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 "got_compat.h"
18 
19 #include <sys/types.h>
20 #include <sys/queue.h>
21 #include <sys/uio.h>
22 #include <sys/stat.h>
23 #include <sys/socket.h>
24 #include <sys/wait.h>
25 
26 #include <dirent.h>
27 #include <errno.h>
28 #include <fcntl.h>
29 #include <stdint.h>
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <limits.h>
34 #include <unistd.h>
35 
36 #include "got_error.h"
37 #include "got_cancel.h"
38 #include "got_object.h"
39 #include "got_reference.h"
40 #include "got_repository.h"
41 #include "got_repository_admin.h"
42 #include "got_opentemp.h"
43 #include "got_path.h"
44 
45 #include "got_lib_delta.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_idset.h"
48 #include "got_lib_object_cache.h"
49 #include "got_lib_pack.h"
50 #include "got_lib_privsep.h"
51 #include "got_lib_repository.h"
52 #include "got_lib_ratelimit.h"
53 #include "got_lib_pack_create.h"
54 #include "got_lib_hash.h"
55 #include "got_lib_lockfile.h"
56 
57 #ifndef nitems
58 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
59 #endif
60 
61 static const struct got_error *
62 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
63     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
64     struct got_repository *repo,
65     got_cancel_cb cancel_cb, void *cancel_arg)
66 {
67 	const struct got_error *err = NULL;
68 	const size_t alloc_chunksz = 256;
69 	size_t nalloc;
70 	struct got_reflist_entry *re;
71 	int i;
72 
73 	*ids = NULL;
74 	*nobjects = 0;
75 
76 	err = got_reflist_sort(refs,
77 	    got_ref_cmp_by_commit_timestamp_descending, repo);
78 	if (err)
79 		return err;
80 
81 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
82 	if (*ids == NULL)
83 		return got_error_from_errno("reallocarray");
84 	nalloc = alloc_chunksz;
85 
86 	TAILQ_FOREACH(re, refs, entry) {
87 		struct got_object_id *id;
88 
89 		if (cancel_cb) {
90 			err = cancel_cb(cancel_arg);
91 			if (err)
92 				goto done;
93 		}
94 
95 		err = got_ref_resolve(&id, repo, re->ref);
96 		if (err)
97 			goto done;
98 
99 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
100 			int obj_type;
101 			err = got_object_get_type(&obj_type, repo, id);
102 			if (err)
103 				goto done;
104 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
105 				free(id);
106 				id = NULL;
107 				continue;
108 			}
109 		}
110 
111 		if (nalloc <= *nobjects) {
112 			struct got_object_id **new;
113 			new = recallocarray(*ids, nalloc,
114 			    nalloc + alloc_chunksz,
115 			    sizeof(struct got_object_id *));
116 			if (new == NULL) {
117 				err = got_error_from_errno(
118 				    "recallocarray");
119 				goto done;
120 			}
121 			*ids = new;
122 			nalloc += alloc_chunksz;
123 		}
124 		(*ids)[*nobjects] = id;
125 		if ((*ids)[*nobjects] == NULL) {
126 			err = got_error_from_errno("got_object_id_dup");
127 			goto done;
128 		}
129 		(*nobjects)++;
130 	}
131 done:
132 	if (err) {
133 		for (i = 0; i < *nobjects; i++)
134 			free((*ids)[i]);
135 		free(*ids);
136 		*ids = NULL;
137 		*nobjects = 0;
138 	}
139 	return err;
140 }
141 
142 const struct got_error *
143 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
144     struct got_reflist_head *include_refs,
145     struct got_reflist_head *exclude_refs, struct got_repository *repo,
146     int loose_obj_only, int force_refdelta,
147     got_pack_progress_cb progress_cb, void *progress_arg,
148     got_cancel_cb cancel_cb, void *cancel_arg)
149 {
150 	const struct got_error *err = NULL;
151 	struct got_object_id **ours = NULL, **theirs = NULL;
152 	int nours = 0, ntheirs = 0, packfd = -1, i;
153 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
154 	char *sha1_str = NULL;
155 	FILE *delta_cache = NULL;
156 	struct got_ratelimit rl;
157 
158 	*packfile = NULL;
159 	*pack_hash = NULL;
160 
161 	got_ratelimit_init(&rl, 0, 500);
162 
163 	if (asprintf(&path, "%s/%s/packing.pack",
164 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
165 		err = got_error_from_errno("asprintf");
166 		goto done;
167 	}
168 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path, "");
169 	if (err)
170 		goto done;
171 
172 	if (fchmod(packfd, GOT_DEFAULT_PACK_MODE) == -1) {
173 		err = got_error_from_errno2("fchmod", tmpfile_path);
174 		goto done;
175 	}
176 
177 	delta_cache = got_opentemp();
178 	if (delta_cache == NULL) {
179 		err = got_error_from_errno("got_opentemp");
180 		goto done;
181 	}
182 
183 	err = get_reflist_object_ids(&ours, &nours,
184 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
185 	    include_refs, repo, cancel_cb, cancel_arg);
186 	if (err)
187 		goto done;
188 
189 	if (nours == 0) {
190 		err = got_error(GOT_ERR_CANNOT_PACK);
191 		goto done;
192 	}
193 
194 	if (!TAILQ_EMPTY(exclude_refs)) {
195 		err = get_reflist_object_ids(&theirs, &ntheirs,
196 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
197 		    exclude_refs, repo,
198 		    cancel_cb, cancel_arg);
199 		if (err)
200 			goto done;
201 	}
202 
203 	*pack_hash = calloc(1, sizeof(**pack_hash));
204 	if (*pack_hash == NULL) {
205 		err = got_error_from_errno("calloc");
206 		goto done;
207 	}
208 
209 	err = got_pack_create((*pack_hash)->sha1, packfd, delta_cache,
210 	    theirs, ntheirs, ours, nours, repo, loose_obj_only,
211 	    0, force_refdelta, progress_cb, progress_arg, &rl,
212 	    cancel_cb, cancel_arg);
213 	if (err)
214 		goto done;
215 
216 	err = got_object_id_str(&sha1_str, *pack_hash);
217 	if (err)
218 		goto done;
219 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
220 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
221 	    sha1_str) == -1) {
222 		err = got_error_from_errno("asprintf");
223 		goto done;
224 	}
225 
226 	if (lseek(packfd, 0L, SEEK_SET) == -1) {
227 		err = got_error_from_errno("lseek");
228 		goto done;
229 	}
230 	if (rename(tmpfile_path, packfile_path) == -1) {
231 		err = got_error_from_errno3("rename", tmpfile_path,
232 		    packfile_path);
233 		goto done;
234 	}
235 	free(tmpfile_path);
236 	tmpfile_path = NULL;
237 
238 	*packfile = fdopen(packfd, "w");
239 	if (*packfile == NULL) {
240 		err = got_error_from_errno2("fdopen", tmpfile_path);
241 		goto done;
242 	}
243 	packfd = -1;
244 done:
245 	for (i = 0; i < nours; i++)
246 		free(ours[i]);
247 	free(ours);
248 	for (i = 0; i < ntheirs; i++)
249 		free(theirs[i]);
250 	free(theirs);
251 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
252 		err = got_error_from_errno2("close", packfile_path);
253 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
254 		err = got_error_from_errno("fclose");
255 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
256 		err = got_error_from_errno2("unlink", tmpfile_path);
257 	free(tmpfile_path);
258 	free(packfile_path);
259 	free(sha1_str);
260 	free(path);
261 	if (err) {
262 		free(*pack_hash);
263 		*pack_hash = NULL;
264 		if (*packfile)
265 			fclose(*packfile);
266 		*packfile = NULL;
267 	}
268 	return err;
269 }
270 
271 const struct got_error *
272 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
273     struct got_repository *repo,
274     got_pack_index_progress_cb progress_cb, void *progress_arg,
275     got_cancel_cb cancel_cb, void *cancel_arg)
276 {
277 	size_t i;
278 	char *path;
279 	int imsg_idxfds[2];
280 	int npackfd = -1, idxfd = -1, nidxfd = -1;
281 	int tmpfds[3];
282 	int idxstatus, done = 0;
283 	const struct got_error *err;
284 	struct imsgbuf idxibuf;
285 	pid_t idxpid;
286 	char *tmpidxpath = NULL;
287 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
288 	const char *repo_path = got_repo_get_path_git_dir(repo);
289 	struct stat sb;
290 
291 	for (i = 0; i < nitems(tmpfds); i++)
292 		tmpfds[i] = -1;
293 
294 	if (asprintf(&path, "%s/%s/indexing.idx",
295 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
296 		err = got_error_from_errno("asprintf");
297 		goto done;
298 	}
299 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
300 	free(path);
301 	if (err)
302 		goto done;
303 	if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
304 		err = got_error_from_errno2("fchmod", tmpidxpath);
305 		goto done;
306 	}
307 
308 	nidxfd = dup(idxfd);
309 	if (nidxfd == -1) {
310 		err = got_error_from_errno("dup");
311 		goto done;
312 	}
313 
314 	for (i = 0; i < nitems(tmpfds); i++) {
315 		tmpfds[i] = got_opentempfd();
316 		if (tmpfds[i] == -1) {
317 			err = got_error_from_errno("got_opentempfd");
318 			goto done;
319 		}
320 	}
321 
322 	err = got_object_id_str(&id_str, pack_hash);
323 	if (err)
324 		goto done;
325 
326 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
327 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
328 		err = got_error_from_errno("asprintf");
329 		goto done;
330 	}
331 
332 	if (fstat(fileno(packfile), &sb) == -1) {
333 		err = got_error_from_errno2("fstat", packfile_path);
334 		goto done;
335 	}
336 
337 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
338 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
339 		err = got_error_from_errno("asprintf");
340 		goto done;
341 	}
342 
343 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
344 		err = got_error_from_errno("socketpair");
345 		goto done;
346 	}
347 	idxpid = fork();
348 	if (idxpid == -1) {
349 		err= got_error_from_errno("fork");
350 		goto done;
351 	} else if (idxpid == 0)
352 		got_privsep_exec_child(imsg_idxfds,
353 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
354 	if (close(imsg_idxfds[1]) == -1) {
355 		err = got_error_from_errno("close");
356 		goto done;
357 	}
358 	imsg_init(&idxibuf, imsg_idxfds[0]);
359 
360 	npackfd = dup(fileno(packfile));
361 	if (npackfd == -1) {
362 		err = got_error_from_errno("dup");
363 		goto done;
364 	}
365 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
366 	    npackfd);
367 	if (err != NULL)
368 		goto done;
369 	npackfd = -1;
370 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
371 	if (err != NULL)
372 		goto done;
373 	nidxfd = -1;
374 	for (i = 0; i < nitems(tmpfds); i++) {
375 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
376 		if (err != NULL)
377 			goto done;
378 		tmpfds[i] = -1;
379 	}
380 	done = 0;
381 	while (!done) {
382 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
383 
384 		if (cancel_cb) {
385 			err = cancel_cb(cancel_arg);
386 			if (err)
387 				goto done;
388 		}
389 
390 		err = got_privsep_recv_index_progress(&done, &nobj_total,
391 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
392 		    &idxibuf);
393 		if (err != NULL)
394 			goto done;
395 		if (nobj_indexed != 0) {
396 			err = progress_cb(progress_arg, sb.st_size,
397 			    nobj_total, nobj_indexed, nobj_loose,
398 			    nobj_resolved);
399 			if (err)
400 				break;
401 		}
402 	}
403 	if (close(imsg_idxfds[0]) == -1) {
404 		err = got_error_from_errno("close");
405 		goto done;
406 	}
407 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
408 		err = got_error_from_errno("waitpid");
409 		goto done;
410 	}
411 
412 	if (rename(tmpidxpath, idxpath) == -1) {
413 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
414 		goto done;
415 	}
416 	free(tmpidxpath);
417 	tmpidxpath = NULL;
418 
419 done:
420 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
421 		err = got_error_from_errno2("unlink", tmpidxpath);
422 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
423 		err = got_error_from_errno("close");
424 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
425 		err = got_error_from_errno("close");
426 	for (i = 0; i < nitems(tmpfds); i++) {
427 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
428 			err = got_error_from_errno("close");
429 	}
430 	free(tmpidxpath);
431 	free(idxpath);
432 	free(packfile_path);
433 	return err;
434 }
435 
436 const struct got_error *
437 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
438     struct got_repository *repo, const char *packfile_path)
439 {
440 	const struct got_error *err = NULL;
441 	const char *packdir_path = NULL;
442 	char *packfile_name = NULL, *p, *dot;
443 	struct got_object_id id;
444 	int packfd = -1;
445 
446 	*packfile = NULL;
447 	*pack_hash = NULL;
448 
449 	packdir_path = got_repo_get_path_objects_pack(repo);
450 	if (packdir_path == NULL)
451 		return got_error_from_errno("got_repo_get_path_objects_pack");
452 
453 	if (!got_path_is_child(packfile_path, packdir_path,
454 	    strlen(packdir_path))) {
455 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
456 		goto done;
457 
458 	}
459 
460 	err = got_path_basename(&packfile_name, packfile_path);
461 	if (err)
462 		goto done;
463 	p = packfile_name;
464 
465 	if (strncmp(p, "pack-", 5) != 0) {
466 		err = got_error_fmt(GOT_ERR_BAD_PATH,
467 		    "'%s' is not a valid pack file name",
468 		    packfile_name);
469 		goto done;
470 	}
471 	p += 5;
472 	dot = strchr(p, '.');
473 	if (dot == NULL) {
474 		err = got_error_fmt(GOT_ERR_BAD_PATH,
475 		    "'%s' is not a valid pack file name",
476 		    packfile_name);
477 		goto done;
478 	}
479 	if (strcmp(dot + 1, "pack") != 0) {
480 		err = got_error_fmt(GOT_ERR_BAD_PATH,
481 		    "'%s' is not a valid pack file name",
482 		    packfile_name);
483 		goto done;
484 	}
485 	*dot = '\0';
486 	if (!got_parse_object_id(&id, p, repo->algo)) {
487 		err = got_error_fmt(GOT_ERR_BAD_PATH,
488 		    "'%s' is not a valid pack file name",
489 		    packfile_name);
490 		goto done;
491 	}
492 
493 	*pack_hash = got_object_id_dup(&id);
494 	if (*pack_hash == NULL) {
495 		err = got_error_from_errno("got_object_id_dup");
496 		goto done;
497 	}
498 
499 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
500 	if (packfd == -1) {
501 		err = got_error_from_errno2("open", packfile_path);
502 		goto done;
503 	}
504 
505 	*packfile = fdopen(packfd, "r");
506 	if (*packfile == NULL) {
507 		err = got_error_from_errno2("fdopen", packfile_path);
508 		goto done;
509 	}
510 	packfd = -1;
511 done:
512 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
513 		err = got_error_from_errno2("close", packfile_path);
514 	free(packfile_name);
515 	if (err) {
516 		free(*pack_hash);
517 		*pack_hash = NULL;
518 	}
519 	return err;
520 }
521 
522 const struct got_error *
523 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
524     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
525     got_cancel_cb cancel_cb, void *cancel_arg)
526 {
527 	const struct got_error *err = NULL;
528 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
529 	struct got_packidx *packidx = NULL;
530 	struct got_pack *pack = NULL;
531 	uint32_t nobj, i;
532 
533 	err = got_object_id_str(&id_str, pack_hash);
534 	if (err)
535 		goto done;
536 
537 	if (asprintf(&packpath, "%s/pack-%s.pack",
538 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
539 		err = got_error_from_errno("asprintf");
540 		goto done;
541 	}
542 	if (asprintf(&idxpath, "%s/pack-%s.idx",
543 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
544 		err = got_error_from_errno("asprintf");
545 		goto done;
546 	}
547 
548 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
549 	if (err)
550 		goto done;
551 
552 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
553 	if (err)
554 		goto done;
555 
556 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
557 	for (i = 0; i < nobj; i++) {
558 		struct got_packidx_object_id *oid;
559 		struct got_object_id id, base_id;
560 		off_t offset, base_offset = 0;
561 		uint8_t type;
562 		uint64_t size;
563 		size_t tslen, len;
564 
565 		if (cancel_cb) {
566 			err = cancel_cb(cancel_arg);
567 			if (err)
568 				break;
569 		}
570 		oid = &packidx->hdr.sorted_ids[i];
571 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
572 
573 		offset = got_packidx_get_object_offset(packidx, i);
574 		if (offset == -1) {
575 			err = got_error(GOT_ERR_BAD_PACKIDX);
576 			goto done;
577 		}
578 
579 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
580 		    pack, offset);
581 		if (err)
582 			goto done;
583 
584 		switch (type) {
585 		case GOT_OBJ_TYPE_OFFSET_DELTA:
586 			err = got_pack_parse_offset_delta(&base_offset, &len,
587 			    pack, offset, tslen);
588 			if (err)
589 				goto done;
590 			break;
591 		case GOT_OBJ_TYPE_REF_DELTA:
592 			err = got_pack_parse_ref_delta(&base_id,
593 			    pack, offset, tslen);
594 			if (err)
595 				goto done;
596 			break;
597 		}
598 		err = (*list_cb)(list_arg, &id, type, offset, size,
599 		    base_offset, &base_id);
600 		if (err)
601 			goto done;
602 	}
603 
604 done:
605 	free(id_str);
606 	free(idxpath);
607 	free(packpath);
608 	if (packidx)
609 		got_packidx_close(packidx);
610 	return err;
611 }
612 
613 static const struct got_error *
614 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
615     void *progress_arg, struct got_ratelimit *rl,
616     int nloose, int ncommits, int npurged)
617 {
618 	const struct got_error *err;
619 	int elapsed;
620 
621 	if (progress_cb == NULL)
622 		return NULL;
623 
624 	err = got_ratelimit_check(&elapsed, rl);
625 	if (err || !elapsed)
626 		return err;
627 
628 	return progress_cb(progress_arg, nloose, ncommits, npurged);
629 }
630 
631 static const struct got_error *
632 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
633     got_cleanup_progress_cb progress_cb, void *progress_arg,
634     struct got_ratelimit *rl, struct got_repository *repo)
635 {
636 	const struct got_error *err = NULL;
637 	char *path_objects = NULL, *path = NULL;
638 	DIR *dir = NULL;
639 	struct got_object *obj = NULL;
640 	struct got_object_id id;
641 	int i, fd = -1;
642 	struct stat sb;
643 
644 	*ondisk_size = 0;
645 	*loose_ids = got_object_idset_alloc();
646 	if (*loose_ids == NULL)
647 		return got_error_from_errno("got_object_idset_alloc");
648 
649 	path_objects = got_repo_get_path_objects(repo);
650 	if (path_objects == NULL) {
651 		err = got_error_from_errno("got_repo_get_path_objects");
652 		goto done;
653 	}
654 
655 	for (i = 0; i <= 0xff; i++) {
656 		struct dirent *dent;
657 
658 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
659 			err = got_error_from_errno("asprintf");
660 			break;
661 		}
662 
663 		dir = opendir(path);
664 		if (dir == NULL) {
665 			if (errno == ENOENT) {
666 				err = NULL;
667 				continue;
668 			}
669 			err = got_error_from_errno2("opendir", path);
670 			break;
671 		}
672 
673 		while ((dent = readdir(dir)) != NULL) {
674 			char *id_str;
675 
676 			if (strcmp(dent->d_name, ".") == 0 ||
677 			    strcmp(dent->d_name, "..") == 0)
678 				continue;
679 
680 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
681 				err = got_error_from_errno("asprintf");
682 				goto done;
683 			}
684 
685 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
686 				free(id_str);
687 				continue;
688 			}
689 			free(id_str);
690 
691 			err = got_object_open_loose_fd(&fd, &id, repo);
692 			if (err)
693 				goto done;
694 			if (fstat(fd, &sb) == -1) {
695 				err = got_error_from_errno("fstat");
696 				goto done;
697 			}
698 			err = got_object_read_header_privsep(&obj, &id, repo,
699 			    fd);
700 			if (err)
701 				goto done;
702 			fd = -1; /* already closed */
703 
704 			switch (obj->type) {
705 			case GOT_OBJ_TYPE_COMMIT:
706 			case GOT_OBJ_TYPE_TREE:
707 			case GOT_OBJ_TYPE_BLOB:
708 			case GOT_OBJ_TYPE_TAG:
709 				break;
710 			default:
711 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
712 				    "%d", obj->type);
713 				goto done;
714 			}
715 			got_object_close(obj);
716 			obj = NULL;
717 			(*ondisk_size) += sb.st_size;
718 			err = got_object_idset_add(*loose_ids, &id, NULL);
719 			if (err)
720 				goto done;
721 			err = report_cleanup_progress(progress_cb,
722 			    progress_arg, rl,
723 			    got_object_idset_num_elements(*loose_ids), -1, -1);
724 			if (err)
725 				goto done;
726 		}
727 
728 		if (closedir(dir) != 0) {
729 			err = got_error_from_errno("closedir");
730 			goto done;
731 		}
732 		dir = NULL;
733 
734 		free(path);
735 		path = NULL;
736 	}
737 done:
738 	if (dir && closedir(dir) != 0 && err == NULL)
739 		err = got_error_from_errno("closedir");
740 	if (fd != -1 && close(fd) == -1 && err == NULL)
741 		err = got_error_from_errno("close");
742 	if (err) {
743 		got_object_idset_free(*loose_ids);
744 		*loose_ids = NULL;
745 	}
746 	if (obj)
747 		got_object_close(obj);
748 	free(path_objects);
749 	free(path);
750 	return err;
751 }
752 
753 static const struct got_error *
754 preserve_loose_object(struct got_object_idset *loose_ids,
755     struct got_object_id *id, struct got_repository *repo, int *npacked)
756 {
757 	const struct got_error *err = NULL;
758 	struct got_object *obj;
759 
760 	if (!got_object_idset_contains(loose_ids, id))
761 		return NULL;
762 
763 	/*
764 	 * Try to open this object from a pack file. This ensures that
765 	 * we do in fact have a valid packed copy of the object. Otherwise
766 	 * we should not delete the loose representation of this object.
767 	 */
768 	err = got_object_open_packed(&obj, id, repo);
769 	if (err == NULL) {
770 		got_object_close(obj);
771 		/*
772 		 * The object is referenced and packed.
773 		 * We can purge the redundantly stored loose object.
774 		 */
775 		(*npacked)++;
776 		return NULL;
777 	} else if (err->code != GOT_ERR_NO_OBJ)
778 		return err;
779 
780 	/*
781 	 * This object is referenced and not packed.
782 	 * Remove it from our purge set.
783 	 */
784 	return got_object_idset_remove(NULL, loose_ids, id);
785 }
786 
787 static const struct got_error *
788 load_tree_entries(struct got_object_id_queue *ids,
789     struct got_object_idset *loose_ids,
790     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
791     const char *dpath, struct got_repository *repo, int *npacked,
792     got_cancel_cb cancel_cb, void *cancel_arg)
793 {
794 	const struct got_error *err;
795 	struct got_tree_object *tree;
796 	char *p = NULL;
797 	int i;
798 
799 	err = got_object_open_as_tree(&tree, repo, tree_id);
800 	if (err)
801 		return err;
802 
803 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
804 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
805 		struct got_object_id *id = got_tree_entry_get_id(e);
806 		mode_t mode = got_tree_entry_get_mode(e);
807 
808 		if (cancel_cb) {
809 			err = (*cancel_cb)(cancel_arg);
810 			if (err)
811 				break;
812 		}
813 
814 		if (got_object_tree_entry_is_symlink(e) ||
815 		    got_object_tree_entry_is_submodule(e) ||
816 		    got_object_idset_contains(traversed_ids, id))
817 			continue;
818 
819 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
820 		    got_tree_entry_get_name(e)) == -1) {
821 			err = got_error_from_errno("asprintf");
822 			break;
823 		}
824 
825 		if (S_ISDIR(mode)) {
826 			struct got_object_qid *qid;
827 			err = got_object_qid_alloc(&qid, id);
828 			if (err)
829 				break;
830 			STAILQ_INSERT_TAIL(ids, qid, entry);
831 		} else if (S_ISREG(mode)) {
832 			/* This blob is referenced. */
833 			err = preserve_loose_object(loose_ids, id, repo,
834 			    npacked);
835 			if (err)
836 				break;
837 			err = got_object_idset_add(traversed_ids, id, NULL);
838 			if (err)
839 				break;
840 		}
841 		free(p);
842 		p = NULL;
843 	}
844 
845 	got_object_tree_close(tree);
846 	free(p);
847 	return err;
848 }
849 
850 static const struct got_error *
851 load_tree(struct got_object_idset *loose_ids,
852     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
853     const char *dpath, struct got_repository *repo, int *npacked,
854     got_cancel_cb cancel_cb, void *cancel_arg)
855 {
856 	const struct got_error *err = NULL;
857 	struct got_object_id_queue tree_ids;
858 	struct got_object_qid *qid;
859 
860 	err = got_object_qid_alloc(&qid, tree_id);
861 	if (err)
862 		return err;
863 
864 	STAILQ_INIT(&tree_ids);
865 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
866 
867 	while (!STAILQ_EMPTY(&tree_ids)) {
868 		if (cancel_cb) {
869 			err = (*cancel_cb)(cancel_arg);
870 			if (err)
871 				break;
872 		}
873 
874 		qid = STAILQ_FIRST(&tree_ids);
875 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
876 
877 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
878 			got_object_qid_free(qid);
879 			continue;
880 		}
881 
882 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
883 		if (err) {
884 			got_object_qid_free(qid);
885 			break;
886 		}
887 
888 		/* This tree is referenced. */
889 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
890 		if (err)
891 			break;
892 
893 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
894 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
895 		got_object_qid_free(qid);
896 		if (err)
897 			break;
898 	}
899 
900 	got_object_id_queue_free(&tree_ids);
901 	return err;
902 }
903 
904 static const struct got_error *
905 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
906     int *npacked, struct got_object_idset *traversed_ids,
907     struct got_object_id *id, struct got_repository *repo,
908     got_cleanup_progress_cb progress_cb, void *progress_arg,
909     struct got_ratelimit *rl, int nloose,
910     got_cancel_cb cancel_cb, void *cancel_arg)
911 {
912 	const struct got_error *err;
913 	struct got_commit_object *commit = NULL;
914 	struct got_tag_object *tag = NULL;
915 	struct got_object_id *tree_id = NULL;
916 	struct got_object_id_queue ids;
917 	struct got_object_qid *qid;
918 	int obj_type;
919 
920 	err = got_object_qid_alloc(&qid, id);
921 	if (err)
922 		return err;
923 
924 	STAILQ_INIT(&ids);
925 	STAILQ_INSERT_TAIL(&ids, qid, entry);
926 
927 	while (!STAILQ_EMPTY(&ids)) {
928 		if (cancel_cb) {
929 			err = (*cancel_cb)(cancel_arg);
930 			if (err)
931 				break;
932 		}
933 
934 		qid = STAILQ_FIRST(&ids);
935 		STAILQ_REMOVE_HEAD(&ids, entry);
936 
937 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
938 			got_object_qid_free(qid);
939 			qid = NULL;
940 			continue;
941 		}
942 
943 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
944 		if (err)
945 			break;
946 
947 		/* This commit or tag is referenced. */
948 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
949 		if (err)
950 			break;
951 
952 		err = got_object_get_type(&obj_type, repo, &qid->id);
953 		if (err)
954 			break;
955 		switch (obj_type) {
956 		case GOT_OBJ_TYPE_COMMIT:
957 			err = got_object_open_as_commit(&commit, repo,
958 			    &qid->id);
959 			if (err)
960 				goto done;
961 			break;
962 		case GOT_OBJ_TYPE_TAG:
963 			err = got_object_open_as_tag(&tag, repo, &qid->id);
964 			if (err)
965 				goto done;
966 			break;
967 		default:
968 			/* should not happen */
969 			err = got_error(GOT_ERR_OBJ_TYPE);
970 			goto done;
971 		}
972 
973 		/* Find a tree object to scan. */
974 		if (commit) {
975 			tree_id = got_object_commit_get_tree_id(commit);
976 		} else if (tag) {
977 			obj_type = got_object_tag_get_object_type(tag);
978 			switch (obj_type) {
979 			case GOT_OBJ_TYPE_COMMIT:
980 				err = got_object_open_as_commit(&commit, repo,
981 				    got_object_tag_get_object_id(tag));
982 				if (err)
983 					goto done;
984 				tree_id = got_object_commit_get_tree_id(commit);
985 				break;
986 			case GOT_OBJ_TYPE_TREE:
987 				tree_id = got_object_tag_get_object_id(tag);
988 				break;
989 			default:
990 				/*
991 				 * Tag points at something other than a
992 				 * commit or tree. Leave this weird tag object
993 				 * and the object it points to on disk.
994 				 */
995 				err = got_object_idset_remove(NULL, loose_ids,
996 				    &qid->id);
997 				if (err && err->code != GOT_ERR_NO_OBJ)
998 					goto done;
999 				err = got_object_idset_remove(NULL, loose_ids,
1000 				    got_object_tag_get_object_id(tag));
1001 				if (err && err->code != GOT_ERR_NO_OBJ)
1002 					goto done;
1003 				err = NULL;
1004 				break;
1005 			}
1006 		}
1007 
1008 		if (tree_id) {
1009 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1010 			    repo, npacked, cancel_cb, cancel_arg);
1011 			if (err)
1012 				break;
1013 		}
1014 
1015 		if (commit || tag)
1016 			(*ncommits)++; /* scanned tags are counted as commits */
1017 
1018 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1019 		    nloose, *ncommits, -1);
1020 		if (err)
1021 			break;
1022 
1023 		if (commit) {
1024 			/* Find parent commits to scan. */
1025 			const struct got_object_id_queue *parent_ids;
1026 			parent_ids = got_object_commit_get_parent_ids(commit);
1027 			err = got_object_id_queue_copy(parent_ids, &ids);
1028 			if (err)
1029 				break;
1030 			got_object_commit_close(commit);
1031 			commit = NULL;
1032 		}
1033 		if (tag) {
1034 			got_object_tag_close(tag);
1035 			tag = NULL;
1036 		}
1037 		got_object_qid_free(qid);
1038 		qid = NULL;
1039 	}
1040 done:
1041 	if (qid)
1042 		got_object_qid_free(qid);
1043 	if (commit)
1044 		got_object_commit_close(commit);
1045 	if (tag)
1046 		got_object_tag_close(tag);
1047 	got_object_id_queue_free(&ids);
1048 	return err;
1049 }
1050 
1051 struct purge_loose_object_arg {
1052 	struct got_repository *repo;
1053 	got_cleanup_progress_cb progress_cb;
1054 	void *progress_arg;
1055 	struct got_ratelimit *rl;
1056 	int nloose;
1057 	int ncommits;
1058 	int npurged;
1059 	off_t size_purged;
1060 	int dry_run;
1061 	time_t max_mtime;
1062 	int ignore_mtime;
1063 };
1064 
1065 static const struct got_error *
1066 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1067 {
1068 	struct purge_loose_object_arg *a = arg;
1069 	const struct got_error *err, *unlock_err = NULL;
1070 	char *path = NULL;
1071 	int fd = -1;
1072 	struct stat sb;
1073 	struct got_lockfile *lf = NULL;
1074 
1075 	err = got_object_get_path(&path, id, a->repo);
1076 	if (err)
1077 		return err;
1078 
1079 	err = got_object_open_loose_fd(&fd, id, a->repo);
1080 	if (err)
1081 		goto done;
1082 
1083 	if (fstat(fd, &sb) == -1) {
1084 		err = got_error_from_errno("fstat");
1085 		goto done;
1086 	}
1087 
1088 	/*
1089 	 * Do not delete objects which are younger than our maximum
1090 	 * modification time threshold. This prevents a race where
1091 	 * new objects which are being added to the repository
1092 	 * concurrently would be deleted.
1093 	 */
1094 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1095 		if (!a->dry_run) {
1096 			err = got_lockfile_lock(&lf, path, -1);
1097 			if (err)
1098 				goto done;
1099 			if (unlink(path) == -1) {
1100 				err = got_error_from_errno2("unlink", path);
1101 				goto done;
1102 			}
1103 		}
1104 
1105 		a->npurged++;
1106 		a->size_purged += sb.st_size;
1107 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1108 		    a->rl, a->nloose, a->ncommits, a->npurged);
1109 		if (err)
1110 			goto done;
1111 	}
1112 done:
1113 	if (fd != -1 && close(fd) == -1 && err == NULL)
1114 		err = got_error_from_errno("close");
1115 	free(path);
1116 	if (lf)
1117 		unlock_err = got_lockfile_unlock(lf, -1);
1118 	return err ? err : unlock_err;
1119 }
1120 
1121 const struct got_error *
1122 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1123     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1124     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1125     got_cancel_cb cancel_cb, void *cancel_arg)
1126 {
1127 	const struct got_error *err;
1128 	struct got_object_idset *loose_ids;
1129 	struct got_object_idset *traversed_ids;
1130 	struct got_object_id **referenced_ids;
1131 	int i, nreferenced, nloose, ncommits = 0;
1132 	struct got_reflist_head refs;
1133 	struct got_reflist_entry *re;
1134 	struct purge_loose_object_arg arg;
1135 	time_t max_mtime = 0;
1136 	struct got_ratelimit rl;
1137 
1138 	TAILQ_INIT(&refs);
1139 	got_ratelimit_init(&rl, 0, 500);
1140 
1141 	*size_before = 0;
1142 	*size_after = 0;
1143 	*npacked = 0;
1144 
1145 	err = get_loose_object_ids(&loose_ids, size_before,
1146 	    progress_cb, progress_arg, &rl, repo);
1147 	if (err)
1148 		return err;
1149 	nloose = got_object_idset_num_elements(loose_ids);
1150 	if (nloose == 0) {
1151 		got_object_idset_free(loose_ids);
1152 		if (progress_cb) {
1153 			err = progress_cb(progress_arg, 0, 0, 0);
1154 			if (err)
1155 				return err;
1156 		}
1157 		return NULL;
1158 	}
1159 
1160 	traversed_ids = got_object_idset_alloc();
1161 	if (traversed_ids == NULL) {
1162 		err = got_error_from_errno("got_object_idset_alloc");
1163 		goto done;
1164 	}
1165 
1166 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1167 	if (err)
1168 		goto done;
1169 	if (!ignore_mtime) {
1170 		TAILQ_FOREACH(re, &refs, entry) {
1171 			time_t mtime = got_ref_get_mtime(re->ref);
1172 			if (mtime > max_mtime)
1173 				max_mtime = mtime;
1174 		}
1175 		/*
1176 		 * For safety, keep objects created within 10 minutes
1177 		 * before the youngest reference was created.
1178 		 */
1179 		if (max_mtime >= 600)
1180 			max_mtime -= 600;
1181 	}
1182 
1183 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1184 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1185 	    &refs, repo, cancel_cb, cancel_arg);
1186 	if (err)
1187 		goto done;
1188 
1189 	for (i = 0; i < nreferenced; i++) {
1190 		struct got_object_id *id = referenced_ids[i];
1191 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1192 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1193 		    nloose, cancel_cb, cancel_arg);
1194 		if (err)
1195 			goto done;
1196 	}
1197 
1198 	/* Any remaining loose objects are unreferenced and can be purged. */
1199 	arg.repo = repo;
1200 	arg.progress_arg = progress_arg;
1201 	arg.progress_cb = progress_cb;
1202 	arg.rl = &rl;
1203 	arg.nloose = nloose;
1204 	arg.npurged = 0;
1205 	arg.size_purged = 0;
1206 	arg.ncommits = ncommits;
1207 	arg.dry_run = dry_run;
1208 	arg.max_mtime = max_mtime;
1209 	arg.ignore_mtime = ignore_mtime;
1210 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1211 	if (err)
1212 		goto done;
1213 	*size_after = *size_before - arg.size_purged;
1214 
1215 	/* Produce a final progress report. */
1216 	if (progress_cb) {
1217 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1218 		if (err)
1219 			goto done;
1220 	}
1221 done:
1222 	got_object_idset_free(loose_ids);
1223 	got_object_idset_free(traversed_ids);
1224 	return err;
1225 }
1226 
1227 static const struct got_error *
1228 remove_packidx(int dir_fd, const char *relpath)
1229 {
1230 	const struct got_error *err, *unlock_err;
1231 	struct got_lockfile *lf;
1232 
1233 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1234 	if (err)
1235 		return err;
1236 	if (unlinkat(dir_fd, relpath, 0) == -1)
1237 		err = got_error_from_errno("unlinkat");
1238 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1239 	return err ? err : unlock_err;
1240 }
1241 
1242 const struct got_error *
1243 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1244     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1245     got_cancel_cb cancel_cb, void *cancel_arg)
1246 {
1247 	const struct got_error *err = NULL;
1248 	DIR *packdir = NULL;
1249 	struct dirent *dent;
1250 	char *pack_relpath = NULL;
1251 	int packdir_fd;
1252 	struct stat sb;
1253 
1254 	packdir_fd = openat(got_repo_get_fd(repo),
1255 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1256 	if (packdir_fd == -1) {
1257 		if (errno == ENOENT)
1258 			return NULL;
1259 		return got_error_from_errno_fmt("openat: %s/%s",
1260 		    got_repo_get_path_git_dir(repo),
1261 		    GOT_OBJECTS_PACK_DIR);
1262 	}
1263 
1264 	packdir = fdopendir(packdir_fd);
1265 	if (packdir == NULL) {
1266 		err = got_error_from_errno("fdopendir");
1267 		goto done;
1268 	}
1269 
1270 	while ((dent = readdir(packdir)) != NULL) {
1271 		if (cancel_cb) {
1272 			err = cancel_cb(cancel_arg);
1273 			if (err)
1274 				goto done;
1275 		}
1276 
1277 		if (!got_repo_is_packidx_filename(dent->d_name,
1278 		    strlen(dent->d_name)))
1279 			continue;
1280 
1281 		err = got_packidx_get_packfile_path(&pack_relpath,
1282 		    dent->d_name);
1283 		if (err)
1284 			goto done;
1285 
1286 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1287 			free(pack_relpath);
1288 			pack_relpath = NULL;
1289 			continue;
1290 		}
1291 		if (errno != ENOENT) {
1292 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1293 			    got_repo_get_path_git_dir(repo),
1294 			    GOT_OBJECTS_PACK_DIR,
1295 			    pack_relpath);
1296 			goto done;
1297 		}
1298 
1299 		if (!dry_run) {
1300 			err = remove_packidx(packdir_fd, dent->d_name);
1301 			if (err)
1302 				goto done;
1303 		}
1304 		if (progress_cb) {
1305 			char *path;
1306 			if (asprintf(&path, "%s/%s/%s",
1307 			    got_repo_get_path_git_dir(repo),
1308 			    GOT_OBJECTS_PACK_DIR,
1309 			    dent->d_name) == -1) {
1310 				err = got_error_from_errno("asprintf");
1311 				goto done;
1312 			}
1313 			err = progress_cb(progress_arg, path);
1314 			free(path);
1315 			if (err)
1316 				goto done;
1317 		}
1318 		free(pack_relpath);
1319 		pack_relpath = NULL;
1320 	}
1321 done:
1322 	if (packdir && closedir(packdir) != 0 && err == NULL)
1323 		err = got_error_from_errno("closedir");
1324 	free(pack_relpath);
1325 	return err;
1326 }