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/uio.h>
20 #include <sys/stat.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 
24 #include <dirent.h>
25 #include <errno.h>
26 #include <fcntl.h>
27 #include <stdint.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <limits.h>
32 #include <unistd.h>
33 
34 #include "got_error.h"
35 #include "got_cancel.h"
36 #include "got_object.h"
37 #include "got_reference.h"
38 #include "got_repository.h"
39 #include "got_repository_admin.h"
40 #include "got_opentemp.h"
41 #include "got_path.h"
42 
43 #include "got_lib_delta.h"
44 #include "got_lib_object.h"
45 #include "got_lib_object_idset.h"
46 #include "got_lib_object_cache.h"
47 #include "got_lib_pack.h"
48 #include "got_lib_privsep.h"
49 #include "got_lib_repository.h"
50 #include "got_lib_ratelimit.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 	err = got_reflist_sort(refs,
75 	    got_ref_cmp_by_commit_timestamp_descending, repo);
76 	if (err)
77 		return err;
78 
79 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
80 	if (*ids == NULL)
81 		return got_error_from_errno("reallocarray");
82 	nalloc = alloc_chunksz;
83 
84 	TAILQ_FOREACH(re, refs, entry) {
85 		struct got_object_id *id;
86 
87 		if (cancel_cb) {
88 			err = cancel_cb(cancel_arg);
89 			if (err)
90 				goto done;
91 		}
92 
93 		err = got_ref_resolve(&id, repo, re->ref);
94 		if (err)
95 			goto done;
96 
97 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
98 			int obj_type;
99 			err = got_object_get_type(&obj_type, repo, id);
100 			if (err)
101 				goto done;
102 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
103 				free(id);
104 				id = NULL;
105 				continue;
106 			}
107 		}
108 
109 		if (nalloc <= *nobjects) {
110 			struct got_object_id **new;
111 			new = recallocarray(*ids, nalloc,
112 			    nalloc + alloc_chunksz,
113 			    sizeof(struct got_object_id *));
114 			if (new == NULL) {
115 				err = got_error_from_errno(
116 				    "recallocarray");
117 				goto done;
118 			}
119 			*ids = new;
120 			nalloc += alloc_chunksz;
121 		}
122 		(*ids)[*nobjects] = id;
123 		if ((*ids)[*nobjects] == NULL) {
124 			err = got_error_from_errno("got_object_id_dup");
125 			goto done;
126 		}
127 		(*nobjects)++;
128 	}
129 done:
130 	if (err) {
131 		for (i = 0; i < *nobjects; i++)
132 			free((*ids)[i]);
133 		free(*ids);
134 		*ids = NULL;
135 		*nobjects = 0;
136 	}
137 	return err;
138 }
139 
140 const struct got_error *
141 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
142     struct got_reflist_head *include_refs,
143     struct got_reflist_head *exclude_refs, struct got_repository *repo,
144     int loose_obj_only,
145     got_pack_progress_cb progress_cb, void *progress_arg,
146     got_cancel_cb cancel_cb, void *cancel_arg)
147 {
148 	const struct got_error *err = NULL;
149 	struct got_object_id **ours = NULL, **theirs = NULL;
150 	int nours = 0, ntheirs = 0, packfd = -1, i;
151 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
152 	char *sha1_str = NULL;
153 	FILE *delta_cache = NULL;
154 	struct got_ratelimit rl;
155 
156 	*packfile = NULL;
157 	*pack_hash = NULL;
158 
159 	got_ratelimit_init(&rl, 0, 500);
160 
161 	if (asprintf(&path, "%s/%s/packing.pack",
162 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
163 		err = got_error_from_errno("asprintf");
164 		goto done;
165 	}
166 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path, "");
167 	if (err)
168 		goto done;
169 
170 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
171 		err = got_error_from_errno2("fchmod", tmpfile_path);
172 		goto done;
173 	}
174 
175 	delta_cache = got_opentemp();
176 	if (delta_cache == NULL) {
177 		err = got_error_from_errno("got_opentemp");
178 		goto done;
179 	}
180 
181 	err = get_reflist_object_ids(&ours, &nours,
182 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
183 	    include_refs, repo, cancel_cb, cancel_arg);
184 	if (err)
185 		goto done;
186 
187 	if (nours == 0) {
188 		err = got_error(GOT_ERR_CANNOT_PACK);
189 		goto done;
190 	}
191 
192 	if (!TAILQ_EMPTY(exclude_refs)) {
193 		err = get_reflist_object_ids(&theirs, &ntheirs,
194 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
195 		    exclude_refs, repo,
196 		    cancel_cb, cancel_arg);
197 		if (err)
198 			goto done;
199 	}
200 
201 	*pack_hash = calloc(1, sizeof(**pack_hash));
202 	if (*pack_hash == NULL) {
203 		err = got_error_from_errno("calloc");
204 		goto done;
205 	}
206 
207 	err = got_pack_create((*pack_hash)->sha1, packfd, delta_cache,
208 	    theirs, ntheirs, ours, nours, repo, loose_obj_only, 0,
209 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
210 	if (err)
211 		goto done;
212 
213 	err = got_object_id_str(&sha1_str, *pack_hash);
214 	if (err)
215 		goto done;
216 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
217 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
218 	    sha1_str) == -1) {
219 		err = got_error_from_errno("asprintf");
220 		goto done;
221 	}
222 
223 	if (lseek(packfd, 0L, SEEK_SET) == -1) {
224 		err = got_error_from_errno("lseek");
225 		goto done;
226 	}
227 	if (rename(tmpfile_path, packfile_path) == -1) {
228 		err = got_error_from_errno3("rename", tmpfile_path,
229 		    packfile_path);
230 		goto done;
231 	}
232 	free(tmpfile_path);
233 	tmpfile_path = NULL;
234 
235 	*packfile = fdopen(packfd, "w");
236 	if (*packfile == NULL) {
237 		err = got_error_from_errno2("fdopen", tmpfile_path);
238 		goto done;
239 	}
240 	packfd = -1;
241 done:
242 	for (i = 0; i < nours; i++)
243 		free(ours[i]);
244 	free(ours);
245 	for (i = 0; i < ntheirs; i++)
246 		free(theirs[i]);
247 	free(theirs);
248 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
249 		err = got_error_from_errno2("close", packfile_path);
250 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
251 		err = got_error_from_errno("fclose");
252 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
253 		err = got_error_from_errno2("unlink", tmpfile_path);
254 	free(tmpfile_path);
255 	free(packfile_path);
256 	free(sha1_str);
257 	free(path);
258 	if (err) {
259 		free(*pack_hash);
260 		*pack_hash = NULL;
261 		if (*packfile)
262 			fclose(*packfile);
263 		*packfile = NULL;
264 	}
265 	return err;
266 }
267 
268 const struct got_error *
269 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
270     struct got_repository *repo,
271     got_pack_index_progress_cb progress_cb, void *progress_arg,
272     got_cancel_cb cancel_cb, void *cancel_arg)
273 {
274 	size_t i;
275 	char *path;
276 	int imsg_idxfds[2];
277 	int npackfd = -1, idxfd = -1, nidxfd = -1;
278 	int tmpfds[3];
279 	int idxstatus, done = 0;
280 	const struct got_error *err;
281 	struct imsgbuf idxibuf;
282 	pid_t idxpid;
283 	char *tmpidxpath = NULL;
284 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
285 	const char *repo_path = got_repo_get_path_git_dir(repo);
286 	struct stat sb;
287 
288 	for (i = 0; i < nitems(tmpfds); i++)
289 		tmpfds[i] = -1;
290 
291 	if (asprintf(&path, "%s/%s/indexing.idx",
292 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
293 		err = got_error_from_errno("asprintf");
294 		goto done;
295 	}
296 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
297 	free(path);
298 	if (err)
299 		goto done;
300 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
301 		err = got_error_from_errno2("fchmod", tmpidxpath);
302 		goto done;
303 	}
304 
305 	nidxfd = dup(idxfd);
306 	if (nidxfd == -1) {
307 		err = got_error_from_errno("dup");
308 		goto done;
309 	}
310 
311 	for (i = 0; i < nitems(tmpfds); i++) {
312 		tmpfds[i] = got_opentempfd();
313 		if (tmpfds[i] == -1) {
314 			err = got_error_from_errno("got_opentempfd");
315 			goto done;
316 		}
317 	}
318 
319 	err = got_object_id_str(&id_str, pack_hash);
320 	if (err)
321 		goto done;
322 
323 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
324 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
325 		err = got_error_from_errno("asprintf");
326 		goto done;
327 	}
328 
329 	if (fstat(fileno(packfile), &sb) == -1) {
330 		err = got_error_from_errno2("fstat", packfile_path);
331 		goto done;
332 	}
333 
334 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
335 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
336 		err = got_error_from_errno("asprintf");
337 		goto done;
338 	}
339 
340 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
341 		err = got_error_from_errno("socketpair");
342 		goto done;
343 	}
344 	idxpid = fork();
345 	if (idxpid == -1) {
346 		err= got_error_from_errno("fork");
347 		goto done;
348 	} else if (idxpid == 0)
349 		got_privsep_exec_child(imsg_idxfds,
350 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
351 	if (close(imsg_idxfds[1]) == -1) {
352 		err = got_error_from_errno("close");
353 		goto done;
354 	}
355 	imsg_init(&idxibuf, imsg_idxfds[0]);
356 
357 	npackfd = dup(fileno(packfile));
358 	if (npackfd == -1) {
359 		err = got_error_from_errno("dup");
360 		goto done;
361 	}
362 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
363 	    npackfd);
364 	if (err != NULL)
365 		goto done;
366 	npackfd = -1;
367 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
368 	if (err != NULL)
369 		goto done;
370 	nidxfd = -1;
371 	for (i = 0; i < nitems(tmpfds); i++) {
372 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
373 		if (err != NULL)
374 			goto done;
375 		tmpfds[i] = -1;
376 	}
377 	done = 0;
378 	while (!done) {
379 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
380 
381 		if (cancel_cb) {
382 			err = cancel_cb(cancel_arg);
383 			if (err)
384 				goto done;
385 		}
386 
387 		err = got_privsep_recv_index_progress(&done, &nobj_total,
388 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
389 		    &idxibuf);
390 		if (err != NULL)
391 			goto done;
392 		if (nobj_indexed != 0) {
393 			err = progress_cb(progress_arg, sb.st_size,
394 			    nobj_total, nobj_indexed, nobj_loose,
395 			    nobj_resolved);
396 			if (err)
397 				break;
398 		}
399 	}
400 	if (close(imsg_idxfds[0]) == -1) {
401 		err = got_error_from_errno("close");
402 		goto done;
403 	}
404 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
405 		err = got_error_from_errno("waitpid");
406 		goto done;
407 	}
408 
409 	if (rename(tmpidxpath, idxpath) == -1) {
410 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
411 		goto done;
412 	}
413 	free(tmpidxpath);
414 	tmpidxpath = NULL;
415 
416 done:
417 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
418 		err = got_error_from_errno2("unlink", tmpidxpath);
419 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
420 		err = got_error_from_errno("close");
421 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
422 		err = got_error_from_errno("close");
423 	for (i = 0; i < nitems(tmpfds); i++) {
424 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
425 			err = got_error_from_errno("close");
426 	}
427 	free(tmpidxpath);
428 	free(idxpath);
429 	free(packfile_path);
430 	return err;
431 }
432 
433 const struct got_error *
434 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
435     struct got_repository *repo, const char *packfile_path)
436 {
437 	const struct got_error *err = NULL;
438 	const char *packdir_path = NULL;
439 	char *packfile_name = NULL, *p, *dot;
440 	struct got_object_id id;
441 	int packfd = -1;
442 
443 	*packfile = NULL;
444 	*pack_hash = NULL;
445 
446 	packdir_path = got_repo_get_path_objects_pack(repo);
447 	if (packdir_path == NULL)
448 		return got_error_from_errno("got_repo_get_path_objects_pack");
449 
450 	if (!got_path_is_child(packfile_path, packdir_path,
451 	    strlen(packdir_path))) {
452 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
453 		goto done;
454 
455 	}
456 
457 	err = got_path_basename(&packfile_name, packfile_path);
458 	if (err)
459 		goto done;
460 	p = packfile_name;
461 
462 	if (strncmp(p, "pack-", 5) != 0) {
463 		err = got_error_fmt(GOT_ERR_BAD_PATH,
464 		    "'%s' is not a valid pack file name",
465 		    packfile_name);
466 		goto done;
467 	}
468 	p += 5;
469 	dot = strchr(p, '.');
470 	if (dot == NULL) {
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 	if (strcmp(dot + 1, "pack") != 0) {
477 		err = got_error_fmt(GOT_ERR_BAD_PATH,
478 		    "'%s' is not a valid pack file name",
479 		    packfile_name);
480 		goto done;
481 	}
482 	*dot = '\0';
483 	if (!got_parse_sha1_digest(id.sha1, p)) {
484 		err = got_error_fmt(GOT_ERR_BAD_PATH,
485 		    "'%s' is not a valid pack file name",
486 		    packfile_name);
487 		goto done;
488 	}
489 
490 	*pack_hash = got_object_id_dup(&id);
491 	if (*pack_hash == NULL) {
492 		err = got_error_from_errno("got_object_id_dup");
493 		goto done;
494 	}
495 
496 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
497 	if (packfd == -1) {
498 		err = got_error_from_errno2("open", packfile_path);
499 		goto done;
500 	}
501 
502 	*packfile = fdopen(packfd, "r");
503 	if (*packfile == NULL) {
504 		err = got_error_from_errno2("fdopen", packfile_path);
505 		goto done;
506 	}
507 	packfd = -1;
508 done:
509 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
510 		err = got_error_from_errno2("close", packfile_path);
511 	free(packfile_name);
512 	if (err) {
513 		free(*pack_hash);
514 		*pack_hash = NULL;
515 	}
516 	return err;
517 }
518 
519 const struct got_error *
520 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
521     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
522     got_cancel_cb cancel_cb, void *cancel_arg)
523 {
524 	const struct got_error *err = NULL;
525 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
526 	struct got_packidx *packidx = NULL;
527 	struct got_pack *pack = NULL;
528 	uint32_t nobj, i;
529 
530 	err = got_object_id_str(&id_str, pack_hash);
531 	if (err)
532 		goto done;
533 
534 	if (asprintf(&packpath, "%s/pack-%s.pack",
535 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
536 		err = got_error_from_errno("asprintf");
537 		goto done;
538 	}
539 	if (asprintf(&idxpath, "%s/pack-%s.idx",
540 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
541 		err = got_error_from_errno("asprintf");
542 		goto done;
543 	}
544 
545 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
546 	if (err)
547 		goto done;
548 
549 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
550 	if (err)
551 		goto done;
552 
553 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
554 	for (i = 0; i < nobj; i++) {
555 		struct got_packidx_object_id *oid;
556 		struct got_object_id id, base_id;
557 		off_t offset, base_offset = 0;
558 		uint8_t type;
559 		uint64_t size;
560 		size_t tslen, len;
561 
562 		if (cancel_cb) {
563 			err = cancel_cb(cancel_arg);
564 			if (err)
565 				break;
566 		}
567 		oid = &packidx->hdr.sorted_ids[i];
568 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
569 
570 		offset = got_packidx_get_object_offset(packidx, i);
571 		if (offset == -1) {
572 			err = got_error(GOT_ERR_BAD_PACKIDX);
573 			goto done;
574 		}
575 
576 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
577 		    pack, offset);
578 		if (err)
579 			goto done;
580 
581 		switch (type) {
582 		case GOT_OBJ_TYPE_OFFSET_DELTA:
583 			err = got_pack_parse_offset_delta(&base_offset, &len,
584 			    pack, offset, tslen);
585 			if (err)
586 				goto done;
587 			break;
588 		case GOT_OBJ_TYPE_REF_DELTA:
589 			err = got_pack_parse_ref_delta(&base_id,
590 			    pack, offset, tslen);
591 			if (err)
592 				goto done;
593 			break;
594 		}
595 		err = (*list_cb)(list_arg, &id, type, offset, size,
596 		    base_offset, &base_id);
597 		if (err)
598 			goto done;
599 	}
600 
601 done:
602 	free(id_str);
603 	free(idxpath);
604 	free(packpath);
605 	if (packidx)
606 		got_packidx_close(packidx);
607 	return err;
608 }
609 
610 static const struct got_error *
611 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
612     void *progress_arg, struct got_ratelimit *rl,
613     int nloose, int ncommits, int npurged)
614 {
615 	const struct got_error *err;
616 	int elapsed;
617 
618 	if (progress_cb == NULL)
619 		return NULL;
620 
621 	err = got_ratelimit_check(&elapsed, rl);
622 	if (err || !elapsed)
623 		return err;
624 
625 	return progress_cb(progress_arg, nloose, ncommits, npurged);
626 }
627 
628 static const struct got_error *
629 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
630     got_cleanup_progress_cb progress_cb, void *progress_arg,
631     struct got_ratelimit *rl, struct got_repository *repo)
632 {
633 	const struct got_error *err = NULL;
634 	char *path_objects = NULL, *path = NULL;
635 	DIR *dir = NULL;
636 	struct got_object *obj = NULL;
637 	struct got_object_id id;
638 	int i, fd = -1;
639 	struct stat sb;
640 
641 	*ondisk_size = 0;
642 	*loose_ids = got_object_idset_alloc();
643 	if (*loose_ids == NULL)
644 		return got_error_from_errno("got_object_idset_alloc");
645 
646 	path_objects = got_repo_get_path_objects(repo);
647 	if (path_objects == NULL) {
648 		err = got_error_from_errno("got_repo_get_path_objects");
649 		goto done;
650 	}
651 
652 	for (i = 0; i <= 0xff; i++) {
653 		struct dirent *dent;
654 
655 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
656 			err = got_error_from_errno("asprintf");
657 			break;
658 		}
659 
660 		dir = opendir(path);
661 		if (dir == NULL) {
662 			if (errno == ENOENT) {
663 				err = NULL;
664 				continue;
665 			}
666 			err = got_error_from_errno2("opendir", path);
667 			break;
668 		}
669 
670 		while ((dent = readdir(dir)) != NULL) {
671 			char *id_str;
672 
673 			if (strcmp(dent->d_name, ".") == 0 ||
674 			    strcmp(dent->d_name, "..") == 0)
675 				continue;
676 
677 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
678 				err = got_error_from_errno("asprintf");
679 				goto done;
680 			}
681 
682 			memset(&id, 0, sizeof(id));
683 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
684 				free(id_str);
685 				continue;
686 			}
687 			free(id_str);
688 
689 			err = got_object_open_loose_fd(&fd, &id, repo);
690 			if (err)
691 				goto done;
692 			if (fstat(fd, &sb) == -1) {
693 				err = got_error_from_errno("fstat");
694 				goto done;
695 			}
696 			err = got_object_read_header_privsep(&obj, &id, repo,
697 			    fd);
698 			if (err)
699 				goto done;
700 			fd = -1; /* already closed */
701 
702 			switch (obj->type) {
703 			case GOT_OBJ_TYPE_COMMIT:
704 			case GOT_OBJ_TYPE_TREE:
705 			case GOT_OBJ_TYPE_BLOB:
706 			case GOT_OBJ_TYPE_TAG:
707 				break;
708 			default:
709 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
710 				    "%d", obj->type);
711 				goto done;
712 			}
713 			got_object_close(obj);
714 			obj = NULL;
715 			(*ondisk_size) += sb.st_size;
716 			err = got_object_idset_add(*loose_ids, &id, NULL);
717 			if (err)
718 				goto done;
719 			err = report_cleanup_progress(progress_cb,
720 			    progress_arg, rl,
721 			    got_object_idset_num_elements(*loose_ids), -1, -1);
722 			if (err)
723 				goto done;
724 		}
725 
726 		if (closedir(dir) != 0) {
727 			err = got_error_from_errno("closedir");
728 			goto done;
729 		}
730 		dir = NULL;
731 
732 		free(path);
733 		path = NULL;
734 	}
735 done:
736 	if (dir && closedir(dir) != 0 && err == NULL)
737 		err = got_error_from_errno("closedir");
738 	if (fd != -1 && close(fd) == -1 && err == NULL)
739 		err = got_error_from_errno("close");
740 	if (err) {
741 		got_object_idset_free(*loose_ids);
742 		*loose_ids = NULL;
743 	}
744 	if (obj)
745 		got_object_close(obj);
746 	free(path_objects);
747 	free(path);
748 	return err;
749 }
750 
751 static const struct got_error *
752 preserve_loose_object(struct got_object_idset *loose_ids,
753     struct got_object_id *id, struct got_repository *repo, int *npacked)
754 {
755 	const struct got_error *err = NULL;
756 	struct got_object *obj;
757 
758 	if (!got_object_idset_contains(loose_ids, id))
759 		return NULL;
760 
761 	/*
762 	 * Try to open this object from a pack file. This ensures that
763 	 * we do in fact have a valid packed copy of the object. Otherwise
764 	 * we should not delete the loose representation of this object.
765 	 */
766 	err = got_object_open_packed(&obj, id, repo);
767 	if (err == NULL) {
768 		got_object_close(obj);
769 		/*
770 		 * The object is referenced and packed.
771 		 * We can purge the redundantly stored loose object.
772 		 */
773 		(*npacked)++;
774 		return NULL;
775 	} else if (err->code != GOT_ERR_NO_OBJ)
776 		return err;
777 
778 	/*
779 	 * This object is referenced and not packed.
780 	 * Remove it from our purge set.
781 	 */
782 	return got_object_idset_remove(NULL, loose_ids, id);
783 }
784 
785 static const struct got_error *
786 load_tree_entries(struct got_object_id_queue *ids,
787     struct got_object_idset *loose_ids,
788     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
789     const char *dpath, struct got_repository *repo, int *npacked,
790     got_cancel_cb cancel_cb, void *cancel_arg)
791 {
792 	const struct got_error *err;
793 	struct got_tree_object *tree;
794 	char *p = NULL;
795 	int i;
796 
797 	err = got_object_open_as_tree(&tree, repo, tree_id);
798 	if (err)
799 		return err;
800 
801 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
802 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
803 		struct got_object_id *id = got_tree_entry_get_id(e);
804 		mode_t mode = got_tree_entry_get_mode(e);
805 
806 		if (cancel_cb) {
807 			err = (*cancel_cb)(cancel_arg);
808 			if (err)
809 				break;
810 		}
811 
812 		if (got_object_tree_entry_is_symlink(e) ||
813 		    got_object_tree_entry_is_submodule(e) ||
814 		    got_object_idset_contains(traversed_ids, id))
815 			continue;
816 
817 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
818 		    got_tree_entry_get_name(e)) == -1) {
819 			err = got_error_from_errno("asprintf");
820 			break;
821 		}
822 
823 		if (S_ISDIR(mode)) {
824 			struct got_object_qid *qid;
825 			err = got_object_qid_alloc(&qid, id);
826 			if (err)
827 				break;
828 			STAILQ_INSERT_TAIL(ids, qid, entry);
829 		} else if (S_ISREG(mode)) {
830 			/* This blob is referenced. */
831 			err = preserve_loose_object(loose_ids, id, repo,
832 			    npacked);
833 			if (err)
834 				break;
835 			err = got_object_idset_add(traversed_ids, id, NULL);
836 			if (err)
837 				break;
838 		}
839 		free(p);
840 		p = NULL;
841 	}
842 
843 	got_object_tree_close(tree);
844 	free(p);
845 	return err;
846 }
847 
848 static const struct got_error *
849 load_tree(struct got_object_idset *loose_ids,
850     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
851     const char *dpath, struct got_repository *repo, int *npacked,
852     got_cancel_cb cancel_cb, void *cancel_arg)
853 {
854 	const struct got_error *err = NULL;
855 	struct got_object_id_queue tree_ids;
856 	struct got_object_qid *qid;
857 
858 	err = got_object_qid_alloc(&qid, tree_id);
859 	if (err)
860 		return err;
861 
862 	STAILQ_INIT(&tree_ids);
863 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
864 
865 	while (!STAILQ_EMPTY(&tree_ids)) {
866 		if (cancel_cb) {
867 			err = (*cancel_cb)(cancel_arg);
868 			if (err)
869 				break;
870 		}
871 
872 		qid = STAILQ_FIRST(&tree_ids);
873 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
874 
875 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
876 			got_object_qid_free(qid);
877 			continue;
878 		}
879 
880 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
881 		if (err) {
882 			got_object_qid_free(qid);
883 			break;
884 		}
885 
886 		/* This tree is referenced. */
887 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
888 		if (err)
889 			break;
890 
891 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
892 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
893 		got_object_qid_free(qid);
894 		if (err)
895 			break;
896 	}
897 
898 	got_object_id_queue_free(&tree_ids);
899 	return err;
900 }
901 
902 static const struct got_error *
903 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
904     int *npacked, struct got_object_idset *traversed_ids,
905     struct got_object_id *id, struct got_repository *repo,
906     got_cleanup_progress_cb progress_cb, void *progress_arg,
907     struct got_ratelimit *rl, int nloose,
908     got_cancel_cb cancel_cb, void *cancel_arg)
909 {
910 	const struct got_error *err;
911 	struct got_commit_object *commit = NULL;
912 	struct got_tag_object *tag = NULL;
913 	struct got_object_id *tree_id = NULL;
914 	struct got_object_id_queue ids;
915 	struct got_object_qid *qid;
916 	int obj_type;
917 
918 	err = got_object_qid_alloc(&qid, id);
919 	if (err)
920 		return err;
921 
922 	STAILQ_INIT(&ids);
923 	STAILQ_INSERT_TAIL(&ids, qid, entry);
924 
925 	while (!STAILQ_EMPTY(&ids)) {
926 		if (cancel_cb) {
927 			err = (*cancel_cb)(cancel_arg);
928 			if (err)
929 				break;
930 		}
931 
932 		qid = STAILQ_FIRST(&ids);
933 		STAILQ_REMOVE_HEAD(&ids, entry);
934 
935 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
936 			got_object_qid_free(qid);
937 			qid = NULL;
938 			continue;
939 		}
940 
941 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
942 		if (err)
943 			break;
944 
945 		/* This commit or tag is referenced. */
946 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
947 		if (err)
948 			break;
949 
950 		err = got_object_get_type(&obj_type, repo, &qid->id);
951 		if (err)
952 			break;
953 		switch (obj_type) {
954 		case GOT_OBJ_TYPE_COMMIT:
955 			err = got_object_open_as_commit(&commit, repo,
956 			    &qid->id);
957 			if (err)
958 				goto done;
959 			break;
960 		case GOT_OBJ_TYPE_TAG:
961 			err = got_object_open_as_tag(&tag, repo, &qid->id);
962 			if (err)
963 				goto done;
964 			break;
965 		default:
966 			/* should not happen */
967 			err = got_error(GOT_ERR_OBJ_TYPE);
968 			goto done;
969 		}
970 
971 		/* Find a tree object to scan. */
972 		if (commit) {
973 			tree_id = got_object_commit_get_tree_id(commit);
974 		} else if (tag) {
975 			obj_type = got_object_tag_get_object_type(tag);
976 			switch (obj_type) {
977 			case GOT_OBJ_TYPE_COMMIT:
978 				err = got_object_open_as_commit(&commit, repo,
979 				    got_object_tag_get_object_id(tag));
980 				if (err)
981 					goto done;
982 				tree_id = got_object_commit_get_tree_id(commit);
983 				break;
984 			case GOT_OBJ_TYPE_TREE:
985 				tree_id = got_object_tag_get_object_id(tag);
986 				break;
987 			default:
988 				/*
989 				 * Tag points at something other than a
990 				 * commit or tree. Leave this weird tag object
991 				 * and the object it points to on disk.
992 				 */
993 				err = got_object_idset_remove(NULL, loose_ids,
994 				    &qid->id);
995 				if (err && err->code != GOT_ERR_NO_OBJ)
996 					goto done;
997 				err = got_object_idset_remove(NULL, loose_ids,
998 				    got_object_tag_get_object_id(tag));
999 				if (err && err->code != GOT_ERR_NO_OBJ)
1000 					goto done;
1001 				err = NULL;
1002 				break;
1003 			}
1004 		}
1005 
1006 		if (tree_id) {
1007 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1008 			    repo, npacked, cancel_cb, cancel_arg);
1009 			if (err)
1010 				break;
1011 		}
1012 
1013 		if (commit || tag)
1014 			(*ncommits)++; /* scanned tags are counted as commits */
1015 
1016 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1017 		    nloose, *ncommits, -1);
1018 		if (err)
1019 			break;
1020 
1021 		if (commit) {
1022 			/* Find parent commits to scan. */
1023 			const struct got_object_id_queue *parent_ids;
1024 			parent_ids = got_object_commit_get_parent_ids(commit);
1025 			err = got_object_id_queue_copy(parent_ids, &ids);
1026 			if (err)
1027 				break;
1028 			got_object_commit_close(commit);
1029 			commit = NULL;
1030 		}
1031 		if (tag) {
1032 			got_object_tag_close(tag);
1033 			tag = NULL;
1034 		}
1035 		got_object_qid_free(qid);
1036 		qid = NULL;
1037 	}
1038 done:
1039 	if (qid)
1040 		got_object_qid_free(qid);
1041 	if (commit)
1042 		got_object_commit_close(commit);
1043 	if (tag)
1044 		got_object_tag_close(tag);
1045 	got_object_id_queue_free(&ids);
1046 	return err;
1047 }
1048 
1049 struct purge_loose_object_arg {
1050 	struct got_repository *repo;
1051 	got_cleanup_progress_cb progress_cb;
1052 	void *progress_arg;
1053 	struct got_ratelimit *rl;
1054 	int nloose;
1055 	int ncommits;
1056 	int npurged;
1057 	off_t size_purged;
1058 	int dry_run;
1059 	time_t max_mtime;
1060 	int ignore_mtime;
1061 };
1062 
1063 static const struct got_error *
1064 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1065 {
1066 	struct purge_loose_object_arg *a = arg;
1067 	const struct got_error *err, *unlock_err = NULL;
1068 	char *path = NULL;
1069 	int fd = -1;
1070 	struct stat sb;
1071 	struct got_lockfile *lf = NULL;
1072 
1073 	err = got_object_get_path(&path, id, a->repo);
1074 	if (err)
1075 		return err;
1076 
1077 	err = got_object_open_loose_fd(&fd, id, a->repo);
1078 	if (err)
1079 		goto done;
1080 
1081 	if (fstat(fd, &sb) == -1) {
1082 		err = got_error_from_errno("fstat");
1083 		goto done;
1084 	}
1085 
1086 	/*
1087 	 * Do not delete objects which are younger than our maximum
1088 	 * modification time threshold. This prevents a race where
1089 	 * new objects which are being added to the repository
1090 	 * concurrently would be deleted.
1091 	 */
1092 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1093 		if (!a->dry_run) {
1094 			err = got_lockfile_lock(&lf, path, -1);
1095 			if (err)
1096 				goto done;
1097 			if (unlink(path) == -1) {
1098 				err = got_error_from_errno2("unlink", path);
1099 				goto done;
1100 			}
1101 		}
1102 
1103 		a->npurged++;
1104 		a->size_purged += sb.st_size;
1105 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1106 		    a->rl, a->nloose, a->ncommits, a->npurged);
1107 		if (err)
1108 			goto done;
1109 	}
1110 done:
1111 	if (fd != -1 && close(fd) == -1 && err == NULL)
1112 		err = got_error_from_errno("close");
1113 	free(path);
1114 	if (lf)
1115 		unlock_err = got_lockfile_unlock(lf, -1);
1116 	return err ? err : unlock_err;
1117 }
1118 
1119 const struct got_error *
1120 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1121     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1122     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1123     got_cancel_cb cancel_cb, void *cancel_arg)
1124 {
1125 	const struct got_error *err;
1126 	struct got_object_idset *loose_ids;
1127 	struct got_object_idset *traversed_ids;
1128 	struct got_object_id **referenced_ids;
1129 	int i, nreferenced, nloose, ncommits = 0;
1130 	struct got_reflist_head refs;
1131 	struct got_reflist_entry *re;
1132 	struct purge_loose_object_arg arg;
1133 	time_t max_mtime = 0;
1134 	struct got_ratelimit rl;
1135 
1136 	TAILQ_INIT(&refs);
1137 	got_ratelimit_init(&rl, 0, 500);
1138 
1139 	*size_before = 0;
1140 	*size_after = 0;
1141 	*npacked = 0;
1142 
1143 	err = get_loose_object_ids(&loose_ids, size_before,
1144 	    progress_cb, progress_arg, &rl, repo);
1145 	if (err)
1146 		return err;
1147 	nloose = got_object_idset_num_elements(loose_ids);
1148 	if (nloose == 0) {
1149 		got_object_idset_free(loose_ids);
1150 		if (progress_cb) {
1151 			err = progress_cb(progress_arg, 0, 0, 0);
1152 			if (err)
1153 				return err;
1154 		}
1155 		return NULL;
1156 	}
1157 
1158 	traversed_ids = got_object_idset_alloc();
1159 	if (traversed_ids == NULL) {
1160 		err = got_error_from_errno("got_object_idset_alloc");
1161 		goto done;
1162 	}
1163 
1164 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1165 	if (err)
1166 		goto done;
1167 	if (!ignore_mtime) {
1168 		TAILQ_FOREACH(re, &refs, entry) {
1169 			time_t mtime = got_ref_get_mtime(re->ref);
1170 			if (mtime > max_mtime)
1171 				max_mtime = mtime;
1172 		}
1173 		/*
1174 		 * For safety, keep objects created within 10 minutes
1175 		 * before the youngest reference was created.
1176 		 */
1177 		if (max_mtime >= 600)
1178 			max_mtime -= 600;
1179 	}
1180 
1181 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1182 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1183 	    &refs, repo, cancel_cb, cancel_arg);
1184 	if (err)
1185 		goto done;
1186 
1187 	for (i = 0; i < nreferenced; i++) {
1188 		struct got_object_id *id = referenced_ids[i];
1189 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1190 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1191 		    nloose, cancel_cb, cancel_arg);
1192 		if (err)
1193 			goto done;
1194 	}
1195 
1196 	/* Any remaining loose objects are unreferenced and can be purged. */
1197 	arg.repo = repo;
1198 	arg.progress_arg = progress_arg;
1199 	arg.progress_cb = progress_cb;
1200 	arg.rl = &rl;
1201 	arg.nloose = nloose;
1202 	arg.npurged = 0;
1203 	arg.size_purged = 0;
1204 	arg.ncommits = ncommits;
1205 	arg.dry_run = dry_run;
1206 	arg.max_mtime = max_mtime;
1207 	arg.ignore_mtime = ignore_mtime;
1208 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1209 	if (err)
1210 		goto done;
1211 	*size_after = *size_before - arg.size_purged;
1212 
1213 	/* Produce a final progress report. */
1214 	if (progress_cb) {
1215 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1216 		if (err)
1217 			goto done;
1218 	}
1219 done:
1220 	got_object_idset_free(loose_ids);
1221 	got_object_idset_free(traversed_ids);
1222 	return err;
1223 }
1224 
1225 static const struct got_error *
1226 remove_packidx(int dir_fd, const char *relpath)
1227 {
1228 	const struct got_error *err, *unlock_err;
1229 	struct got_lockfile *lf;
1230 
1231 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1232 	if (err)
1233 		return err;
1234 	if (unlinkat(dir_fd, relpath, 0) == -1)
1235 		err = got_error_from_errno("unlinkat");
1236 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1237 	return err ? err : unlock_err;
1238 }
1239 
1240 const struct got_error *
1241 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1242     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1243     got_cancel_cb cancel_cb, void *cancel_arg)
1244 {
1245 	const struct got_error *err = NULL;
1246 	DIR *packdir = NULL;
1247 	struct dirent *dent;
1248 	char *pack_relpath = NULL;
1249 	int packdir_fd;
1250 	struct stat sb;
1251 
1252 	packdir_fd = openat(got_repo_get_fd(repo),
1253 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1254 	if (packdir_fd == -1) {
1255 		if (errno == ENOENT)
1256 			return NULL;
1257 		return got_error_from_errno_fmt("openat: %s/%s",
1258 		    got_repo_get_path_git_dir(repo),
1259 		    GOT_OBJECTS_PACK_DIR);
1260 	}
1261 
1262 	packdir = fdopendir(packdir_fd);
1263 	if (packdir == NULL) {
1264 		err = got_error_from_errno("fdopendir");
1265 		goto done;
1266 	}
1267 
1268 	while ((dent = readdir(packdir)) != NULL) {
1269 		if (cancel_cb) {
1270 			err = cancel_cb(cancel_arg);
1271 			if (err)
1272 				goto done;
1273 		}
1274 
1275 		if (!got_repo_is_packidx_filename(dent->d_name,
1276 		    strlen(dent->d_name)))
1277 			continue;
1278 
1279 		err = got_packidx_get_packfile_path(&pack_relpath,
1280 		    dent->d_name);
1281 		if (err)
1282 			goto done;
1283 
1284 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1285 			free(pack_relpath);
1286 			pack_relpath = NULL;
1287 			continue;
1288 		}
1289 		if (errno != ENOENT) {
1290 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1291 			    got_repo_get_path_git_dir(repo),
1292 			    GOT_OBJECTS_PACK_DIR,
1293 			    pack_relpath);
1294 			goto done;
1295 		}
1296 
1297 		if (!dry_run) {
1298 			err = remove_packidx(packdir_fd, dent->d_name);
1299 			if (err)
1300 				goto done;
1301 		}
1302 		if (progress_cb) {
1303 			char *path;
1304 			if (asprintf(&path, "%s/%s/%s",
1305 			    got_repo_get_path_git_dir(repo),
1306 			    GOT_OBJECTS_PACK_DIR,
1307 			    dent->d_name) == -1) {
1308 				err = got_error_from_errno("asprintf");
1309 				goto done;
1310 			}
1311 			err = progress_cb(progress_arg, path);
1312 			free(path);
1313 			if (err)
1314 				goto done;
1315 		}
1316 		free(pack_relpath);
1317 		pack_relpath = NULL;
1318 	}
1319 done:
1320 	if (packdir && closedir(packdir) != 0 && err == NULL)
1321 		err = got_error_from_errno("closedir");
1322 	free(pack_relpath);
1323 	return err;
1324 }