op public repos

Blob

Date:: Wed Jan 5 20:24:13 2022 UTC
Message:: print status output even if 'gotadmin cleanup' finds no objects to remove
Actions:: History | Blame | Raw File
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 <endian.h>
27 #include <errno.h>
28 #include <fcntl.h>
29 #include <stdint.h>
30 #include <sha1.h>
31 #include <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <limits.h>
35 #include <unistd.h>
36 #include <imsg.h>
37 
38 #include "got_error.h"
39 #include "got_cancel.h"
40 #include "got_object.h"
41 #include "got_reference.h"
42 #include "got_repository.h"
43 #include "got_repository_admin.h"
44 #include "got_opentemp.h"
45 #include "got_path.h"
46 
47 #include "got_lib_delta.h"
48 #include "got_lib_object.h"
49 #include "got_lib_object_idset.h"
50 #include "got_lib_object_cache.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_repository.h"
54 #include "got_lib_pack_create.h"
55 #include "got_lib_sha1.h"
56 #include "got_lib_lockfile.h"
57 #include "got_lib_ratelimit.h"
58 
59 #ifndef nitems
60 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
61 #endif
62 
63 static const struct got_error *
64 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
65     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
66     struct got_repository *repo,
67     got_cancel_cb cancel_cb, void *cancel_arg)
68 {
69 	const struct got_error *err = NULL;
70 	const size_t alloc_chunksz = 256;
71 	size_t nalloc;
72 	struct got_reflist_entry *re;
73 	int i;
74 
75 	*ids = NULL;
76 	*nobjects = 0;
77 
78 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
79 	if (*ids == NULL)
80 		return got_error_from_errno("reallocarray");
81 	nalloc = alloc_chunksz;
82 
83 	TAILQ_FOREACH(re, refs, entry) {
84 		struct got_object_id *id;
85 
86 		if (cancel_cb) {
87 			err = cancel_cb(cancel_arg);
88 			if (err)
89 				goto done;
90 		}
91 
92 		err = got_ref_resolve(&id, repo, re->ref);
93 		if (err)
94 			goto done;
95 
96 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
97 			int obj_type;
98 			err = got_object_get_type(&obj_type, repo, id);
99 			if (err)
100 				goto done;
101 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
102 				free(id);
103 				id = NULL;
104 				continue;
105 			}
106 		}
107 
108 		if (nalloc <= *nobjects) {
109 			struct got_object_id **new;
110 			new = recallocarray(*ids, nalloc,
111 			    nalloc + alloc_chunksz,
112 			    sizeof(struct got_object_id *));
113 			if (new == NULL) {
114 				err = got_error_from_errno(
115 				    "recallocarray");
116 				goto done;
117 			}
118 			*ids = new;
119 			nalloc += alloc_chunksz;
120 		}
121 		(*ids)[*nobjects] = id;
122 		if ((*ids)[*nobjects] == NULL) {
123 			err = got_error_from_errno("got_object_id_dup");
124 			goto done;
125 		}
126 		(*nobjects)++;
127 	}
128 done:
129 	if (err) {
130 		for (i = 0; i < *nobjects; i++)
131 			free((*ids)[i]);
132 		free(*ids);
133 		*ids = NULL;
134 		*nobjects = 0;
135 	}
136 	return err;
137 }
138 
139 const struct got_error *
140 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
141     struct got_reflist_head *include_refs,
142     struct got_reflist_head *exclude_refs, struct got_repository *repo,
143     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
144     got_cancel_cb cancel_cb, void *cancel_arg)
145 {
146 	const struct got_error *err = NULL;
147 	struct got_object_id **ours = NULL, **theirs = NULL;
148 	int nours = 0, ntheirs = 0, packfd = -1, i;
149 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
150 	char *sha1_str = NULL;
151 
152 	*packfile = NULL;
153 	*pack_hash = NULL;
154 
155 	if (asprintf(&path, "%s/%s/packing.pack",
156 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
157 		err = got_error_from_errno("asprintf");
158 		goto done;
159 	}
160 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
161 	if (err)
162 		goto done;
163 
164 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
165 		err = got_error_from_errno2("fchmod", tmpfile_path);
166 		goto done;
167 	}
168 
169 	*packfile = fdopen(packfd, "w");
170 	if (*packfile == NULL) {
171 		err = got_error_from_errno2("fdopen", tmpfile_path);
172 		goto done;
173 	}
174 	packfd = -1;
175 
176 	err = get_reflist_object_ids(&ours, &nours,
177 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
178 	    include_refs, repo, cancel_cb, cancel_arg);
179 	if (err)
180 		goto done;
181 
182 	if (nours == 0) {
183 		err = got_error(GOT_ERR_CANNOT_PACK);
184 		goto done;
185 	}
186 
187 	if (!TAILQ_EMPTY(exclude_refs)) {
188 		err = get_reflist_object_ids(&theirs, &ntheirs,
189 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
190 		    exclude_refs, repo,
191 		    cancel_cb, cancel_arg);
192 		if (err)
193 			goto done;
194 	}
195 
196 	*pack_hash = calloc(1, sizeof(**pack_hash));
197 	if (*pack_hash == NULL) {
198 		err = got_error_from_errno("calloc");
199 		goto done;
200 	}
201 
202 	err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
203 	    ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
204 	    cancel_cb, cancel_arg);
205 	if (err)
206 		goto done;
207 
208 	err = got_object_id_str(&sha1_str, *pack_hash);
209 	if (err)
210 		goto done;
211 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
212 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
213 	    sha1_str) == -1) {
214 		err = got_error_from_errno("asprintf");
215 		goto done;
216 	}
217 
218 	if (fflush(*packfile) == -1) {
219 		err = got_error_from_errno("fflush");
220 		goto done;
221 	}
222 	if (fseek(*packfile, 0L, SEEK_SET) == -1) {
223 		err = got_error_from_errno("fseek");
224 		goto done;
225 	}
226 	if (rename(tmpfile_path, packfile_path) == -1) {
227 		err = got_error_from_errno3("rename", tmpfile_path,
228 		    packfile_path);
229 		goto done;
230 	}
231 	free(tmpfile_path);
232 	tmpfile_path = NULL;
233 done:
234 	for (i = 0; i < nours; i++)
235 		free(ours[i]);
236 	free(ours);
237 	for (i = 0; i < ntheirs; i++)
238 		free(theirs[i]);
239 	free(theirs);
240 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
241 		err = got_error_from_errno2("close", packfile_path);
242 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
243 		err = got_error_from_errno2("unlink", tmpfile_path);
244 	free(tmpfile_path);
245 	free(packfile_path);
246 	free(sha1_str);
247 	free(path);
248 	if (err) {
249 		free(*pack_hash);
250 		*pack_hash = NULL;
251 		if (*packfile)
252 			fclose(*packfile);
253 		*packfile = NULL;
254 	}
255 	return err;
256 }
257 
258 const struct got_error *
259 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
260     struct got_repository *repo,
261     got_pack_index_progress_cb progress_cb, void *progress_arg,
262     got_cancel_cb cancel_cb, void *cancel_arg)
263 {
264 	size_t i;
265 	char *path;
266 	int imsg_idxfds[2];
267 	int npackfd = -1, idxfd = -1, nidxfd = -1;
268 	int tmpfds[3];
269 	int idxstatus, done = 0;
270 	const struct got_error *err;
271 	struct imsgbuf idxibuf;
272 	pid_t idxpid;
273 	char *tmpidxpath = NULL;
274 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
275 	const char *repo_path = got_repo_get_path_git_dir(repo);
276 	struct stat sb;
277 
278 	for (i = 0; i < nitems(tmpfds); i++)
279 		tmpfds[i] = -1;
280 
281 	if (asprintf(&path, "%s/%s/indexing.idx",
282 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
283 		err = got_error_from_errno("asprintf");
284 		goto done;
285 	}
286 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
287 	free(path);
288 	if (err)
289 		goto done;
290 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
291 		err = got_error_from_errno2("fchmod", tmpidxpath);
292 		goto done;
293 	}
294 
295 	nidxfd = dup(idxfd);
296 	if (nidxfd == -1) {
297 		err = got_error_from_errno("dup");
298 		goto done;
299 	}
300 
301 	for (i = 0; i < nitems(tmpfds); i++) {
302 		tmpfds[i] = got_opentempfd();
303 		if (tmpfds[i] == -1) {
304 			err = got_error_from_errno("got_opentempfd");
305 			goto done;
306 		}
307 	}
308 
309 	err = got_object_id_str(&id_str, pack_hash);
310 	if (err)
311 		goto done;
312 
313 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
314 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
315 		err = got_error_from_errno("asprintf");
316 		goto done;
317 	}
318 
319 	if (fstat(fileno(packfile), &sb) == -1) {
320 		err = got_error_from_errno2("fstat", packfile_path);
321 		goto done;
322 	}
323 
324 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
325 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
326 		err = got_error_from_errno("asprintf");
327 		goto done;
328 	}
329 
330 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
331 		err = got_error_from_errno("socketpair");
332 		goto done;
333 	}
334 	idxpid = fork();
335 	if (idxpid == -1) {
336 		err= got_error_from_errno("fork");
337 		goto done;
338 	} else if (idxpid == 0)
339 		got_privsep_exec_child(imsg_idxfds,
340 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
341 	if (close(imsg_idxfds[1]) == -1) {
342 		err = got_error_from_errno("close");
343 		goto done;
344 	}
345 	imsg_init(&idxibuf, imsg_idxfds[0]);
346 
347 	npackfd = dup(fileno(packfile));
348 	if (npackfd == -1) {
349 		err = got_error_from_errno("dup");
350 		goto done;
351 	}
352 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
353 	    npackfd);
354 	if (err != NULL)
355 		goto done;
356 	npackfd = -1;
357 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
358 	if (err != NULL)
359 		goto done;
360 	nidxfd = -1;
361 	for (i = 0; i < nitems(tmpfds); i++) {
362 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
363 		if (err != NULL)
364 			goto done;
365 		tmpfds[i] = -1;
366 	}
367 	done = 0;
368 	while (!done) {
369 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
370 
371 		if (cancel_cb) {
372 			err = cancel_cb(cancel_arg);
373 			if (err)
374 				goto done;
375 		}
376 
377 		err = got_privsep_recv_index_progress(&done, &nobj_total,
378 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
379 		    &idxibuf);
380 		if (err != NULL)
381 			goto done;
382 		if (nobj_indexed != 0) {
383 			err = progress_cb(progress_arg, sb.st_size,
384 			    nobj_total, nobj_indexed, nobj_loose,
385 			    nobj_resolved);
386 			if (err)
387 				break;
388 		}
389 		imsg_clear(&idxibuf);
390 	}
391 	if (close(imsg_idxfds[0]) == -1) {
392 		err = got_error_from_errno("close");
393 		goto done;
394 	}
395 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
396 		err = got_error_from_errno("waitpid");
397 		goto done;
398 	}
399 
400 	if (rename(tmpidxpath, idxpath) == -1) {
401 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
402 		goto done;
403 	}
404 	free(tmpidxpath);
405 	tmpidxpath = NULL;
406 
407 done:
408 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
409 		err = got_error_from_errno2("unlink", tmpidxpath);
410 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
411 		err = got_error_from_errno("close");
412 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
413 		err = got_error_from_errno("close");
414 	for (i = 0; i < nitems(tmpfds); i++) {
415 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
416 			err = got_error_from_errno("close");
417 	}
418 	free(tmpidxpath);
419 	free(idxpath);
420 	free(packfile_path);
421 	return err;
422 }
423 
424 const struct got_error *
425 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
426     struct got_repository *repo, const char *packfile_path)
427 {
428 	const struct got_error *err = NULL;
429 	const char *packdir_path = NULL;
430 	char *packfile_name = NULL, *p, *dot;
431 	struct got_object_id id;
432 	int packfd = -1;
433 
434 	*packfile = NULL;
435 	*pack_hash = NULL;
436 
437 	packdir_path = got_repo_get_path_objects_pack(repo);
438 	if (packdir_path == NULL)
439 		return got_error_from_errno("got_repo_get_path_objects_pack");
440 
441 	if (!got_path_is_child(packfile_path, packdir_path,
442 	    strlen(packdir_path))) {
443 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
444 		goto done;
445 
446 	}
447 
448 	err = got_path_basename(&packfile_name, packfile_path);
449 	if (err)
450 		goto done;
451 	p = packfile_name;
452 
453 	if (strncmp(p, "pack-", 5) != 0) {
454 		err = got_error_fmt(GOT_ERR_BAD_PATH,
455 		    "'%s' is not a valid pack file name",
456 		    packfile_name);
457 		goto done;
458 	}
459 	p += 5;
460 	dot = strchr(p, '.');
461 	if (dot == NULL) {
462 		err = got_error_fmt(GOT_ERR_BAD_PATH,
463 		    "'%s' is not a valid pack file name",
464 		    packfile_name);
465 		goto done;
466 	}
467 	if (strcmp(dot + 1, "pack") != 0) {
468 		err = got_error_fmt(GOT_ERR_BAD_PATH,
469 		    "'%s' is not a valid pack file name",
470 		    packfile_name);
471 		goto done;
472 	}
473 	*dot = '\0';
474 	if (!got_parse_sha1_digest(id.sha1, p)) {
475 		err = got_error_fmt(GOT_ERR_BAD_PATH,
476 		    "'%s' is not a valid pack file name",
477 		    packfile_name);
478 		goto done;
479 	}
480 
481 	*pack_hash = got_object_id_dup(&id);
482 	if (*pack_hash == NULL) {
483 		err = got_error_from_errno("got_object_id_dup");
484 		goto done;
485 	}
486 
487 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
488 	if (packfd == -1) {
489 		err = got_error_from_errno2("open", packfile_path);
490 		goto done;
491 	}
492 
493 	*packfile = fdopen(packfd, "r");
494 	if (*packfile == NULL) {
495 		err = got_error_from_errno2("fdopen", packfile_path);
496 		goto done;
497 	}
498 	packfd = -1;
499 done:
500 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
501 		err = got_error_from_errno2("close", packfile_path);
502 	free(packfile_name);
503 	if (err) {
504 		free(*pack_hash);
505 		*pack_hash = NULL;
506 	}
507 	return err;
508 }
509 
510 const struct got_error *
511 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
512     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
513     got_cancel_cb cancel_cb, void *cancel_arg)
514 {
515 	const struct got_error *err = NULL;
516 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
517 	struct got_packidx *packidx = NULL;
518 	struct got_pack *pack = NULL;
519 	uint32_t nobj, i;
520 
521 	err = got_object_id_str(&id_str, pack_hash);
522 	if (err)
523 		goto done;
524 
525 	if (asprintf(&packpath, "%s/pack-%s.pack",
526 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
527 		err = got_error_from_errno("asprintf");
528 		goto done;
529 	}
530 	if (asprintf(&idxpath, "%s/pack-%s.idx",
531 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
532 		err = got_error_from_errno("asprintf");
533 		goto done;
534 	}
535 
536 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
537 	if (err)
538 		goto done;
539 
540 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
541 	if (err)
542 		goto done;
543 
544 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
545 	for (i = 0; i < nobj; i++) {
546 		struct got_packidx_object_id *oid;
547 		struct got_object_id id, base_id;
548 		off_t offset, base_offset = 0;
549 		uint8_t type;
550 		uint64_t size;
551 		size_t tslen, len;
552 
553 		if (cancel_cb) {
554 			err = cancel_cb(cancel_arg);
555 			if (err)
556 				break;
557 		}
558 		oid = &packidx->hdr.sorted_ids[i];
559 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
560 
561 		offset = got_packidx_get_object_offset(packidx, i);
562 		if (offset == -1) {
563 			err = got_error(GOT_ERR_BAD_PACKIDX);
564 			goto done;
565 		}
566 
567 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
568 		    pack, offset);
569 		if (err)
570 			goto done;
571 
572 		switch (type) {
573 		case GOT_OBJ_TYPE_OFFSET_DELTA:
574 			err = got_pack_parse_offset_delta(&base_offset, &len,
575 			    pack, offset, tslen);
576 			if (err)
577 				goto done;
578 			break;
579 		case GOT_OBJ_TYPE_REF_DELTA:
580 			err = got_pack_parse_ref_delta(&base_id,
581 			    pack, offset, tslen);
582 			if (err)
583 				goto done;
584 			break;
585 		}
586 		err = (*list_cb)(list_arg, &id, type, offset, size,
587 		    base_offset, &base_id);
588 		if (err)
589 			goto done;
590 	}
591 
592 done:
593 	free(id_str);
594 	free(idxpath);
595 	free(packpath);
596 	if (packidx)
597 		got_packidx_close(packidx);
598 	return err;
599 }
600 
601 static const struct got_error *
602 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
603     void *progress_arg, struct got_ratelimit *rl,
604     int nloose, int ncommits, int npurged)
605 {
606 	const struct got_error *err;
607 	int elapsed;
608 
609 	if (progress_cb == NULL)
610 		return NULL;
611 
612 	err = got_ratelimit_check(&elapsed, rl);
613 	if (err || !elapsed)
614 		return err;
615 
616 	return progress_cb(progress_arg, nloose, ncommits, npurged);
617 }
618 
619 static const struct got_error *
620 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
621     got_cleanup_progress_cb progress_cb, void *progress_arg,
622     struct got_ratelimit *rl, struct got_repository *repo)
623 {
624 	const struct got_error *err = NULL;
625 	char *path_objects = NULL, *path = NULL;
626 	DIR *dir = NULL;
627 	struct got_object *obj = NULL;
628 	struct got_object_id id;
629 	int i, fd = -1;
630 	struct stat sb;
631 
632 	*ondisk_size = 0;
633 	*loose_ids = got_object_idset_alloc();
634 	if (*loose_ids == NULL)
635 		return got_error_from_errno("got_object_idset_alloc");
636 
637 	path_objects = got_repo_get_path_objects(repo);
638 	if (path_objects == NULL) {
639 		err = got_error_from_errno("got_repo_get_path_objects");
640 		goto done;
641 	}
642 
643 	for (i = 0; i <= 0xff; i++) {
644 		struct dirent *dent;
645 
646 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
647 			err = got_error_from_errno("asprintf");
648 			break;
649 		}
650 
651 		dir = opendir(path);
652 		if (dir == NULL) {
653 			if (errno == ENOENT) {
654 				err = NULL;
655 				continue;
656 			}
657 			err = got_error_from_errno2("opendir", path);
658 			break;
659 		}
660 
661 		while ((dent = readdir(dir)) != NULL) {
662 			char *id_str;
663 
664 			if (strcmp(dent->d_name, ".") == 0 ||
665 			    strcmp(dent->d_name, "..") == 0)
666 				continue;
667 
668 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
669 				err = got_error_from_errno("asprintf");
670 				goto done;
671 			}
672 
673 			memset(&id, 0, sizeof(id));
674 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
675 				free(id_str);
676 				continue;
677 			}
678 			free(id_str);
679 
680 			err = got_object_open_loose_fd(&fd, &id, repo);
681 			if (err)
682 				goto done;
683 			if (fstat(fd, &sb) == -1) {
684 				err = got_error_from_errno("fstat");
685 				goto done;
686 			}
687 			err = got_object_read_header_privsep(&obj, &id, repo,
688 			    fd);
689 			if (err)
690 				goto done;
691 			fd = -1; /* already closed */
692 
693 			switch (obj->type) {
694 			case GOT_OBJ_TYPE_COMMIT:
695 			case GOT_OBJ_TYPE_TREE:
696 			case GOT_OBJ_TYPE_BLOB:
697 			case GOT_OBJ_TYPE_TAG:
698 				break;
699 			default:
700 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
701 				    "%d", obj->type);
702 				goto done;
703 			}
704 			got_object_close(obj);
705 			obj = NULL;
706 			(*ondisk_size) += sb.st_size;
707 			err = got_object_idset_add(*loose_ids, &id, NULL);
708 			if (err)
709 				goto done;
710 			err = report_cleanup_progress(progress_cb,
711 			    progress_arg, rl,
712 			    got_object_idset_num_elements(*loose_ids), -1, -1);
713 			if (err)
714 				goto done;
715 		}
716 
717 		if (closedir(dir) != 0) {
718 			err = got_error_from_errno("closedir");
719 			goto done;
720 		}
721 		dir = NULL;
722 
723 		free(path);
724 		path = NULL;
725 	}
726 done:
727 	if (dir && closedir(dir) != 0 && err == NULL)
728 		err = got_error_from_errno("closedir");
729 	if (fd != -1 && close(fd) == -1 && err == NULL)
730 		err = got_error_from_errno("close");
731 	if (err) {
732 		got_object_idset_free(*loose_ids);
733 		*loose_ids = NULL;
734 	}
735 	if (obj)
736 		got_object_close(obj);
737 	free(path_objects);
738 	free(path);
739 	return err;
740 }
741 
742 static const struct got_error *
743 preserve_loose_object(struct got_object_idset *loose_ids,
744     struct got_object_id *id, struct got_repository *repo, int *npacked)
745 {
746 	const struct got_error *err = NULL;
747 	struct got_object *obj;
748 
749 	if (!got_object_idset_contains(loose_ids, id))
750 		return NULL;
751 
752 	/*
753 	 * Try to open this object from a pack file. This ensures that
754 	 * we do in fact have a valid packed copy of the object. Otherwise
755 	 * we should not delete the loose representation of this object.
756 	 */
757 	err = got_object_open_packed(&obj, id, repo);
758 	if (err == NULL) {
759 		got_object_close(obj);
760 		/*
761 		 * The object is referenced and packed.
762 		 * We can purge the redundantly stored loose object.
763 		 */
764 		(*npacked)++;
765 		return NULL;
766 	} else if (err->code != GOT_ERR_NO_OBJ)
767 		return err;
768 
769 	/*
770 	 * This object is referenced and not packed.
771 	 * Remove it from our purge set.
772 	 */
773 	return got_object_idset_remove(NULL, loose_ids, id);
774 }
775 
776 static const struct got_error *
777 load_tree_entries(struct got_object_id_queue *ids,
778     struct got_object_idset *loose_ids,
779     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
780     const char *dpath, struct got_repository *repo, int *npacked,
781     got_cancel_cb cancel_cb, void *cancel_arg)
782 {
783 	const struct got_error *err;
784 	struct got_tree_object *tree;
785 	char *p = NULL;
786 	int i;
787 
788 	err = got_object_open_as_tree(&tree, repo, tree_id);
789 	if (err)
790 		return err;
791 
792 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
793 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
794 		struct got_object_id *id = got_tree_entry_get_id(e);
795 		mode_t mode = got_tree_entry_get_mode(e);
796 
797 		if (cancel_cb) {
798 			err = (*cancel_cb)(cancel_arg);
799 			if (err)
800 				break;
801 		}
802 
803 		if (got_object_tree_entry_is_symlink(e) ||
804 		    got_object_tree_entry_is_submodule(e) ||
805 		    got_object_idset_contains(traversed_ids, id))
806 			continue;
807 
808 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
809 		    got_tree_entry_get_name(e)) == -1) {
810 			err = got_error_from_errno("asprintf");
811 			break;
812 		}
813 
814 		if (S_ISDIR(mode)) {
815 			struct got_object_qid *qid;
816 			err = got_object_qid_alloc(&qid, id);
817 			if (err)
818 				break;
819 			STAILQ_INSERT_TAIL(ids, qid, entry);
820 		} else if (S_ISREG(mode)) {
821 			/* This blob is referenced. */
822 			err = preserve_loose_object(loose_ids, id, repo,
823 			    npacked);
824 			if (err)
825 				break;
826 			err = got_object_idset_add(traversed_ids, id, NULL);
827 			if (err)
828 				break;
829 			
830 		}
831 		free(p);
832 		p = NULL;
833 	}
834 
835 	got_object_tree_close(tree);
836 	free(p);
837 	return err;
838 }
839 
840 static const struct got_error *
841 load_tree(struct got_object_idset *loose_ids,
842     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
843     const char *dpath, struct got_repository *repo, int *npacked,
844     got_cancel_cb cancel_cb, void *cancel_arg)
845 {
846 	const struct got_error *err = NULL;
847 	struct got_object_id_queue tree_ids;
848 	struct got_object_qid *qid;
849 
850 	err = got_object_qid_alloc(&qid, tree_id);
851 	if (err)
852 		return err;
853 
854 	STAILQ_INIT(&tree_ids);
855 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
856 
857 	while (!STAILQ_EMPTY(&tree_ids)) {
858 		if (cancel_cb) {
859 			err = (*cancel_cb)(cancel_arg);
860 			if (err)
861 				break;
862 		}
863 
864 		qid = STAILQ_FIRST(&tree_ids);
865 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
866 
867 		if (got_object_idset_contains(traversed_ids, qid->id)) {
868 			got_object_qid_free(qid);
869 			continue;
870 		}
871 
872 		err = got_object_idset_add(traversed_ids, qid->id, NULL);
873 		if (err) {
874 			got_object_qid_free(qid);
875 			break;
876 		}
877 
878 		/* This tree is referenced. */
879 		err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
880 		if (err)
881 			break;
882 
883 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
884 		    qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
885 		got_object_qid_free(qid);
886 		if (err)
887 			break;
888 	}
889 
890 	got_object_id_queue_free(&tree_ids);
891 	return err;
892 }
893 
894 static const struct got_error *
895 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
896     int *npacked, struct got_object_idset *traversed_ids,
897     struct got_object_id *id, struct got_repository *repo,
898     got_cleanup_progress_cb progress_cb, void *progress_arg,
899     struct got_ratelimit *rl, int nloose,
900     got_cancel_cb cancel_cb, void *cancel_arg)
901 {
902 	const struct got_error *err;
903 	struct got_commit_object *commit = NULL;
904 	struct got_tag_object *tag = NULL;
905 	struct got_object_id *tree_id = NULL;
906 	struct got_object_id_queue ids;
907 	struct got_object_qid *qid;
908 	int obj_type;
909 
910 	err = got_object_qid_alloc(&qid, id);
911 	if (err)
912 		return err;
913 
914 	STAILQ_INIT(&ids);
915 	STAILQ_INSERT_TAIL(&ids, qid, entry);
916 
917 	while (!STAILQ_EMPTY(&ids)) {
918 		if (cancel_cb) {
919 			err = (*cancel_cb)(cancel_arg);
920 			if (err)
921 				break;
922 		}
923 
924 		qid = STAILQ_FIRST(&ids);
925 		STAILQ_REMOVE_HEAD(&ids, entry);
926 
927 		if (got_object_idset_contains(traversed_ids, qid->id)) {
928 			got_object_qid_free(qid);
929 			qid = NULL;
930 			continue;
931 		}
932 
933 		err = got_object_idset_add(traversed_ids, qid->id, NULL);
934 		if (err)
935 			break;
936 
937 		/* This commit or tag is referenced. */
938 		err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
939 		if (err)
940 			break;
941 
942 		err = got_object_get_type(&obj_type, repo, qid->id);
943 		if (err)
944 			break;
945 		switch (obj_type) {
946 		case GOT_OBJ_TYPE_COMMIT:
947 			err = got_object_open_as_commit(&commit, repo, qid->id);
948 			if (err)
949 				goto done;
950 			break;
951 		case GOT_OBJ_TYPE_TAG:
952 			err = got_object_open_as_tag(&tag, repo, qid->id);
953 			if (err)
954 				goto done;
955 			break;
956 		default:
957 			/* should not happen */
958 			err = got_error(GOT_ERR_OBJ_TYPE);
959 			goto done;
960 		}
961 
962 		/* Find a tree object to scan. */
963 		if (commit) {
964 			tree_id = got_object_commit_get_tree_id(commit);
965 		} else if (tag) {
966 			obj_type = got_object_tag_get_object_type(tag);
967 			switch (obj_type) {
968 			case GOT_OBJ_TYPE_COMMIT:
969 				err = got_object_open_as_commit(&commit, repo,
970 				    got_object_tag_get_object_id(tag));
971 				if (err)
972 					goto done;
973 				tree_id = got_object_commit_get_tree_id(commit);
974 				break;
975 			case GOT_OBJ_TYPE_TREE:
976 				tree_id = got_object_tag_get_object_id(tag);
977 				break;
978 			default:
979 				/*
980 				 * Tag points at something other than a
981 				 * commit or tree. Leave this weird tag object
982 				 * and the object it points to on disk.
983 				 */
984 				err = got_object_idset_remove(NULL, loose_ids,
985 				    qid->id);
986 				if (err && err->code != GOT_ERR_NO_OBJ)
987 					goto done;
988 				err = got_object_idset_remove(NULL, loose_ids,
989 				    got_object_tag_get_object_id(tag));
990 				if (err && err->code != GOT_ERR_NO_OBJ)
991 					goto done;
992 				err = NULL;
993 				break;
994 			}
995 		}
996 
997 		if (tree_id) {
998 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
999 			    repo, npacked, cancel_cb, cancel_arg);
1000 			if (err)
1001 				break;
1002 		}
1003 
1004 		if (commit || tag)
1005 			(*ncommits)++; /* scanned tags are counted as commits */
1006 
1007 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1008 		    nloose, *ncommits, -1);
1009 		if (err)
1010 			break;
1011  
1012 		if (commit) {
1013 			/* Find parent commits to scan. */
1014 			const struct got_object_id_queue *parent_ids;
1015 			parent_ids = got_object_commit_get_parent_ids(commit);
1016 			err = got_object_id_queue_copy(parent_ids, &ids);
1017 			if (err)
1018 				break;
1019 			got_object_commit_close(commit);
1020 			commit = NULL;
1021 		}
1022 		if (tag) {
1023 			got_object_tag_close(tag);
1024 			tag = NULL;
1025 		}
1026 		got_object_qid_free(qid);
1027 		qid = NULL;
1028 	}
1029 done:
1030 	if (qid)
1031 		got_object_qid_free(qid);
1032 	if (commit)
1033 		got_object_commit_close(commit);
1034 	if (tag)
1035 		got_object_tag_close(tag);
1036 	got_object_id_queue_free(&ids);
1037 	return err;
1038 }
1039 
1040 struct purge_loose_object_arg {
1041 	struct got_repository *repo;
1042 	got_cleanup_progress_cb progress_cb;
1043 	void *progress_arg;
1044 	struct got_ratelimit *rl;
1045 	int nloose;
1046 	int ncommits;
1047 	int npurged;
1048 	off_t size_purged;
1049 	int dry_run;
1050 	time_t max_mtime;
1051 	int ignore_mtime;
1052 };
1053 
1054 static const struct got_error *
1055 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1056 {
1057 	struct purge_loose_object_arg *a = arg;
1058 	const struct got_error *err, *unlock_err = NULL;
1059 	char *path = NULL;
1060 	int fd = -1;
1061 	struct stat sb;
1062 	struct got_lockfile *lf = NULL;
1063 
1064 	err = got_object_get_path(&path, id, a->repo);
1065 	if (err)
1066 		return err;
1067 
1068 	err = got_object_open_loose_fd(&fd, id, a->repo);
1069 	if (err)
1070 		goto done;
1071 
1072 	if (fstat(fd, &sb) == -1) {
1073 		err = got_error_from_errno("fstat");
1074 		goto done;
1075 	}
1076 
1077 	/*
1078 	 * Do not delete objects which are younger than our maximum
1079 	 * modification time threshold. This prevents a race where
1080 	 * new objects which are being added to the repository
1081 	 * concurrently would be deleted.
1082 	 */
1083 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1084 		if (!a->dry_run) {
1085 			err = got_lockfile_lock(&lf, path, -1);
1086 			if (err)
1087 				goto done;
1088 			if (unlink(path) == -1) {
1089 				err = got_error_from_errno2("unlink", path);
1090 				goto done;
1091 			}
1092 		}
1093 
1094 		a->npurged++;
1095 		a->size_purged += sb.st_size;
1096 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1097 		    a->rl, a->nloose, a->ncommits, a->npurged);
1098 		if (err)
1099 			goto done;
1100 	}
1101 done:
1102 	if (fd != -1 && close(fd) == -1 && err == NULL)
1103 		err = got_error_from_errno("close");
1104 	free(path);
1105 	if (lf)
1106 		unlock_err = got_lockfile_unlock(lf, -1);
1107 	return err ? err : unlock_err;
1108 }
1109 
1110 const struct got_error *
1111 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1112     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1113     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1114     got_cancel_cb cancel_cb, void *cancel_arg)
1115 {
1116 	const struct got_error *err;
1117 	struct got_object_idset *loose_ids;
1118 	struct got_object_idset *traversed_ids;
1119 	struct got_object_id **referenced_ids;
1120 	int i, nreferenced, nloose, ncommits = 0;
1121 	struct got_reflist_head refs;
1122 	struct got_reflist_entry *re;
1123 	struct purge_loose_object_arg arg;
1124 	time_t max_mtime = 0;
1125 	struct got_ratelimit rl;
1126 
1127 	TAILQ_INIT(&refs);
1128 	got_ratelimit_init(&rl, 0, 500);
1129 
1130 	*size_before = 0;
1131 	*size_after = 0;
1132 	*npacked = 0;
1133 
1134 	err = get_loose_object_ids(&loose_ids, size_before,
1135 	    progress_cb, progress_arg, &rl, repo);
1136 	if (err)
1137 		return err;
1138 	nloose = got_object_idset_num_elements(loose_ids);
1139 	if (nloose == 0) {
1140 		got_object_idset_free(loose_ids);
1141 		if (progress_cb) {
1142 			err = progress_cb(progress_arg, 0, 0, 0);
1143 			if (err)
1144 				return err;
1145 		}
1146 		return NULL;
1147 	}
1148 
1149 	traversed_ids = got_object_idset_alloc();
1150 	if (traversed_ids == NULL) {
1151 		err = got_error_from_errno("got_object_idset_alloc");
1152 		goto done;
1153 	}
1154 
1155 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1156 	if (err)
1157 		goto done;
1158 	if (!ignore_mtime) {
1159 		TAILQ_FOREACH(re, &refs, entry) {
1160 			time_t mtime = got_ref_get_mtime(re->ref);
1161 			if (mtime > max_mtime)
1162 				max_mtime = mtime;
1163 		}
1164 		/*
1165 		 * For safety, keep objects created within 10 minutes
1166 		 * before the youngest reference was created.
1167 		 */
1168 		if (max_mtime >= 600)
1169 			max_mtime -= 600;
1170 	}
1171 
1172 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1173 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1174 	    &refs, repo, cancel_cb, cancel_arg);
1175 	if (err)
1176 		goto done;
1177 
1178 	for (i = 0; i < nreferenced; i++) {
1179 		struct got_object_id *id = referenced_ids[i];
1180 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1181 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1182 		    nloose, cancel_cb, cancel_arg);
1183 		if (err)
1184 			goto done;
1185 	}
1186 
1187 	/* Any remaining loose objects are unreferenced and can be purged. */
1188 	arg.repo = repo;
1189 	arg.progress_arg = progress_arg;
1190 	arg.progress_cb = progress_cb;
1191 	arg.rl = &rl;
1192 	arg.nloose = nloose;
1193 	arg.npurged = 0;
1194 	arg.size_purged = 0;
1195 	arg.ncommits = ncommits;
1196 	arg.dry_run = dry_run;
1197 	arg.max_mtime = max_mtime;
1198 	arg.ignore_mtime = ignore_mtime;
1199 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1200 	if (err)
1201 		goto done;
1202 	*size_after = *size_before - arg.size_purged;
1203 
1204 	/* Produce a final progress report. */
1205 	if (progress_cb) {
1206 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1207 		if (err)
1208 			goto done;
1209 	}
1210 done:
1211 	got_object_idset_free(loose_ids);
1212 	got_object_idset_free(traversed_ids);
1213 	return err;
1214 }
1215 
1216 static const struct got_error *
1217 remove_packidx(int dir_fd, const char *relpath)
1218 {
1219 	const struct got_error *err, *unlock_err;
1220 	struct got_lockfile *lf;
1221 
1222 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1223 	if (err)
1224 		return err;
1225 	if (unlinkat(dir_fd, relpath, 0) == -1)
1226 		err = got_error_from_errno("unlinkat");
1227 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1228 	return err ? err : unlock_err;
1229 }
1230 
1231 const struct got_error *
1232 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1233     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1234     got_cancel_cb cancel_cb, void *cancel_arg)
1235 {
1236 	const struct got_error *err;
1237 	DIR *packdir = NULL;
1238 	struct dirent *dent;
1239 	char *pack_relpath = NULL;
1240 	int packdir_fd;
1241 	struct stat sb;
1242 
1243 	packdir_fd = openat(got_repo_get_fd(repo),
1244 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1245 	if (packdir_fd == -1) {
1246 		if (errno == ENOENT)
1247 			return NULL;
1248 		return got_error_from_errno_fmt("openat: %s/%s",
1249 		    got_repo_get_path_git_dir(repo),
1250 		    GOT_OBJECTS_PACK_DIR);
1251 	}
1252 
1253 	packdir = fdopendir(packdir_fd);
1254 	if (packdir == NULL) {
1255 		err = got_error_from_errno("fdopendir");
1256 		goto done;
1257 	}
1258 
1259 	while ((dent = readdir(packdir)) != NULL) {
1260 		if (cancel_cb) {
1261 			err = cancel_cb(cancel_arg);
1262 			if (err)
1263 				goto done;
1264 		}
1265 
1266 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1267 			continue;
1268 
1269 		err = got_packidx_get_packfile_path(&pack_relpath,
1270 		    dent->d_name);
1271 		if (err)
1272 			goto done;
1273 
1274 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1275 			free(pack_relpath);
1276 			pack_relpath = NULL;
1277 			continue;
1278 		}
1279 		if (errno != ENOENT) {
1280 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1281 			    got_repo_get_path_git_dir(repo),
1282 			    GOT_OBJECTS_PACK_DIR,
1283 			    pack_relpath);
1284 			goto done;
1285 		}
1286 
1287 		if (!dry_run) {
1288 			err = remove_packidx(packdir_fd, dent->d_name);
1289 			if (err)
1290 				goto done;
1291 		}
1292 		if (progress_cb) {
1293 			char *path;
1294 			if (asprintf(&path, "%s/%s/%s",
1295 			    got_repo_get_path_git_dir(repo),
1296 			    GOT_OBJECTS_PACK_DIR,
1297 			    dent->d_name) == -1) {
1298 				err = got_error_from_errno("asprintf");
1299 				goto done;
1300 			}
1301 			err = progress_cb(progress_arg, path);
1302 			free(path);
1303 			if (err)
1304 				goto done;
1305 		}
1306 		free(pack_relpath);
1307 		pack_relpath = NULL;
1308 	}
1309 done:
1310 	if (packdir && closedir(packdir) != 0 && err == NULL)
1311 		err = got_error_from_errno("closedir");
1312 	free(pack_relpath);
1313 	return err;
1314 }