1 /*
2  * Copyright (c) 2020 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/tree.h>
20 #include <sys/uio.h>
21 #include <sys/stat.h>
22 #include <sys/socket.h>
23 #include <sys/wait.h>
24 
25 #include <dirent.h>
26 #include <errno.h>
27 #include <fcntl.h>
28 #include <stdint.h>
29 #include <stdio.h>
30 #include <stdlib.h>
31 #include <string.h>
32 #include <limits.h>
33 #include <unistd.h>
34 
35 #include "got_error.h"
36 #include "got_cancel.h"
37 #include "got_object.h"
38 #include "got_reference.h"
39 #include "got_repository.h"
40 #include "got_repository_admin.h"
41 #include "got_opentemp.h"
42 #include "got_path.h"
43 
44 #include "got_lib_delta.h"
45 #include "got_lib_object.h"
46 #include "got_lib_object_idset.h"
47 #include "got_lib_object_cache.h"
48 #include "got_lib_pack.h"
49 #include "got_lib_privsep.h"
50 #include "got_lib_repository.h"
51 #include "got_lib_pack_create.h"
52 #include "got_lib_sha1.h"
53 #include "got_lib_lockfile.h"
54 
55 #ifndef nitems
56 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
57 #endif
58 
59 static const struct got_error *
60 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
61     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
62     struct got_repository *repo,
63     got_cancel_cb cancel_cb, void *cancel_arg)
64 {
65 	const struct got_error *err = NULL;
66 	const size_t alloc_chunksz = 256;
67 	size_t nalloc;
68 	struct got_reflist_entry *re;
69 	int i;
70 
71 	*ids = NULL;
72 	*nobjects = 0;
73 
74 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
75 	if (*ids == NULL)
76 		return got_error_from_errno("reallocarray");
77 	nalloc = alloc_chunksz;
78 
79 	TAILQ_FOREACH(re, refs, entry) {
80 		struct got_object_id *id;
81 
82 		if (cancel_cb) {
83 			err = cancel_cb(cancel_arg);
84 			if (err)
85 				goto done;
86 		}
87 
88 		err = got_ref_resolve(&id, repo, re->ref);
89 		if (err)
90 			goto done;
91 
92 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
93 			int obj_type;
94 			err = got_object_get_type(&obj_type, repo, id);
95 			if (err)
96 				goto done;
97 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
98 				free(id);
99 				id = NULL;
100 				continue;
101 			}
102 		}
103 
104 		if (nalloc <= *nobjects) {
105 			struct got_object_id **new;
106 			new = recallocarray(*ids, nalloc,
107 			    nalloc + alloc_chunksz,
108 			    sizeof(struct got_object_id *));
109 			if (new == NULL) {
110 				err = got_error_from_errno(
111 				    "recallocarray");
112 				goto done;
113 			}
114 			*ids = new;
115 			nalloc += alloc_chunksz;
116 		}
117 		(*ids)[*nobjects] = id;
118 		if ((*ids)[*nobjects] == NULL) {
119 			err = got_error_from_errno("got_object_id_dup");
120 			goto done;
121 		}
122 		(*nobjects)++;
123 	}
124 done:
125 	if (err) {
126 		for (i = 0; i < *nobjects; i++)
127 			free((*ids)[i]);
128 		free(*ids);
129 		*ids = NULL;
130 		*nobjects = 0;
131 	}
132 	return err;
133 }
134 
135 const struct got_error *
136 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
137     struct got_reflist_head *include_refs,
138     struct got_reflist_head *exclude_refs, struct got_repository *repo,
139     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
140     got_cancel_cb cancel_cb, void *cancel_arg)
141 {
142 	const struct got_error *err = NULL;
143 	struct got_object_id **ours = NULL, **theirs = NULL;
144 	int nours = 0, ntheirs = 0, packfd = -1, i;
145 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
146 	char *sha1_str = NULL;
147 
148 	*packfile = NULL;
149 	*pack_hash = NULL;
150 
151 	if (asprintf(&path, "%s/%s/packing.pack",
152 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
153 		err = got_error_from_errno("asprintf");
154 		goto done;
155 	}
156 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
157 	if (err)
158 		goto done;
159 
160 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
161 		err = got_error_from_errno2("fchmod", tmpfile_path);
162 		goto done;
163 	}
164 
165 	*packfile = fdopen(packfd, "w");
166 	if (*packfile == NULL) {
167 		err = got_error_from_errno2("fdopen", tmpfile_path);
168 		goto done;
169 	}
170 	packfd = -1;
171 
172 	err = get_reflist_object_ids(&ours, &nours,
173 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
174 	    include_refs, repo, cancel_cb, cancel_arg);
175 	if (err)
176 		goto done;
177 
178 	if (nours == 0) {
179 		err = got_error(GOT_ERR_CANNOT_PACK);
180 		goto done;
181 	}
182 
183 	if (!TAILQ_EMPTY(exclude_refs)) {
184 		err = get_reflist_object_ids(&theirs, &ntheirs,
185 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
186 		    exclude_refs, repo,
187 		    cancel_cb, cancel_arg);
188 		if (err)
189 			goto done;
190 	}
191 
192 	*pack_hash = calloc(1, sizeof(**pack_hash));
193 	if (*pack_hash == NULL) {
194 		err = got_error_from_errno("calloc");
195 		goto done;
196 	}
197 
198 	err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
199 	    ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
200 	    cancel_cb, cancel_arg);
201 	if (err)
202 		goto done;
203 
204 	err = got_object_id_str(&sha1_str, *pack_hash);
205 	if (err)
206 		goto done;
207 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
208 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
209 	    sha1_str) == -1) {
210 		err = got_error_from_errno("asprintf");
211 		goto done;
212 	}
213 
214 	if (fflush(*packfile) == -1) {
215 		err = got_error_from_errno("fflush");
216 		goto done;
217 	}
218 	if (fseek(*packfile, 0L, SEEK_SET) == -1) {
219 		err = got_error_from_errno("fseek");
220 		goto done;
221 	}
222 	if (rename(tmpfile_path, packfile_path) == -1) {
223 		err = got_error_from_errno3("rename", tmpfile_path,
224 		    packfile_path);
225 		goto done;
226 	}
227 	free(tmpfile_path);
228 	tmpfile_path = NULL;
229 done:
230 	for (i = 0; i < nours; i++)
231 		free(ours[i]);
232 	free(ours);
233 	for (i = 0; i < ntheirs; i++)
234 		free(theirs[i]);
235 	free(theirs);
236 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
237 		err = got_error_from_errno2("close", packfile_path);
238 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
239 		err = got_error_from_errno2("unlink", tmpfile_path);
240 	free(tmpfile_path);
241 	free(packfile_path);
242 	free(sha1_str);
243 	free(path);
244 	if (err) {
245 		free(*pack_hash);
246 		*pack_hash = NULL;
247 		if (*packfile)
248 			fclose(*packfile);
249 		*packfile = NULL;
250 	}
251 	return err;
252 }
253 
254 const struct got_error *
255 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
256     struct got_repository *repo,
257     got_pack_index_progress_cb progress_cb, void *progress_arg,
258     got_cancel_cb cancel_cb, void *cancel_arg)
259 {
260 	size_t i;
261 	char *path;
262 	int imsg_idxfds[2];
263 	int npackfd = -1, idxfd = -1, nidxfd = -1;
264 	int tmpfds[3];
265 	int idxstatus, done = 0;
266 	const struct got_error *err;
267 	struct imsgbuf idxibuf;
268 	pid_t idxpid;
269 	char *tmpidxpath = NULL;
270 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
271 	const char *repo_path = got_repo_get_path_git_dir(repo);
272 	struct stat sb;
273 
274 	for (i = 0; i < nitems(tmpfds); i++)
275 		tmpfds[i] = -1;
276 
277 	if (asprintf(&path, "%s/%s/indexing.idx",
278 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
279 		err = got_error_from_errno("asprintf");
280 		goto done;
281 	}
282 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
283 	free(path);
284 	if (err)
285 		goto done;
286 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
287 		err = got_error_from_errno2("fchmod", tmpidxpath);
288 		goto done;
289 	}
290 
291 	nidxfd = dup(idxfd);
292 	if (nidxfd == -1) {
293 		err = got_error_from_errno("dup");
294 		goto done;
295 	}
296 
297 	for (i = 0; i < nitems(tmpfds); i++) {
298 		tmpfds[i] = got_opentempfd();
299 		if (tmpfds[i] == -1) {
300 			err = got_error_from_errno("got_opentempfd");
301 			goto done;
302 		}
303 	}
304 
305 	err = got_object_id_str(&id_str, pack_hash);
306 	if (err)
307 		goto done;
308 
309 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
310 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
311 		err = got_error_from_errno("asprintf");
312 		goto done;
313 	}
314 
315 	if (fstat(fileno(packfile), &sb) == -1) {
316 		err = got_error_from_errno2("fstat", packfile_path);
317 		goto done;
318 	}
319 
320 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
321 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
322 		err = got_error_from_errno("asprintf");
323 		goto done;
324 	}
325 
326 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
327 		err = got_error_from_errno("socketpair");
328 		goto done;
329 	}
330 	idxpid = fork();
331 	if (idxpid == -1) {
332 		err= got_error_from_errno("fork");
333 		goto done;
334 	} else if (idxpid == 0)
335 		got_privsep_exec_child(imsg_idxfds,
336 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
337 	if (close(imsg_idxfds[1]) == -1) {
338 		err = got_error_from_errno("close");
339 		goto done;
340 	}
341 	imsg_init(&idxibuf, imsg_idxfds[0]);
342 
343 	npackfd = dup(fileno(packfile));
344 	if (npackfd == -1) {
345 		err = got_error_from_errno("dup");
346 		goto done;
347 	}
348 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
349 	    npackfd);
350 	if (err != NULL)
351 		goto done;
352 	npackfd = -1;
353 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
354 	if (err != NULL)
355 		goto done;
356 	nidxfd = -1;
357 	for (i = 0; i < nitems(tmpfds); i++) {
358 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
359 		if (err != NULL)
360 			goto done;
361 		tmpfds[i] = -1;
362 	}
363 	done = 0;
364 	while (!done) {
365 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
366 
367 		if (cancel_cb) {
368 			err = cancel_cb(cancel_arg);
369 			if (err)
370 				goto done;
371 		}
372 
373 		err = got_privsep_recv_index_progress(&done, &nobj_total,
374 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
375 		    &idxibuf);
376 		if (err != NULL)
377 			goto done;
378 		if (nobj_indexed != 0) {
379 			err = progress_cb(progress_arg, sb.st_size,
380 			    nobj_total, nobj_indexed, nobj_loose,
381 			    nobj_resolved);
382 			if (err)
383 				break;
384 		}
385 		imsg_clear(&idxibuf);
386 	}
387 	if (close(imsg_idxfds[0]) == -1) {
388 		err = got_error_from_errno("close");
389 		goto done;
390 	}
391 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
392 		err = got_error_from_errno("waitpid");
393 		goto done;
394 	}
395 
396 	if (rename(tmpidxpath, idxpath) == -1) {
397 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
398 		goto done;
399 	}
400 	free(tmpidxpath);
401 	tmpidxpath = NULL;
402 
403 done:
404 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
405 		err = got_error_from_errno2("unlink", tmpidxpath);
406 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
407 		err = got_error_from_errno("close");
408 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
409 		err = got_error_from_errno("close");
410 	for (i = 0; i < nitems(tmpfds); i++) {
411 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
412 			err = got_error_from_errno("close");
413 	}
414 	free(tmpidxpath);
415 	free(idxpath);
416 	free(packfile_path);
417 	return err;
418 }
419 
420 const struct got_error *
421 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
422     struct got_repository *repo, const char *packfile_path)
423 {
424 	const struct got_error *err = NULL;
425 	const char *packdir_path = NULL;
426 	char *packfile_name = NULL, *p, *dot;
427 	struct got_object_id id;
428 	int packfd = -1;
429 
430 	*packfile = NULL;
431 	*pack_hash = NULL;
432 
433 	packdir_path = got_repo_get_path_objects_pack(repo);
434 	if (packdir_path == NULL)
435 		return got_error_from_errno("got_repo_get_path_objects_pack");
436 
437 	if (!got_path_is_child(packfile_path, packdir_path,
438 	    strlen(packdir_path))) {
439 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
440 		goto done;
441 
442 	}
443 
444 	err = got_path_basename(&packfile_name, packfile_path);
445 	if (err)
446 		goto done;
447 	p = packfile_name;
448 
449 	if (strncmp(p, "pack-", 5) != 0) {
450 		err = got_error_fmt(GOT_ERR_BAD_PATH,
451 		    "'%s' is not a valid pack file name",
452 		    packfile_name);
453 		goto done;
454 	}
455 	p += 5;
456 	dot = strchr(p, '.');
457 	if (dot == NULL) {
458 		err = got_error_fmt(GOT_ERR_BAD_PATH,
459 		    "'%s' is not a valid pack file name",
460 		    packfile_name);
461 		goto done;
462 	}
463 	if (strcmp(dot + 1, "pack") != 0) {
464 		err = got_error_fmt(GOT_ERR_BAD_PATH,
465 		    "'%s' is not a valid pack file name",
466 		    packfile_name);
467 		goto done;
468 	}
469 	*dot = '\0';
470 	if (!got_parse_sha1_digest(id.sha1, p)) {
471 		err = got_error_fmt(GOT_ERR_BAD_PATH,
472 		    "'%s' is not a valid pack file name",
473 		    packfile_name);
474 		goto done;
475 	}
476 
477 	*pack_hash = got_object_id_dup(&id);
478 	if (*pack_hash == NULL) {
479 		err = got_error_from_errno("got_object_id_dup");
480 		goto done;
481 	}
482 
483 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW);
484 	if (packfd == -1) {
485 		err = got_error_from_errno2("open", packfile_path);
486 		goto done;
487 	}
488 
489 	*packfile = fdopen(packfd, "r");
490 	if (*packfile == NULL) {
491 		err = got_error_from_errno2("fdopen", packfile_path);
492 		goto done;
493 	}
494 	packfd = -1;
495 done:
496 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
497 		err = got_error_from_errno2("close", packfile_path);
498 	free(packfile_name);
499 	if (err) {
500 		free(*pack_hash);
501 		*pack_hash = NULL;
502 	}
503 	return err;
504 }
505 
506 const struct got_error *
507 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
508     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
509     got_cancel_cb cancel_cb, void *cancel_arg)
510 {
511 	const struct got_error *err = NULL;
512 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
513 	struct got_packidx *packidx = NULL;
514 	struct got_pack *pack = NULL;
515 	uint32_t nobj, i;
516 
517 	err = got_object_id_str(&id_str, pack_hash);
518 	if (err)
519 		goto done;
520 
521 	if (asprintf(&packpath, "%s/pack-%s.pack",
522 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
523 		err = got_error_from_errno("asprintf");
524 		goto done;
525 	}
526 	if (asprintf(&idxpath, "%s/pack-%s.idx",
527 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
528 		err = got_error_from_errno("asprintf");
529 		goto done;
530 	}
531 
532 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
533 	if (err)
534 		goto done;
535 
536 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
537 	if (err)
538 		goto done;
539 
540 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
541 	for (i = 0; i < nobj; i++) {
542 		struct got_packidx_object_id *oid;
543 		struct got_object_id id, base_id;
544 		off_t offset, base_offset = 0;
545 		uint8_t type;
546 		uint64_t size;
547 		size_t tslen, len;
548 
549 		if (cancel_cb) {
550 			err = cancel_cb(cancel_arg);
551 			if (err)
552 				break;
553 		}
554 		oid = &packidx->hdr.sorted_ids[i];
555 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
556 
557 		offset = got_packidx_get_object_offset(packidx, i);
558 		if (offset == -1) {
559 			err = got_error(GOT_ERR_BAD_PACKIDX);
560 			goto done;
561 		}
562 
563 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
564 		    pack, offset);
565 		if (err)
566 			goto done;
567 
568 		switch (type) {
569 		case GOT_OBJ_TYPE_OFFSET_DELTA:
570 			err = got_pack_parse_offset_delta(&base_offset, &len,
571 			    pack, offset, tslen);
572 			if (err)
573 				goto done;
574 			break;
575 		case GOT_OBJ_TYPE_REF_DELTA:
576 			err = got_pack_parse_ref_delta(&base_id,
577 			    pack, offset, tslen);
578 			if (err)
579 				goto done;
580 			break;
581 		}
582 		err = (*list_cb)(list_arg, &id, type, offset, size,
583 		    base_offset, &base_id);
584 		if (err)
585 			goto done;
586 	}
587 
588 done:
589 	free(id_str);
590 	free(idxpath);
591 	free(packpath);
592 	if (packidx)
593 		got_packidx_close(packidx);
594 	return err;
595 }
596 
597 static const struct got_error *
598 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
599     got_cleanup_progress_cb progress_cb, void *progress_arg,
600     struct got_repository *repo)
601 {
602 	const struct got_error *err = NULL;
603 	char *path_objects = NULL, *path = NULL;
604 	DIR *dir = NULL;
605 	struct got_object *obj = NULL;
606 	struct got_object_id id;
607 	int i, fd = -1;
608 	struct stat sb;
609 
610 	*ondisk_size = 0;
611 	*loose_ids = got_object_idset_alloc();
612 	if (*loose_ids == NULL)
613 		return got_error_from_errno("got_object_idset_alloc");
614 
615 	path_objects = got_repo_get_path_objects(repo);
616 	if (path_objects == NULL) {
617 		err = got_error_from_errno("got_repo_get_path_objects");
618 		goto done;
619 	}
620 
621 	for (i = 0; i <= 0xff; i++) {
622 		struct dirent *dent;
623 
624 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
625 			err = got_error_from_errno("asprintf");
626 			break;
627 		}
628 
629 		dir = opendir(path);
630 		if (dir == NULL) {
631 			if (errno == ENOENT) {
632 				err = NULL;
633 				continue;
634 			}
635 			err = got_error_from_errno2("opendir", path);
636 			break;
637 		}
638 
639 		while ((dent = readdir(dir)) != NULL) {
640 			char *id_str;
641 
642 			if (strcmp(dent->d_name, ".") == 0 ||
643 			    strcmp(dent->d_name, "..") == 0)
644 				continue;
645 
646 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
647 				err = got_error_from_errno("asprintf");
648 				goto done;
649 			}
650 
651 			memset(&id, 0, sizeof(id));
652 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
653 				free(id_str);
654 				continue;
655 			}
656 			free(id_str);
657 
658 			err = got_object_open_loose_fd(&fd, &id, repo);
659 			if (err)
660 				goto done;
661 			if (fstat(fd, &sb) == -1) {
662 				err = got_error_from_errno("fstat");
663 				goto done;
664 			}
665 			err = got_object_read_header_privsep(&obj, &id, repo,
666 			    fd);
667 			if (err)
668 				goto done;
669 			fd = -1; /* already closed */
670 
671 			switch (obj->type) {
672 			case GOT_OBJ_TYPE_COMMIT:
673 			case GOT_OBJ_TYPE_TREE:
674 			case GOT_OBJ_TYPE_BLOB:
675 			case GOT_OBJ_TYPE_TAG:
676 				break;
677 			default:
678 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
679 				    "%d", obj->type);
680 				goto done;
681 			}
682 			got_object_close(obj);
683 			obj = NULL;
684 			(*ondisk_size) += sb.st_size;
685 			err = got_object_idset_add(*loose_ids, &id, NULL);
686 			if (err)
687 				goto done;
688 			if (progress_cb) {
689 				err = progress_cb(progress_arg,
690 				    got_object_idset_num_elements(*loose_ids),
691 				    -1, -1);
692 				if (err)
693 					goto done;
694 			}
695 		}
696 
697 		if (closedir(dir) != 0) {
698 			err = got_error_from_errno("closedir");
699 			goto done;
700 		}
701 		dir = NULL;
702 
703 		free(path);
704 		path = NULL;
705 	}
706 done:
707 	if (dir && closedir(dir) != 0 && err == NULL)
708 		err = got_error_from_errno("closedir");
709 	if (fd != -1 && close(fd) == -1 && err == NULL)
710 		err = got_error_from_errno("close");
711 	if (err) {
712 		got_object_idset_free(*loose_ids);
713 		*loose_ids = NULL;
714 	}
715 	if (obj)
716 		got_object_close(obj);
717 	free(path_objects);
718 	free(path);
719 	return err;
720 }
721 
722 static const struct got_error *
723 preserve_loose_object(struct got_object_idset *loose_ids,
724     struct got_object_id *id, struct got_repository *repo, int *npacked)
725 {
726 	const struct got_error *err = NULL;
727 	struct got_object *obj;
728 
729 	if (!got_object_idset_contains(loose_ids, id))
730 		return NULL;
731 
732 	/*
733 	 * Try to open this object from a pack file. This ensures that
734 	 * we do in fact have a valid packed copy of the object. Otherwise
735 	 * we should not delete the loose representation of this object.
736 	 */
737 	err = got_object_open_packed(&obj, id, repo);
738 	if (err == NULL) {
739 		got_object_close(obj);
740 		/*
741 		 * The object is referenced and packed.
742 		 * We can purge the redundantly stored loose object.
743 		 */
744 		(*npacked)++;
745 		return NULL;
746 	} else if (err->code != GOT_ERR_NO_OBJ)
747 		return err;
748 
749 	/*
750 	 * This object is referenced and not packed.
751 	 * Remove it from our purge set.
752 	 */
753 	return got_object_idset_remove(NULL, loose_ids, id);
754 }
755 
756 static const struct got_error *
757 load_tree_entries(struct got_object_id_queue *ids,
758     struct got_object_idset *loose_ids,
759     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
760     const char *dpath, struct got_repository *repo, int *npacked,
761     got_cancel_cb cancel_cb, void *cancel_arg)
762 {
763 	const struct got_error *err;
764 	struct got_tree_object *tree;
765 	char *p = NULL;
766 	int i;
767 
768 	err = got_object_open_as_tree(&tree, repo, tree_id);
769 	if (err)
770 		return err;
771 
772 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
773 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
774 		struct got_object_id *id = got_tree_entry_get_id(e);
775 		mode_t mode = got_tree_entry_get_mode(e);
776 
777 		if (cancel_cb) {
778 			err = (*cancel_cb)(cancel_arg);
779 			if (err)
780 				break;
781 		}
782 
783 		if (got_object_tree_entry_is_symlink(e) ||
784 		    got_object_tree_entry_is_submodule(e) ||
785 		    got_object_idset_contains(traversed_ids, id))
786 			continue;
787 
788 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
789 		    got_tree_entry_get_name(e)) == -1) {
790 			err = got_error_from_errno("asprintf");
791 			break;
792 		}
793 
794 		if (S_ISDIR(mode)) {
795 			struct got_object_qid *qid;
796 			err = got_object_qid_alloc(&qid, id);
797 			if (err)
798 				break;
799 			STAILQ_INSERT_TAIL(ids, qid, entry);
800 		} else if (S_ISREG(mode)) {
801 			/* This blob is referenced. */
802 			err = preserve_loose_object(loose_ids, id, repo,
803 			    npacked);
804 			if (err)
805 				break;
806 			err = got_object_idset_add(traversed_ids, id, NULL);
807 			if (err)
808 				break;
809 			
810 		}
811 		free(p);
812 		p = NULL;
813 	}
814 
815 	got_object_tree_close(tree);
816 	free(p);
817 	return err;
818 }
819 
820 static const struct got_error *
821 load_tree(struct got_object_idset *loose_ids,
822     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
823     const char *dpath, struct got_repository *repo, int *npacked,
824     got_cancel_cb cancel_cb, void *cancel_arg)
825 {
826 	const struct got_error *err = NULL;
827 	struct got_object_id_queue tree_ids;
828 	struct got_object_qid *qid;
829 
830 	err = got_object_qid_alloc(&qid, tree_id);
831 	if (err)
832 		return err;
833 
834 	STAILQ_INIT(&tree_ids);
835 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
836 
837 	while (!STAILQ_EMPTY(&tree_ids)) {
838 		if (cancel_cb) {
839 			err = (*cancel_cb)(cancel_arg);
840 			if (err)
841 				break;
842 		}
843 
844 		qid = STAILQ_FIRST(&tree_ids);
845 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
846 
847 		if (got_object_idset_contains(traversed_ids, qid->id)) {
848 			got_object_qid_free(qid);
849 			continue;
850 		}
851 
852 		err = got_object_idset_add(traversed_ids, qid->id, NULL);
853 		if (err) {
854 			got_object_qid_free(qid);
855 			break;
856 		}
857 
858 		/* This tree is referenced. */
859 		err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
860 		if (err)
861 			break;
862 
863 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
864 		    qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
865 		got_object_qid_free(qid);
866 		if (err)
867 			break;
868 	}
869 
870 	got_object_id_queue_free(&tree_ids);
871 	return err;
872 }
873 
874 static const struct got_error *
875 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
876     int *npacked, struct got_object_idset *traversed_ids,
877     struct got_object_id *id, struct got_repository *repo,
878     got_cleanup_progress_cb progress_cb, void *progress_arg, int nloose,
879     got_cancel_cb cancel_cb, void *cancel_arg)
880 {
881 	const struct got_error *err;
882 	struct got_commit_object *commit = NULL;
883 	struct got_tag_object *tag = NULL;
884 	struct got_object_id *tree_id = NULL;
885 	struct got_object_id_queue ids;
886 	struct got_object_qid *qid;
887 	int obj_type;
888 
889 	err = got_object_qid_alloc(&qid, id);
890 	if (err)
891 		return err;
892 
893 	STAILQ_INIT(&ids);
894 	STAILQ_INSERT_TAIL(&ids, qid, entry);
895 
896 	while (!STAILQ_EMPTY(&ids)) {
897 		if (cancel_cb) {
898 			err = (*cancel_cb)(cancel_arg);
899 			if (err)
900 				break;
901 		}
902 
903 		qid = STAILQ_FIRST(&ids);
904 		STAILQ_REMOVE_HEAD(&ids, entry);
905 
906 		if (got_object_idset_contains(traversed_ids, qid->id)) {
907 			got_object_qid_free(qid);
908 			qid = NULL;
909 			continue;
910 		}
911 
912 		err = got_object_idset_add(traversed_ids, qid->id, NULL);
913 		if (err)
914 			break;
915 
916 		/* This commit or tag is referenced. */
917 		err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
918 		if (err)
919 			break;
920 
921 		err = got_object_get_type(&obj_type, repo, qid->id);
922 		if (err)
923 			break;
924 		switch (obj_type) {
925 		case GOT_OBJ_TYPE_COMMIT:
926 			err = got_object_open_as_commit(&commit, repo, qid->id);
927 			if (err)
928 				goto done;
929 			break;
930 		case GOT_OBJ_TYPE_TAG:
931 			err = got_object_open_as_tag(&tag, repo, qid->id);
932 			if (err)
933 				goto done;
934 			break;
935 		default:
936 			/* should not happen */
937 			err = got_error(GOT_ERR_OBJ_TYPE);
938 			goto done;
939 		}
940 
941 		/* Find a tree object to scan. */
942 		if (commit) {
943 			tree_id = got_object_commit_get_tree_id(commit);
944 		} else if (tag) {
945 			obj_type = got_object_tag_get_object_type(tag);
946 			switch (obj_type) {
947 			case GOT_OBJ_TYPE_COMMIT:
948 				err = got_object_open_as_commit(&commit, repo,
949 				    got_object_tag_get_object_id(tag));
950 				if (err)
951 					goto done;
952 				tree_id = got_object_commit_get_tree_id(commit);
953 				break;
954 			case GOT_OBJ_TYPE_TREE:
955 				tree_id = got_object_tag_get_object_id(tag);
956 				break;
957 			default:
958 				/*
959 				 * Tag points at something other than a
960 				 * commit or tree. Leave this weird tag object
961 				 * and the object it points to on disk.
962 				 */
963 				err = got_object_idset_remove(NULL, loose_ids,
964 				    qid->id);
965 				if (err && err->code != GOT_ERR_NO_OBJ)
966 					goto done;
967 				err = got_object_idset_remove(NULL, loose_ids,
968 				    got_object_tag_get_object_id(tag));
969 				if (err && err->code != GOT_ERR_NO_OBJ)
970 					goto done;
971 				err = NULL;
972 				break;
973 			}
974 		}
975 
976 		if (tree_id) {
977 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
978 			    repo, npacked, cancel_cb, cancel_arg);
979 			if (err)
980 				break;
981 		}
982 
983 		if (commit || tag)
984 			(*ncommits)++; /* scanned tags are counted as commits */
985 
986 		if (progress_cb) {
987 			err = progress_cb(progress_arg, nloose, *ncommits, -1);
988 			if (err)
989 				break;
990 		}
991  
992 		if (commit) {
993 			/* Find parent commits to scan. */
994 			const struct got_object_id_queue *parent_ids;
995 			parent_ids = got_object_commit_get_parent_ids(commit);
996 			err = got_object_id_queue_copy(parent_ids, &ids);
997 			if (err)
998 				break;
999 			got_object_commit_close(commit);
1000 			commit = NULL;
1001 		}
1002 		if (tag) {
1003 			got_object_tag_close(tag);
1004 			tag = NULL;
1005 		}
1006 		got_object_qid_free(qid);
1007 		qid = NULL;
1008 	}
1009 done:
1010 	if (qid)
1011 		got_object_qid_free(qid);
1012 	if (commit)
1013 		got_object_commit_close(commit);
1014 	if (tag)
1015 		got_object_tag_close(tag);
1016 	got_object_id_queue_free(&ids);
1017 	return err;
1018 }
1019 
1020 struct purge_loose_object_arg {
1021 	struct got_repository *repo;
1022 	got_cleanup_progress_cb progress_cb;
1023 	void *progress_arg;
1024 	int nloose;
1025 	int ncommits;
1026 	int npurged;
1027 	off_t size_purged;
1028 	int dry_run;
1029 	time_t max_mtime;
1030 	int ignore_mtime;
1031 };
1032 
1033 static const struct got_error *
1034 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1035 {
1036 	struct purge_loose_object_arg *a = arg;
1037 	const struct got_error *err, *unlock_err = NULL;
1038 	char *path = NULL;
1039 	int fd = -1;
1040 	struct stat sb;
1041 	struct got_lockfile *lf = NULL;
1042 
1043 	err = got_object_get_path(&path, id, a->repo);
1044 	if (err)
1045 		return err;
1046 
1047 	err = got_object_open_loose_fd(&fd, id, a->repo);
1048 	if (err)
1049 		goto done;
1050 
1051 	if (fstat(fd, &sb) == -1) {
1052 		err = got_error_from_errno("fstat");
1053 		goto done;
1054 	}
1055 
1056 	/*
1057 	 * Do not delete objects which are younger than our maximum
1058 	 * modification time threshold. This prevents a race where
1059 	 * new objects which are being added to the repository
1060 	 * concurrently would be deleted.
1061 	 */
1062 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1063 		if (!a->dry_run) {
1064 			err = got_lockfile_lock(&lf, path, -1);
1065 			if (err)
1066 				goto done;
1067 			if (unlink(path) == -1) {
1068 				err = got_error_from_errno2("unlink", path);
1069 				goto done;
1070 			}
1071 		}
1072 
1073 		a->npurged++;
1074 		a->size_purged += sb.st_size;
1075 		if (a->progress_cb) {
1076 			err = a->progress_cb(a->progress_arg, a->nloose,
1077 			    a->ncommits, a->npurged);
1078 		}
1079 	}
1080 done:
1081 	if (fd != -1 && close(fd) == -1 && err == NULL)
1082 		err = got_error_from_errno("close");
1083 	free(path);
1084 	if (lf)
1085 		unlock_err = got_lockfile_unlock(lf, -1);
1086 	return err ? err : unlock_err;
1087 }
1088 
1089 const struct got_error *
1090 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1091     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1092     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1093     got_cancel_cb cancel_cb, void *cancel_arg)
1094 {
1095 	const struct got_error *err;
1096 	struct got_object_idset *loose_ids;
1097 	struct got_object_idset *traversed_ids;
1098 	struct got_object_id **referenced_ids;
1099 	int i, nreferenced, nloose, ncommits = 0;
1100 	struct got_reflist_head refs;
1101 	struct got_reflist_entry *re;
1102 	struct purge_loose_object_arg arg;
1103 	time_t max_mtime = 0;
1104 
1105 	TAILQ_INIT(&refs);
1106 
1107 	*size_before = 0;
1108 	*size_after = 0;
1109 	*npacked = 0;
1110 
1111 	err = get_loose_object_ids(&loose_ids, size_before,
1112 	    progress_cb, progress_arg, repo);
1113 	if (err)
1114 		return err;
1115 	nloose = got_object_idset_num_elements(loose_ids);
1116 	if (nloose == 0) {
1117 		got_object_idset_free(loose_ids);
1118 		return NULL;
1119 	}
1120 
1121 	traversed_ids = got_object_idset_alloc();
1122 	if (traversed_ids == NULL) {
1123 		err = got_error_from_errno("got_object_idset_alloc");
1124 		goto done;
1125 	}
1126 
1127 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1128 	if (err)
1129 		goto done;
1130 	if (!ignore_mtime) {
1131 		TAILQ_FOREACH(re, &refs, entry) {
1132 			time_t mtime = got_ref_get_mtime(re->ref);
1133 			if (mtime > max_mtime)
1134 				max_mtime = mtime;
1135 		}
1136 		/*
1137 		 * For safety, keep objects created within 10 minutes
1138 		 * before the youngest reference was created.
1139 		 */
1140 		if (max_mtime >= 600)
1141 			max_mtime -= 600;
1142 	}
1143 
1144 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1145 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1146 	    &refs, repo, cancel_cb, cancel_arg);
1147 	if (err)
1148 		goto done;
1149 
1150 	for (i = 0; i < nreferenced; i++) {
1151 		struct got_object_id *id = referenced_ids[i];
1152 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1153 		    traversed_ids, id, repo, progress_cb, progress_arg, nloose, 
1154 		    cancel_cb, cancel_arg);
1155 		if (err)
1156 			goto done;
1157 	}
1158 
1159 	/* Produce a final progress report in case no objects can be purged. */
1160 	if (got_object_idset_num_elements(loose_ids) == 0 && progress_cb) {
1161 		err = progress_cb(progress_arg, nloose, ncommits, 0);
1162 		if (err)
1163 			goto done;
1164 	}
1165 
1166 	/* Any remaining loose objects are unreferenced and can be purged. */
1167 	arg.repo = repo;
1168 	arg.progress_arg = progress_arg;
1169 	arg.progress_cb = progress_cb;
1170 	arg.nloose = nloose;
1171 	arg.npurged = 0;
1172 	arg.size_purged = 0;
1173 	arg.ncommits = ncommits;
1174 	arg.dry_run = dry_run;
1175 	arg.max_mtime = max_mtime;
1176 	arg.ignore_mtime = ignore_mtime;
1177 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1178 	if (err)
1179 		goto done;
1180 	*size_after = *size_before - arg.size_purged;
1181 done:
1182 	got_object_idset_free(loose_ids);
1183 	got_object_idset_free(traversed_ids);
1184 	return err;
1185 }
1186 
1187 static const struct got_error *
1188 remove_packidx(int dir_fd, const char *relpath)
1189 {
1190 	const struct got_error *err, *unlock_err;
1191 	struct got_lockfile *lf;
1192 
1193 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1194 	if (err)
1195 		return err;
1196 	if (unlinkat(dir_fd, relpath, 0) == -1)
1197 		err = got_error_from_errno("unlinkat");
1198 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1199 	return err ? err : unlock_err;
1200 }
1201 
1202 const struct got_error *
1203 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1204     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1205     got_cancel_cb cancel_cb, void *cancel_arg)
1206 {
1207 	const struct got_error *err;
1208 	DIR *packdir = NULL;
1209 	struct dirent *dent;
1210 	char *pack_relpath = NULL;
1211 	int packdir_fd;
1212 	struct stat sb;
1213 
1214 	packdir_fd = openat(got_repo_get_fd(repo),
1215 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1216 	if (packdir_fd == -1) {
1217 		if (errno == ENOENT)
1218 			return NULL;
1219 		return got_error_from_errno_fmt("openat: %s/%s",
1220 		    got_repo_get_path_git_dir(repo),
1221 		    GOT_OBJECTS_PACK_DIR);
1222 	}
1223 
1224 	packdir = fdopendir(packdir_fd);
1225 	if (packdir == NULL) {
1226 		err = got_error_from_errno("fdopendir");
1227 		goto done;
1228 	}
1229 
1230 	while ((dent = readdir(packdir)) != NULL) {
1231 		if (cancel_cb) {
1232 			err = cancel_cb(cancel_arg);
1233 			if (err)
1234 				goto done;
1235 		}
1236 
1237 		if (!got_repo_is_packidx_filename(dent->d_name,
1238 		    strlen(dent->d_name)))
1239 			continue;
1240 
1241 		err = got_packidx_get_packfile_path(&pack_relpath,
1242 		    dent->d_name);
1243 		if (err)
1244 			goto done;
1245 
1246 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1247 			free(pack_relpath);
1248 			pack_relpath = NULL;
1249 			continue;
1250 		}
1251 		if (errno != ENOENT) {
1252 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1253 			    got_repo_get_path_git_dir(repo),
1254 			    GOT_OBJECTS_PACK_DIR,
1255 			    pack_relpath);
1256 			goto done;
1257 		}
1258 
1259 		if (!dry_run) {
1260 			err = remove_packidx(packdir_fd, dent->d_name);
1261 			if (err)
1262 				goto done;
1263 		}
1264 		if (progress_cb) {
1265 			char *path;
1266 			if (asprintf(&path, "%s/%s/%s",
1267 			    got_repo_get_path_git_dir(repo),
1268 			    GOT_OBJECTS_PACK_DIR,
1269 			    dent->d_name) == -1) {
1270 				err = got_error_from_errno("asprintf");
1271 				goto done;
1272 			}
1273 			err = progress_cb(progress_arg, path);
1274 			free(path);
1275 			if (err)
1276 				goto done;
1277 		}
1278 		free(pack_relpath);
1279 		pack_relpath = NULL;
1280 	}
1281 done:
1282 	if (packdir && closedir(packdir) != 0 && err == NULL)
1283 		err = got_error_from_errno("closedir");
1284 	free(pack_relpath);
1285 	return err;
1286 }