}
870 
871 	err = got_opentemp_named(&new_fileindex_path, &new_index,
872 	    fileindex_path);
873 	if (err)
874 		goto done;
875 
876 	err = got_object_open_as_commit(&commit, repo,
877 	   worktree->base_commit_id);
878 	if (err)
879 		goto done;
880 
881 	err = got_object_id_by_path(&tree_id, repo,
882 	    worktree->base_commit_id, worktree->path_prefix);
883 	if (err)
884 		goto done;
885 
886 	err = got_object_open_as_tree(&tree, repo, tree_id);
887 	if (err)
888 		goto done;
889 
890 	diff_cb.diff_old_new = diff_old_new;
891 	diff_cb.diff_old = diff_old;
892 	diff_cb.diff_new = diff_new;
893 	arg.fileindex = fileindex;
894 	arg.worktree = worktree;
895 	arg.repo = repo;
896 	arg.progress_cb = progress_cb;
897 	arg.progress_arg = progress_arg;
898 	arg.cancel_cb = cancel_cb;
899 	arg.cancel_arg = cancel_arg;
900 	checkout_err = got_fileindex_diff_tree(fileindex, tree, repo,
901 	    &diff_cb, &arg);
902 
903 	/* Try to sync the fileindex back to disk in any case. */
904 	err = got_fileindex_write(fileindex, new_index);
905 	if (err)
906 		goto done;
907 
908 	if (rename(new_fileindex_path, fileindex_path) != 0) {
909 		err = got_error_from_errno();
910 		goto done;
911 	}
912 
913 	free(new_fileindex_path);
914 	new_fileindex_path = NULL;
915 
916 done:
917 	if (tree)
918 		got_object_tree_close(tree);
919 	if (commit)
920 		got_object_commit_close(commit);
921 	if (new_fileindex_path)
922 		unlink(new_fileindex_path);
923 	if (new_index)
924 		fclose(new_index);
925 	free(new_fileindex_path);
926 	free(fileindex_path);
927 	got_fileindex_free(fileindex);
928 	if (checkout_err)
929 		err = checkout_err;
930 	unlockerr = lock_worktree(worktree, LOCK_SH);
931 	if (unlockerr && err == NULL)
932 		err = unlockerr;
933 	return err;
934 }
935 
936 struct diff_dir_cb_arg {
937     struct got_fileindex *fileindex;
938     struct got_worktree *worktree;
939     struct got_repository *repo;
940     got_worktree_status_cb status_cb;
941     void *status_arg;
942     got_worktree_cancel_cb cancel_cb;
943     void *cancel_arg;
944 };
945 
946 static const struct got_error *
947 get_file_status(unsigned char *status, struct got_fileindex_entry *ie,
948     const char *abspath, struct got_repository *repo)
949 {
950 	const struct got_error *err = NULL;
951 	struct got_object_id id;
952 	size_t hdrlen;
953 	FILE *f = NULL;
954 	uint8_t fbuf[8192];
955 	struct got_blob_object *blob = NULL;
956 	size_t flen, blen;
957 	struct stat sb;
958 
959 	*status = GOT_STATUS_NO_CHANGE;
960 
961 	if (lstat(abspath, &sb) == -1)
962 		return got_error_from_errno();
963 
964 	if (!S_ISREG(sb.st_mode)) {
965 		*status = GOT_STATUS_OBSTRUCTED;
966 		return NULL;
967 	}
968 
969 	if (ie->ctime_sec == sb.st_ctime &&
970 	    ie->ctime_nsec == sb.st_ctimensec &&
971 	    ie->mtime_sec == sb.st_mtime &&
972 	    ie->mtime_sec == sb.st_mtime &&
973 	    ie->mtime_nsec == sb.st_mtimensec &&
974 	    ie->size == (sb.st_size & 0xffffffff))
975 		return NULL;
976 
977 	memcpy(id.sha1, ie->blob_sha1, sizeof(id.sha1));
978 	err = got_object_open_as_blob(&blob, repo, &id, sizeof(fbuf));
979 	if (err)
980 		return err;
981 
982 	f = fopen(abspath, "r");
983 	if (f == NULL) {
984 		err = got_error_from_errno();
985 		goto done;
986 	}
987 	hdrlen = got_object_blob_get_hdrlen(blob);
988 	while (1) {
989 		const uint8_t *bbuf = got_object_blob_get_read_buf(blob);
990 		err = got_object_blob_read_block(&blen, blob);
991 		if (err)
992 			break;
993 		flen = fread(fbuf, 1, sizeof(fbuf), f);
994 		if (blen == 0) {
995 			if (flen != 0)
996 				*status = GOT_STATUS_MODIFIY;
997 			break;
998 		} else if (flen == 0) {
999 			if (blen != 0)
1000 				*status = GOT_STATUS_MODIFIY;
1001 			break;
1002 		} else if (blen - hdrlen == flen) {
1003 			/* Skip blob object header first time around. */
1004 			if (memcmp(bbuf + hdrlen, fbuf, flen) != 0) {
1005 				*status = GOT_STATUS_MODIFIY;
1006 				break;
1007 			}
1008 		} else {
1009 			*status = GOT_STATUS_MODIFIY;
1010 			break;
1011 		}
1012 		hdrlen = 0;
1013 	}
1014 done:
1015 	if (blob)
1016 		got_object_blob_close(blob);
1017 	if (f)
1018 		fclose(f);
1019 	return err;
1020 }
1021 
1022 static const struct got_error *
1023 status_old_new(void *arg, struct got_fileindex_entry *ie,
1024     struct dirent *de, const char *parent_path)
1025 {
1026 	const struct got_error *err = NULL;
1027 	struct diff_dir_cb_arg *a = arg;
1028 	char *abspath;
1029 	unsigned char status = GOT_STATUS_NO_CHANGE;
1030 
1031 	if (parent_path[0]) {
1032 		if (asprintf(&abspath, "%s/%s/%s", a->worktree->root_path,
1033 		    parent_path, de->d_name) == -1)
1034 			return got_error_from_errno();
1035 	} else {
1036 		if (asprintf(&abspath, "%s/%s", a->worktree->root_path,
1037 		    de->d_name) == -1)
1038 			return got_error_from_errno();
1039 	}
1040 
1041 	err = get_file_status(&status, ie, abspath, a->repo);
1042 	if (err == NULL && status != GOT_STATUS_NO_CHANGE) {
1043 		struct got_object_id id;
1044 		memcpy(id.sha1, ie->blob_sha1, SHA1_DIGEST_LENGTH);
1045 		err = (*a->status_cb)(a->status_arg, status, ie->path, &id);
1046 	}
1047 	free(abspath);
1048 	return err;
1049 }
1050 
1051 static const struct got_error *
1052 status_old(void *arg, struct got_fileindex_entry *ie, const char *parent_path)
1053 {
1054 	struct diff_dir_cb_arg *a = arg;
1055 	struct got_object_id id;
1056 	memcpy(id.sha1, ie->blob_sha1, SHA1_DIGEST_LENGTH);
1057 	return (*a->status_cb)(a->status_arg, GOT_STATUS_MISSING, ie->path,
1058 	    &id);
1059 }
1060 
1061 static const struct got_error *
1062 status_new(void *arg, struct dirent *de, const char *parent_path)
1063 {
1064 	const struct got_error *err = NULL;
1065 	struct diff_dir_cb_arg *a = arg;
1066 	char *path = NULL;
1067 
1068 	if (de->d_type == DT_DIR)
1069 		return NULL;
1070 
1071 	if (parent_path[0]) {
1072 		if (asprintf(&path, "%s/%s", parent_path, de->d_name) == -1)
1073 			return got_error_from_errno();
1074 	} else {
1075 		path = de->d_name;
1076 	}
1077 
1078 	err = (*a->status_cb)(a->status_arg, GOT_STATUS_UNVERSIONED, path,
1079 	    NULL);
1080 	if (parent_path[0])
1081 		free(path);
1082 	return err;
1083 }
1084 
1085 const struct got_error *
1086 got_worktree_status(struct got_worktree *worktree,
1087     struct got_repository *repo, got_worktree_status_cb status_cb,
1088     void *status_arg, got_worktree_cancel_cb cancel_cb, void *cancel_arg)
1089 {
1090 	const struct got_error *err = NULL;
1091 	DIR *workdir = NULL;
1092 	char *fileindex_path = NULL;
1093 	struct got_fileindex *fileindex = NULL;
1094 	FILE *index = NULL;
1095 	struct got_fileindex_diff_dir_cb fdiff_cb;
1096 	struct diff_dir_cb_arg arg;
1097 
1098 	fileindex = got_fileindex_alloc();
1099 	if (fileindex == NULL) {
1100 		err = got_error_from_errno();
1101 		goto done;
1102 	}
1103 
1104 	if (asprintf(&fileindex_path, "%s/%s/%s", worktree->root_path,
1105 	    GOT_WORKTREE_GOT_DIR, GOT_WORKTREE_FILE_INDEX) == -1) {
1106 		err = got_error_from_errno();
1107 		fileindex_path = NULL;
1108 		goto done;
1109 	}
1110 
1111 	index = fopen(fileindex_path, "rb");
1112 	if (index == NULL) {
1113 		if (errno != ENOENT) {
1114 			err = got_error_from_errno();
1115 			goto done;
1116 		}
1117 	} else {
1118 		err = got_fileindex_read(fileindex, index);
1119 		fclose(index);
1120 		if (err)
1121 			goto done;
1122 	}
1123 
1124 	workdir = opendir(worktree->root_path);
1125 	if (workdir == NULL) {
1126 		err = got_error_from_errno();
1127 		goto done;
1128 	}
1129 	fdiff_cb.diff_old_new = status_old_new;
1130 	fdiff_cb.diff_old = status_old;
1131 	fdiff_cb.diff_new = status_new;
1132 	arg.fileindex = fileindex;
1133 	arg.worktree = worktree;
1134 	arg.repo = repo;
1135 	arg.status_cb = status_cb;
1136 	arg.status_arg = status_arg;
1137 	arg.cancel_cb = cancel_cb;
1138 	arg.cancel_arg = cancel_arg;
1139 	err = got_fileindex_diff_dir(fileindex, workdir, worktree->root_path,
1140 	    repo, &fdiff_cb, &arg);
1141 done:
1142 	if (workdir)
1143 		closedir(workdir);
1144 	free(fileindex_path);
1145 	got_fileindex_free(fileindex);
1146 	return err;
1147 }