Gotweb

Blob

Date:: Fri Apr 22 11:37:09 2022 UTC
Message:: inline struct got_object_id in struct got_object_qid Saves us from doing a malloc/free call for every item on the list. ok op@
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/uio.h>
19 #include <sys/stat.h>
20 #include <sys/socket.h>
21 #include <sys/wait.h>
22 
23 #include <dirent.h>
24 #include <errno.h>
25 #include <fcntl.h>
26 #include <stdint.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <limits.h>
31 #include <unistd.h>
32 
33 #include "got_error.h"
34 #include "got_cancel.h"
35 #include "got_object.h"
36 #include "got_reference.h"
37 #include "got_repository.h"
38 #include "got_repository_admin.h"
39 #include "got_opentemp.h"
40 #include "got_path.h"
41 
42 #include "got_lib_delta.h"
43 #include "got_lib_object.h"
44 #include "got_lib_object_idset.h"
45 #include "got_lib_object_cache.h"
46 #include "got_lib_pack.h"
47 #include "got_lib_privsep.h"
48 #include "got_lib_repository.h"
49 #include "got_lib_pack_create.h"
50 #include "got_lib_sha1.h"
51 #include "got_lib_lockfile.h"
52 #include "got_lib_ratelimit.h"
53 
54 #ifndef nitems
55 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
56 #endif
57 
58 static const struct got_error *
59 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
60     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
61     struct got_repository *repo,
62     got_cancel_cb cancel_cb, void *cancel_arg)
63 {
64 	const struct got_error *err = NULL;
65 	const size_t alloc_chunksz = 256;
66 	size_t nalloc;
67 	struct got_reflist_entry *re;
68 	int i;
69 
70 	*ids = NULL;
71 	*nobjects = 0;
72 
73 	err = got_reflist_sort(refs,
74 	    got_ref_cmp_by_commit_timestamp_descending, repo);
75 	if (err)
76 		return err;
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,
144     got_pack_progress_cb progress_cb, void *progress_arg,
145     got_cancel_cb cancel_cb, void *cancel_arg)
146 {
147 	const struct got_error *err = NULL;
148 	struct got_object_id **ours = NULL, **theirs = NULL;
149 	int nours = 0, ntheirs = 0, packfd = -1, i;
150 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
151 	char *sha1_str = NULL;
152 
153 	*packfile = NULL;
154 	*pack_hash = NULL;
155 
156 	if (asprintf(&path, "%s/%s/packing.pack",
157 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
158 		err = got_error_from_errno("asprintf");
159 		goto done;
160 	}
161 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
162 	if (err)
163 		goto done;
164 
165 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
166 		err = got_error_from_errno2("fchmod", tmpfile_path);
167 		goto done;
168 	}
169 
170 	*packfile = fdopen(packfd, "w");
171 	if (*packfile == NULL) {
172 		err = got_error_from_errno2("fdopen", tmpfile_path);
173 		goto done;
174 	}
175 	packfd = -1;
176 
177 	err = get_reflist_object_ids(&ours, &nours,
178 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
179 	    include_refs, repo, cancel_cb, cancel_arg);
180 	if (err)
181 		goto done;
182 
183 	if (nours == 0) {
184 		err = got_error(GOT_ERR_CANNOT_PACK);
185 		goto done;
186 	}
187 
188 	if (!TAILQ_EMPTY(exclude_refs)) {
189 		err = get_reflist_object_ids(&theirs, &ntheirs,
190 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
191 		    exclude_refs, repo,
192 		    cancel_cb, cancel_arg);
193 		if (err)
194 			goto done;
195 	}
196 
197 	*pack_hash = calloc(1, sizeof(**pack_hash));
198 	if (*pack_hash == NULL) {
199 		err = got_error_from_errno("calloc");
200 		goto done;
201 	}
202 
203 	err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
204 	    ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
205 	    cancel_cb, cancel_arg);
206 	if (err)
207 		goto done;
208 
209 	err = got_object_id_str(&sha1_str, *pack_hash);
210 	if (err)
211 		goto done;
212 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
213 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
214 	    sha1_str) == -1) {
215 		err = got_error_from_errno("asprintf");
216 		goto done;
217 	}
218 
219 	if (fflush(*packfile) == -1) {
220 		err = got_error_from_errno("fflush");
221 		goto done;
222 	}
223 	if (fseek(*packfile, 0L, SEEK_SET) == -1) {
224 		err = got_error_from_errno("fseek");
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 done:
235 	for (i = 0; i < nours; i++)
236 		free(ours[i]);
237 	free(ours);
238 	for (i = 0; i < ntheirs; i++)
239 		free(theirs[i]);
240 	free(theirs);
241 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
242 		err = got_error_from_errno2("close", packfile_path);
243 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
244 		err = got_error_from_errno2("unlink", tmpfile_path);
245 	free(tmpfile_path);
246 	free(packfile_path);
247 	free(sha1_str);
248 	free(path);
249 	if (err) {
250 		free(*pack_hash);
251 		*pack_hash = NULL;
252 		if (*packfile)
253 			fclose(*packfile);
254 		*packfile = NULL;
255 	}
256 	return err;
257 }
258 
259 const struct got_error *
260 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
261     struct got_repository *repo,
262     got_pack_index_progress_cb progress_cb, void *progress_arg,
263     got_cancel_cb cancel_cb, void *cancel_arg)
264 {
265 	size_t i;
266 	char *path;
267 	int imsg_idxfds[2];
268 	int npackfd = -1, idxfd = -1, nidxfd = -1;
269 	int tmpfds[3];
270 	int idxstatus, done = 0;
271 	const struct got_error *err;
272 	struct imsgbuf idxibuf;
273 	pid_t idxpid;
274 	char *tmpidxpath = NULL;
275 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
276 	const char *repo_path = got_repo_get_path_git_dir(repo);
277 	struct stat sb;
278 
279 	for (i = 0; i < nitems(tmpfds); i++)
280 		tmpfds[i] = -1;
281 
282 	if (asprintf(&path, "%s/%s/indexing.idx",
283 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
284 		err = got_error_from_errno("asprintf");
285 		goto done;
286 	}
287 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
288 	free(path);
289 	if (err)
290 		goto done;
291 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
292 		err = got_error_from_errno2("fchmod", tmpidxpath);
293 		goto done;
294 	}
295 
296 	nidxfd = dup(idxfd);
297 	if (nidxfd == -1) {
298 		err = got_error_from_errno("dup");
299 		goto done;
300 	}
301 
302 	for (i = 0; i < nitems(tmpfds); i++) {
303 		tmpfds[i] = got_opentempfd();
304 		if (tmpfds[i] == -1) {
305 			err = got_error_from_errno("got_opentempfd");
306 			goto done;
307 		}
308 	}
309 
310 	err = got_object_id_str(&id_str, pack_hash);
311 	if (err)
312 		goto done;
313 
314 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
315 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
316 		err = got_error_from_errno("asprintf");
317 		goto done;
318 	}
319 
320 	if (fstat(fileno(packfile), &sb) == -1) {
321 		err = got_error_from_errno2("fstat", packfile_path);
322 		goto done;
323 	}
324 
325 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
326 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
327 		err = got_error_from_errno("asprintf");
328 		goto done;
329 	}
330 
331 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
332 		err = got_error_from_errno("socketpair");
333 		goto done;
334 	}
335 	idxpid = fork();
336 	if (idxpid == -1) {
337 		err= got_error_from_errno("fork");
338 		goto done;
339 	} else if (idxpid == 0)
340 		got_privsep_exec_child(imsg_idxfds,
341 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
342 	if (close(imsg_idxfds[1]) == -1) {
343 		err = got_error_from_errno("close");
344 		goto done;
345 	}
346 	imsg_init(&idxibuf, imsg_idxfds[0]);
347 
348 	npackfd = dup(fileno(packfile));
349 	if (npackfd == -1) {
350 		err = got_error_from_errno("dup");
351 		goto done;
352 	}
353 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
354 	    npackfd);
355 	if (err != NULL)
356 		goto done;
357 	npackfd = -1;
358 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
359 	if (err != NULL)
360 		goto done;
361 	nidxfd = -1;
362 	for (i = 0; i < nitems(tmpfds); i++) {
363 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
364 		if (err != NULL)
365 			goto done;
366 		tmpfds[i] = -1;
367 	}
368 	done = 0;
369 	while (!done) {
370 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
371 
372 		if (cancel_cb) {
373 			err = cancel_cb(cancel_arg);
374 			if (err)
375 				goto done;
376 		}
377 
378 		err = got_privsep_recv_index_progress(&done, &nobj_total,
379 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
380 		    &idxibuf);
381 		if (err != NULL)
382 			goto done;
383 		if (nobj_indexed != 0) {
384 			err = progress_cb(progress_arg, sb.st_size,
385 			    nobj_total, nobj_indexed, nobj_loose,
386 			    nobj_resolved);
387 			if (err)
388 				break;
389 		}
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,
948 			    &qid->id);
949 			if (err)
950 				goto done;
951 			break;
952 		case GOT_OBJ_TYPE_TAG:
953 			err = got_object_open_as_tag(&tag, repo, &qid->id);
954 			if (err)
955 				goto done;
956 			break;
957 		default:
958 			/* should not happen */
959 			err = got_error(GOT_ERR_OBJ_TYPE);
960 			goto done;
961 		}
962 
963 		/* Find a tree object to scan. */
964 		if (commit) {
965 			tree_id = got_object_commit_get_tree_id(commit);
966 		} else if (tag) {
967 			obj_type = got_object_tag_get_object_type(tag);
968 			switch (obj_type) {
969 			case GOT_OBJ_TYPE_COMMIT:
970 				err = got_object_open_as_commit(&commit, repo,
971 				    got_object_tag_get_object_id(tag));
972 				if (err)
973 					goto done;
974 				tree_id = got_object_commit_get_tree_id(commit);
975 				break;
976 			case GOT_OBJ_TYPE_TREE:
977 				tree_id = got_object_tag_get_object_id(tag);
978 				break;
979 			default:
980 				/*
981 				 * Tag points at something other than a
982 				 * commit or tree. Leave this weird tag object
983 				 * and the object it points to on disk.
984 				 */
985 				err = got_object_idset_remove(NULL, loose_ids,
986 				    &qid->id);
987 				if (err && err->code != GOT_ERR_NO_OBJ)
988 					goto done;
989 				err = got_object_idset_remove(NULL, loose_ids,
990 				    got_object_tag_get_object_id(tag));
991 				if (err && err->code != GOT_ERR_NO_OBJ)
992 					goto done;
993 				err = NULL;
994 				break;
995 			}
996 		}
997 
998 		if (tree_id) {
999 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1000 			    repo, npacked, cancel_cb, cancel_arg);
1001 			if (err)
1002 				break;
1003 		}
1004 
1005 		if (commit || tag)
1006 			(*ncommits)++; /* scanned tags are counted as commits */
1007 
1008 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1009 		    nloose, *ncommits, -1);
1010 		if (err)
1011 			break;
1012  
1013 		if (commit) {
1014 			/* Find parent commits to scan. */
1015 			const struct got_object_id_queue *parent_ids;
1016 			parent_ids = got_object_commit_get_parent_ids(commit);
1017 			err = got_object_id_queue_copy(parent_ids, &ids);
1018 			if (err)
1019 				break;
1020 			got_object_commit_close(commit);
1021 			commit = NULL;
1022 		}
1023 		if (tag) {
1024 			got_object_tag_close(tag);
1025 			tag = NULL;
1026 		}
1027 		got_object_qid_free(qid);
1028 		qid = NULL;
1029 	}
1030 done:
1031 	if (qid)
1032 		got_object_qid_free(qid);
1033 	if (commit)
1034 		got_object_commit_close(commit);
1035 	if (tag)
1036 		got_object_tag_close(tag);
1037 	got_object_id_queue_free(&ids);
1038 	return err;
1039 }
1040 
1041 struct purge_loose_object_arg {
1042 	struct got_repository *repo;
1043 	got_cleanup_progress_cb progress_cb;
1044 	void *progress_arg;
1045 	struct got_ratelimit *rl;
1046 	int nloose;
1047 	int ncommits;
1048 	int npurged;
1049 	off_t size_purged;
1050 	int dry_run;
1051 	time_t max_mtime;
1052 	int ignore_mtime;
1053 };
1054 
1055 static const struct got_error *
1056 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1057 {
1058 	struct purge_loose_object_arg *a = arg;
1059 	const struct got_error *err, *unlock_err = NULL;
1060 	char *path = NULL;
1061 	int fd = -1;
1062 	struct stat sb;
1063 	struct got_lockfile *lf = NULL;
1064 
1065 	err = got_object_get_path(&path, id, a->repo);
1066 	if (err)
1067 		return err;
1068 
1069 	err = got_object_open_loose_fd(&fd, id, a->repo);
1070 	if (err)
1071 		goto done;
1072 
1073 	if (fstat(fd, &sb) == -1) {
1074 		err = got_error_from_errno("fstat");
1075 		goto done;
1076 	}
1077 
1078 	/*
1079 	 * Do not delete objects which are younger than our maximum
1080 	 * modification time threshold. This prevents a race where
1081 	 * new objects which are being added to the repository
1082 	 * concurrently would be deleted.
1083 	 */
1084 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1085 		if (!a->dry_run) {
1086 			err = got_lockfile_lock(&lf, path, -1);
1087 			if (err)
1088 				goto done;
1089 			if (unlink(path) == -1) {
1090 				err = got_error_from_errno2("unlink", path);
1091 				goto done;
1092 			}
1093 		}
1094 
1095 		a->npurged++;
1096 		a->size_purged += sb.st_size;
1097 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1098 		    a->rl, a->nloose, a->ncommits, a->npurged);
1099 		if (err)
1100 			goto done;
1101 	}
1102 done:
1103 	if (fd != -1 && close(fd) == -1 && err == NULL)
1104 		err = got_error_from_errno("close");
1105 	free(path);
1106 	if (lf)
1107 		unlock_err = got_lockfile_unlock(lf, -1);
1108 	return err ? err : unlock_err;
1109 }
1110 
1111 const struct got_error *
1112 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1113     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1114     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1115     got_cancel_cb cancel_cb, void *cancel_arg)
1116 {
1117 	const struct got_error *err;
1118 	struct got_object_idset *loose_ids;
1119 	struct got_object_idset *traversed_ids;
1120 	struct got_object_id **referenced_ids;
1121 	int i, nreferenced, nloose, ncommits = 0;
1122 	struct got_reflist_head refs;
1123 	struct got_reflist_entry *re;
1124 	struct purge_loose_object_arg arg;
1125 	time_t max_mtime = 0;
1126 	struct got_ratelimit rl;
1127 
1128 	TAILQ_INIT(&refs);
1129 	got_ratelimit_init(&rl, 0, 500);
1130 
1131 	*size_before = 0;
1132 	*size_after = 0;
1133 	*npacked = 0;
1134 
1135 	err = get_loose_object_ids(&loose_ids, size_before,
1136 	    progress_cb, progress_arg, &rl, repo);
1137 	if (err)
1138 		return err;
1139 	nloose = got_object_idset_num_elements(loose_ids);
1140 	if (nloose == 0) {
1141 		got_object_idset_free(loose_ids);
1142 		if (progress_cb) {
1143 			err = progress_cb(progress_arg, 0, 0, 0);
1144 			if (err)
1145 				return err;
1146 		}
1147 		return NULL;
1148 	}
1149 
1150 	traversed_ids = got_object_idset_alloc();
1151 	if (traversed_ids == NULL) {
1152 		err = got_error_from_errno("got_object_idset_alloc");
1153 		goto done;
1154 	}
1155 
1156 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1157 	if (err)
1158 		goto done;
1159 	if (!ignore_mtime) {
1160 		TAILQ_FOREACH(re, &refs, entry) {
1161 			time_t mtime = got_ref_get_mtime(re->ref);
1162 			if (mtime > max_mtime)
1163 				max_mtime = mtime;
1164 		}
1165 		/*
1166 		 * For safety, keep objects created within 10 minutes
1167 		 * before the youngest reference was created.
1168 		 */
1169 		if (max_mtime >= 600)
1170 			max_mtime -= 600;
1171 	}
1172 
1173 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1174 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1175 	    &refs, repo, cancel_cb, cancel_arg);
1176 	if (err)
1177 		goto done;
1178 
1179 	for (i = 0; i < nreferenced; i++) {
1180 		struct got_object_id *id = referenced_ids[i];
1181 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1182 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1183 		    nloose, cancel_cb, cancel_arg);
1184 		if (err)
1185 			goto done;
1186 	}
1187 
1188 	/* Any remaining loose objects are unreferenced and can be purged. */
1189 	arg.repo = repo;
1190 	arg.progress_arg = progress_arg;
1191 	arg.progress_cb = progress_cb;
1192 	arg.rl = &rl;
1193 	arg.nloose = nloose;
1194 	arg.npurged = 0;
1195 	arg.size_purged = 0;
1196 	arg.ncommits = ncommits;
1197 	arg.dry_run = dry_run;
1198 	arg.max_mtime = max_mtime;
1199 	arg.ignore_mtime = ignore_mtime;
1200 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1201 	if (err)
1202 		goto done;
1203 	*size_after = *size_before - arg.size_purged;
1204 
1205 	/* Produce a final progress report. */
1206 	if (progress_cb) {
1207 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1208 		if (err)
1209 			goto done;
1210 	}
1211 done:
1212 	got_object_idset_free(loose_ids);
1213 	got_object_idset_free(traversed_ids);
1214 	return err;
1215 }
1216 
1217 static const struct got_error *
1218 remove_packidx(int dir_fd, const char *relpath)
1219 {
1220 	const struct got_error *err, *unlock_err;
1221 	struct got_lockfile *lf;
1222 
1223 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1224 	if (err)
1225 		return err;
1226 	if (unlinkat(dir_fd, relpath, 0) == -1)
1227 		err = got_error_from_errno("unlinkat");
1228 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1229 	return err ? err : unlock_err;
1230 }
1231 
1232 const struct got_error *
1233 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1234     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1235     got_cancel_cb cancel_cb, void *cancel_arg)
1236 {
1237 	const struct got_error *err;
1238 	DIR *packdir = NULL;
1239 	struct dirent *dent;
1240 	char *pack_relpath = NULL;
1241 	int packdir_fd;
1242 	struct stat sb;
1243 
1244 	packdir_fd = openat(got_repo_get_fd(repo),
1245 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1246 	if (packdir_fd == -1) {
1247 		if (errno == ENOENT)
1248 			return NULL;
1249 		return got_error_from_errno_fmt("openat: %s/%s",
1250 		    got_repo_get_path_git_dir(repo),
1251 		    GOT_OBJECTS_PACK_DIR);
1252 	}
1253 
1254 	packdir = fdopendir(packdir_fd);
1255 	if (packdir == NULL) {
1256 		err = got_error_from_errno("fdopendir");
1257 		goto done;
1258 	}
1259 
1260 	while ((dent = readdir(packdir)) != NULL) {
1261 		if (cancel_cb) {
1262 			err = cancel_cb(cancel_arg);
1263 			if (err)
1264 				goto done;
1265 		}
1266 
1267 		if (!got_repo_is_packidx_filename(dent->d_name,
1268 		    strlen(dent->d_name)))
1269 			continue;
1270 
1271 		err = got_packidx_get_packfile_path(&pack_relpath,
1272 		    dent->d_name);
1273 		if (err)
1274 			goto done;
1275 
1276 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1277 			free(pack_relpath);
1278 			pack_relpath = NULL;
1279 			continue;
1280 		}
1281 		if (errno != ENOENT) {
1282 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1283 			    got_repo_get_path_git_dir(repo),
1284 			    GOT_OBJECTS_PACK_DIR,
1285 			    pack_relpath);
1286 			goto done;
1287 		}
1288 
1289 		if (!dry_run) {
1290 			err = remove_packidx(packdir_fd, dent->d_name);
1291 			if (err)
1292 				goto done;
1293 		}
1294 		if (progress_cb) {
1295 			char *path;
1296 			if (asprintf(&path, "%s/%s/%s",
1297 			    got_repo_get_path_git_dir(repo),
1298 			    GOT_OBJECTS_PACK_DIR,
1299 			    dent->d_name) == -1) {
1300 				err = got_error_from_errno("asprintf");
1301 				goto done;
1302 			}
1303 			err = progress_cb(progress_arg, path);
1304 			free(path);
1305 			if (err)
1306 				goto done;
1307 		}
1308 		free(pack_relpath);
1309 		pack_relpath = NULL;
1310 	}
1311 done:
1312 	if (packdir && closedir(packdir) != 0 && err == NULL)
1313 		err = got_error_from_errno("closedir");
1314 	free(pack_relpath);
1315 	return err;
1316 }