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_hash.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, int force_refdelta,
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_PACK_MODE) == -1) {
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,
209 	    0, force_refdelta, progress_cb, progress_arg, &rl,
210 	    cancel_cb, cancel_arg);
211 	if (err)
212 		goto done;
213 
214 	err = got_object_id_str(&sha1_str, *pack_hash);
215 	if (err)
216 		goto done;
217 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
218 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
219 	    sha1_str) == -1) {
220 		err = got_error_from_errno("asprintf");
221 		goto done;
222 	}
223 
224 	if (lseek(packfd, 0L, SEEK_SET) == -1) {
225 		err = got_error_from_errno("lseek");
226 		goto done;
227 	}
228 	if (rename(tmpfile_path, packfile_path) == -1) {
229 		err = got_error_from_errno3("rename", tmpfile_path,
230 		    packfile_path);
231 		goto done;
232 	}
233 	free(tmpfile_path);
234 	tmpfile_path = NULL;
235 
236 	*packfile = fdopen(packfd, "w");
237 	if (*packfile == NULL) {
238 		err = got_error_from_errno2("fdopen", tmpfile_path);
239 		goto done;
240 	}
241 	packfd = -1;
242 done:
243 	for (i = 0; i < nours; i++)
244 		free(ours[i]);
245 	free(ours);
246 	for (i = 0; i < ntheirs; i++)
247 		free(theirs[i]);
248 	free(theirs);
249 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
250 		err = got_error_from_errno2("close", packfile_path);
251 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
252 		err = got_error_from_errno("fclose");
253 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
254 		err = got_error_from_errno2("unlink", tmpfile_path);
255 	free(tmpfile_path);
256 	free(packfile_path);
257 	free(sha1_str);
258 	free(path);
259 	if (err) {
260 		free(*pack_hash);
261 		*pack_hash = NULL;
262 		if (*packfile)
263 			fclose(*packfile);
264 		*packfile = NULL;
265 	}
266 	return err;
267 }
268 
269 const struct got_error *
270 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
271     struct got_repository *repo,
272     got_pack_index_progress_cb progress_cb, void *progress_arg,
273     got_cancel_cb cancel_cb, void *cancel_arg)
274 {
275 	size_t i;
276 	char *path;
277 	int imsg_idxfds[2];
278 	int npackfd = -1, idxfd = -1, nidxfd = -1;
279 	int tmpfds[3];
280 	int idxstatus, done = 0;
281 	const struct got_error *err;
282 	struct imsgbuf idxibuf;
283 	pid_t idxpid;
284 	char *tmpidxpath = NULL;
285 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
286 	const char *repo_path = got_repo_get_path_git_dir(repo);
287 	struct stat sb;
288 
289 	for (i = 0; i < nitems(tmpfds); i++)
290 		tmpfds[i] = -1;
291 
292 	if (asprintf(&path, "%s/%s/indexing.idx",
293 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
294 		err = got_error_from_errno("asprintf");
295 		goto done;
296 	}
297 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
298 	free(path);
299 	if (err)
300 		goto done;
301 	if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
302 		err = got_error_from_errno2("fchmod", tmpidxpath);
303 		goto done;
304 	}
305 
306 	nidxfd = dup(idxfd);
307 	if (nidxfd == -1) {
308 		err = got_error_from_errno("dup");
309 		goto done;
310 	}
311 
312 	for (i = 0; i < nitems(tmpfds); i++) {
313 		tmpfds[i] = got_opentempfd();
314 		if (tmpfds[i] == -1) {
315 			err = got_error_from_errno("got_opentempfd");
316 			goto done;
317 		}
318 	}
319 
320 	err = got_object_id_str(&id_str, pack_hash);
321 	if (err)
322 		goto done;
323 
324 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
325 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
326 		err = got_error_from_errno("asprintf");
327 		goto done;
328 	}
329 
330 	if (fstat(fileno(packfile), &sb) == -1) {
331 		err = got_error_from_errno2("fstat", packfile_path);
332 		goto done;
333 	}
334 
335 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
336 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
337 		err = got_error_from_errno("asprintf");
338 		goto done;
339 	}
340 
341 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
342 		err = got_error_from_errno("socketpair");
343 		goto done;
344 	}
345 	idxpid = fork();
346 	if (idxpid == -1) {
347 		err= got_error_from_errno("fork");
348 		goto done;
349 	} else if (idxpid == 0)
350 		got_privsep_exec_child(imsg_idxfds,
351 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
352 	if (close(imsg_idxfds[1]) == -1) {
353 		err = got_error_from_errno("close");
354 		goto done;
355 	}
356 	imsg_init(&idxibuf, imsg_idxfds[0]);
357 
358 	npackfd = dup(fileno(packfile));
359 	if (npackfd == -1) {
360 		err = got_error_from_errno("dup");
361 		goto done;
362 	}
363 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
364 	    npackfd);
365 	if (err != NULL)
366 		goto done;
367 	npackfd = -1;
368 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
369 	if (err != NULL)
370 		goto done;
371 	nidxfd = -1;
372 	for (i = 0; i < nitems(tmpfds); i++) {
373 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
374 		if (err != NULL)
375 			goto done;
376 		tmpfds[i] = -1;
377 	}
378 	done = 0;
379 	while (!done) {
380 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
381 
382 		if (cancel_cb) {
383 			err = cancel_cb(cancel_arg);
384 			if (err)
385 				goto done;
386 		}
387 
388 		err = got_privsep_recv_index_progress(&done, &nobj_total,
389 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
390 		    &idxibuf);
391 		if (err != NULL)
392 			goto done;
393 		if (nobj_indexed != 0) {
394 			err = progress_cb(progress_arg, sb.st_size,
395 			    nobj_total, nobj_indexed, nobj_loose,
396 			    nobj_resolved);
397 			if (err)
398 				break;
399 		}
400 	}
401 	if (close(imsg_idxfds[0]) == -1) {
402 		err = got_error_from_errno("close");
403 		goto done;
404 	}
405 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
406 		err = got_error_from_errno("waitpid");
407 		goto done;
408 	}
409 
410 	if (rename(tmpidxpath, idxpath) == -1) {
411 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
412 		goto done;
413 	}
414 	free(tmpidxpath);
415 	tmpidxpath = NULL;
416 
417 done:
418 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
419 		err = got_error_from_errno2("unlink", tmpidxpath);
420 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
421 		err = got_error_from_errno("close");
422 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
423 		err = got_error_from_errno("close");
424 	for (i = 0; i < nitems(tmpfds); i++) {
425 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
426 			err = got_error_from_errno("close");
427 	}
428 	free(tmpidxpath);
429 	free(idxpath);
430 	free(packfile_path);
431 	return err;
432 }
433 
434 const struct got_error *
435 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
436     struct got_repository *repo, const char *packfile_path)
437 {
438 	const struct got_error *err = NULL;
439 	const char *packdir_path = NULL;
440 	char *packfile_name = NULL, *p, *dot;
441 	struct got_object_id id;
442 	int packfd = -1;
443 
444 	*packfile = NULL;
445 	*pack_hash = NULL;
446 
447 	packdir_path = got_repo_get_path_objects_pack(repo);
448 	if (packdir_path == NULL)
449 		return got_error_from_errno("got_repo_get_path_objects_pack");
450 
451 	if (!got_path_is_child(packfile_path, packdir_path,
452 	    strlen(packdir_path))) {
453 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
454 		goto done;
455 
456 	}
457 
458 	err = got_path_basename(&packfile_name, packfile_path);
459 	if (err)
460 		goto done;
461 	p = packfile_name;
462 
463 	if (strncmp(p, "pack-", 5) != 0) {
464 		err = got_error_fmt(GOT_ERR_BAD_PATH,
465 		    "'%s' is not a valid pack file name",
466 		    packfile_name);
467 		goto done;
468 	}
469 	p += 5;
470 	dot = strchr(p, '.');
471 	if (dot == NULL) {
472 		err = got_error_fmt(GOT_ERR_BAD_PATH,
473 		    "'%s' is not a valid pack file name",
474 		    packfile_name);
475 		goto done;
476 	}
477 	if (strcmp(dot + 1, "pack") != 0) {
478 		err = got_error_fmt(GOT_ERR_BAD_PATH,
479 		    "'%s' is not a valid pack file name",
480 		    packfile_name);
481 		goto done;
482 	}
483 	*dot = '\0';
484 	if (!got_parse_object_id(&id, p, GOT_HASH_SHA1)) {
485 		err = got_error_fmt(GOT_ERR_BAD_PATH,
486 		    "'%s' is not a valid pack file name",
487 		    packfile_name);
488 		goto done;
489 	}
490 
491 	*pack_hash = got_object_id_dup(&id);
492 	if (*pack_hash == NULL) {
493 		err = got_error_from_errno("got_object_id_dup");
494 		goto done;
495 	}
496 
497 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
498 	if (packfd == -1) {
499 		err = got_error_from_errno2("open", packfile_path);
500 		goto done;
501 	}
502 
503 	*packfile = fdopen(packfd, "r");
504 	if (*packfile == NULL) {
505 		err = got_error_from_errno2("fdopen", packfile_path);
506 		goto done;
507 	}
508 	packfd = -1;
509 done:
510 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
511 		err = got_error_from_errno2("close", packfile_path);
512 	free(packfile_name);
513 	if (err) {
514 		free(*pack_hash);
515 		*pack_hash = NULL;
516 	}
517 	return err;
518 }
519 
520 const struct got_error *
521 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
522     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
523     got_cancel_cb cancel_cb, void *cancel_arg)
524 {
525 	const struct got_error *err = NULL;
526 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
527 	struct got_packidx *packidx = NULL;
528 	struct got_pack *pack = NULL;
529 	uint32_t nobj, i;
530 
531 	err = got_object_id_str(&id_str, pack_hash);
532 	if (err)
533 		goto done;
534 
535 	if (asprintf(&packpath, "%s/pack-%s.pack",
536 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
537 		err = got_error_from_errno("asprintf");
538 		goto done;
539 	}
540 	if (asprintf(&idxpath, "%s/pack-%s.idx",
541 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
542 		err = got_error_from_errno("asprintf");
543 		goto done;
544 	}
545 
546 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
547 	if (err)
548 		goto done;
549 
550 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
551 	if (err)
552 		goto done;
553 
554 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
555 	for (i = 0; i < nobj; i++) {
556 		struct got_packidx_object_id *oid;
557 		struct got_object_id id, base_id;
558 		off_t offset, base_offset = 0;
559 		uint8_t type;
560 		uint64_t size;
561 		size_t tslen, len;
562 
563 		if (cancel_cb) {
564 			err = cancel_cb(cancel_arg);
565 			if (err)
566 				break;
567 		}
568 		oid = &packidx->hdr.sorted_ids[i];
569 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
570 
571 		offset = got_packidx_get_object_offset(packidx, i);
572 		if (offset == -1) {
573 			err = got_error(GOT_ERR_BAD_PACKIDX);
574 			goto done;
575 		}
576 
577 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
578 		    pack, offset);
579 		if (err)
580 			goto done;
581 
582 		switch (type) {
583 		case GOT_OBJ_TYPE_OFFSET_DELTA:
584 			err = got_pack_parse_offset_delta(&base_offset, &len,
585 			    pack, offset, tslen);
586 			if (err)
587 				goto done;
588 			break;
589 		case GOT_OBJ_TYPE_REF_DELTA:
590 			err = got_pack_parse_ref_delta(&base_id,
591 			    pack, offset, tslen);
592 			if (err)
593 				goto done;
594 			break;
595 		}
596 		err = (*list_cb)(list_arg, &id, type, offset, size,
597 		    base_offset, &base_id);
598 		if (err)
599 			goto done;
600 	}
601 
602 done:
603 	free(id_str);
604 	free(idxpath);
605 	free(packpath);
606 	if (packidx)
607 		got_packidx_close(packidx);
608 	return err;
609 }
610 
611 static const struct got_error *
612 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
613     void *progress_arg, struct got_ratelimit *rl,
614     int nloose, int ncommits, int npurged)
615 {
616 	const struct got_error *err;
617 	int elapsed;
618 
619 	if (progress_cb == NULL)
620 		return NULL;
621 
622 	err = got_ratelimit_check(&elapsed, rl);
623 	if (err || !elapsed)
624 		return err;
625 
626 	return progress_cb(progress_arg, nloose, ncommits, npurged);
627 }
628 
629 static const struct got_error *
630 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
631     got_cleanup_progress_cb progress_cb, void *progress_arg,
632     struct got_ratelimit *rl, struct got_repository *repo)
633 {
634 	const struct got_error *err = NULL;
635 	char *path_objects = NULL, *path = NULL;
636 	DIR *dir = NULL;
637 	struct got_object *obj = NULL;
638 	struct got_object_id id;
639 	int i, fd = -1;
640 	struct stat sb;
641 
642 	*ondisk_size = 0;
643 	*loose_ids = got_object_idset_alloc();
644 	if (*loose_ids == NULL)
645 		return got_error_from_errno("got_object_idset_alloc");
646 
647 	path_objects = got_repo_get_path_objects(repo);
648 	if (path_objects == NULL) {
649 		err = got_error_from_errno("got_repo_get_path_objects");
650 		goto done;
651 	}
652 
653 	for (i = 0; i <= 0xff; i++) {
654 		struct dirent *dent;
655 
656 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
657 			err = got_error_from_errno("asprintf");
658 			break;
659 		}
660 
661 		dir = opendir(path);
662 		if (dir == NULL) {
663 			if (errno == ENOENT) {
664 				err = NULL;
665 				continue;
666 			}
667 			err = got_error_from_errno2("opendir", path);
668 			break;
669 		}
670 
671 		while ((dent = readdir(dir)) != NULL) {
672 			char *id_str;
673 
674 			if (strcmp(dent->d_name, ".") == 0 ||
675 			    strcmp(dent->d_name, "..") == 0)
676 				continue;
677 
678 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
679 				err = got_error_from_errno("asprintf");
680 				goto done;
681 			}
682 
683 			if (!got_parse_object_id(&id, id_str,
684 			    GOT_HASH_SHA1)) {
685 				free(id_str);
686 				continue;
687 			}
688 			free(id_str);
689 
690 			err = got_object_open_loose_fd(&fd, &id, repo);
691 			if (err)
692 				goto done;
693 			if (fstat(fd, &sb) == -1) {
694 				err = got_error_from_errno("fstat");
695 				goto done;
696 			}
697 			err = got_object_read_header_privsep(&obj, &id, repo,
698 			    fd);
699 			if (err)
700 				goto done;
701 			fd = -1; /* already closed */
702 
703 			switch (obj->type) {
704 			case GOT_OBJ_TYPE_COMMIT:
705 			case GOT_OBJ_TYPE_TREE:
706 			case GOT_OBJ_TYPE_BLOB:
707 			case GOT_OBJ_TYPE_TAG:
708 				break;
709 			default:
710 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
711 				    "%d", obj->type);
712 				goto done;
713 			}
714 			got_object_close(obj);
715 			obj = NULL;
716 			(*ondisk_size) += sb.st_size;
717 			err = got_object_idset_add(*loose_ids, &id, NULL);
718 			if (err)
719 				goto done;
720 			err = report_cleanup_progress(progress_cb,
721 			    progress_arg, rl,
722 			    got_object_idset_num_elements(*loose_ids), -1, -1);
723 			if (err)
724 				goto done;
725 		}
726 
727 		if (closedir(dir) != 0) {
728 			err = got_error_from_errno("closedir");
729 			goto done;
730 		}
731 		dir = NULL;
732 
733 		free(path);
734 		path = NULL;
735 	}
736 done:
737 	if (dir && closedir(dir) != 0 && err == NULL)
738 		err = got_error_from_errno("closedir");
739 	if (fd != -1 && close(fd) == -1 && err == NULL)
740 		err = got_error_from_errno("close");
741 	if (err) {
742 		got_object_idset_free(*loose_ids);
743 		*loose_ids = NULL;
744 	}
745 	if (obj)
746 		got_object_close(obj);
747 	free(path_objects);
748 	free(path);
749 	return err;
750 }
751 
752 static const struct got_error *
753 preserve_loose_object(struct got_object_idset *loose_ids,
754     struct got_object_id *id, struct got_repository *repo, int *npacked)
755 {
756 	const struct got_error *err = NULL;
757 	struct got_object *obj;
758 
759 	if (!got_object_idset_contains(loose_ids, id))
760 		return NULL;
761 
762 	/*
763 	 * Try to open this object from a pack file. This ensures that
764 	 * we do in fact have a valid packed copy of the object. Otherwise
765 	 * we should not delete the loose representation of this object.
766 	 */
767 	err = got_object_open_packed(&obj, id, repo);
768 	if (err == NULL) {
769 		got_object_close(obj);
770 		/*
771 		 * The object is referenced and packed.
772 		 * We can purge the redundantly stored loose object.
773 		 */
774 		(*npacked)++;
775 		return NULL;
776 	} else if (err->code != GOT_ERR_NO_OBJ)
777 		return err;
778 
779 	/*
780 	 * This object is referenced and not packed.
781 	 * Remove it from our purge set.
782 	 */
783 	return got_object_idset_remove(NULL, loose_ids, id);
784 }
785 
786 static const struct got_error *
787 load_tree_entries(struct got_object_id_queue *ids,
788     struct got_object_idset *loose_ids,
789     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
790     const char *dpath, struct got_repository *repo, int *npacked,
791     got_cancel_cb cancel_cb, void *cancel_arg)
792 {
793 	const struct got_error *err;
794 	struct got_tree_object *tree;
795 	char *p = NULL;
796 	int i;
797 
798 	err = got_object_open_as_tree(&tree, repo, tree_id);
799 	if (err)
800 		return err;
801 
802 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
803 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
804 		struct got_object_id *id = got_tree_entry_get_id(e);
805 		mode_t mode = got_tree_entry_get_mode(e);
806 
807 		if (cancel_cb) {
808 			err = (*cancel_cb)(cancel_arg);
809 			if (err)
810 				break;
811 		}
812 
813 		if (got_object_tree_entry_is_symlink(e) ||
814 		    got_object_tree_entry_is_submodule(e) ||
815 		    got_object_idset_contains(traversed_ids, id))
816 			continue;
817 
818 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
819 		    got_tree_entry_get_name(e)) == -1) {
820 			err = got_error_from_errno("asprintf");
821 			break;
822 		}
823 
824 		if (S_ISDIR(mode)) {
825 			struct got_object_qid *qid;
826 			err = got_object_qid_alloc(&qid, id);
827 			if (err)
828 				break;
829 			STAILQ_INSERT_TAIL(ids, qid, entry);
830 		} else if (S_ISREG(mode)) {
831 			/* This blob is referenced. */
832 			err = preserve_loose_object(loose_ids, id, repo,
833 			    npacked);
834 			if (err)
835 				break;
836 			err = got_object_idset_add(traversed_ids, id, NULL);
837 			if (err)
838 				break;
839 		}
840 		free(p);
841 		p = NULL;
842 	}
843 
844 	got_object_tree_close(tree);
845 	free(p);
846 	return err;
847 }
848 
849 static const struct got_error *
850 load_tree(struct got_object_idset *loose_ids,
851     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
852     const char *dpath, struct got_repository *repo, int *npacked,
853     got_cancel_cb cancel_cb, void *cancel_arg)
854 {
855 	const struct got_error *err = NULL;
856 	struct got_object_id_queue tree_ids;
857 	struct got_object_qid *qid;
858 
859 	err = got_object_qid_alloc(&qid, tree_id);
860 	if (err)
861 		return err;
862 
863 	STAILQ_INIT(&tree_ids);
864 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
865 
866 	while (!STAILQ_EMPTY(&tree_ids)) {
867 		if (cancel_cb) {
868 			err = (*cancel_cb)(cancel_arg);
869 			if (err)
870 				break;
871 		}
872 
873 		qid = STAILQ_FIRST(&tree_ids);
874 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
875 
876 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
877 			got_object_qid_free(qid);
878 			continue;
879 		}
880 
881 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
882 		if (err) {
883 			got_object_qid_free(qid);
884 			break;
885 		}
886 
887 		/* This tree is referenced. */
888 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
889 		if (err)
890 			break;
891 
892 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
893 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
894 		got_object_qid_free(qid);
895 		if (err)
896 			break;
897 	}
898 
899 	got_object_id_queue_free(&tree_ids);
900 	return err;
901 }
902 
903 static const struct got_error *
904 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
905     int *npacked, struct got_object_idset *traversed_ids,
906     struct got_object_id *id, struct got_repository *repo,
907     got_cleanup_progress_cb progress_cb, void *progress_arg,
908     struct got_ratelimit *rl, int nloose,
909     got_cancel_cb cancel_cb, void *cancel_arg)
910 {
911 	const struct got_error *err;
912 	struct got_commit_object *commit = NULL;
913 	struct got_tag_object *tag = NULL;
914 	struct got_object_id *tree_id = NULL;
915 	struct got_object_id_queue ids;
916 	struct got_object_qid *qid;
917 	int obj_type;
918 
919 	err = got_object_qid_alloc(&qid, id);
920 	if (err)
921 		return err;
922 
923 	STAILQ_INIT(&ids);
924 	STAILQ_INSERT_TAIL(&ids, qid, entry);
925 
926 	while (!STAILQ_EMPTY(&ids)) {
927 		if (cancel_cb) {
928 			err = (*cancel_cb)(cancel_arg);
929 			if (err)
930 				break;
931 		}
932 
933 		qid = STAILQ_FIRST(&ids);
934 		STAILQ_REMOVE_HEAD(&ids, entry);
935 
936 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
937 			got_object_qid_free(qid);
938 			qid = NULL;
939 			continue;
940 		}
941 
942 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
943 		if (err)
944 			break;
945 
946 		/* This commit or tag is referenced. */
947 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
948 		if (err)
949 			break;
950 
951 		err = got_object_get_type(&obj_type, repo, &qid->id);
952 		if (err)
953 			break;
954 		switch (obj_type) {
955 		case GOT_OBJ_TYPE_COMMIT:
956 			err = got_object_open_as_commit(&commit, repo,
957 			    &qid->id);
958 			if (err)
959 				goto done;
960 			break;
961 		case GOT_OBJ_TYPE_TAG:
962 			err = got_object_open_as_tag(&tag, repo, &qid->id);
963 			if (err)
964 				goto done;
965 			break;
966 		default:
967 			/* should not happen */
968 			err = got_error(GOT_ERR_OBJ_TYPE);
969 			goto done;
970 		}
971 
972 		/* Find a tree object to scan. */
973 		if (commit) {
974 			tree_id = got_object_commit_get_tree_id(commit);
975 		} else if (tag) {
976 			obj_type = got_object_tag_get_object_type(tag);
977 			switch (obj_type) {
978 			case GOT_OBJ_TYPE_COMMIT:
979 				err = got_object_open_as_commit(&commit, repo,
980 				    got_object_tag_get_object_id(tag));
981 				if (err)
982 					goto done;
983 				tree_id = got_object_commit_get_tree_id(commit);
984 				break;
985 			case GOT_OBJ_TYPE_TREE:
986 				tree_id = got_object_tag_get_object_id(tag);
987 				break;
988 			default:
989 				/*
990 				 * Tag points at something other than a
991 				 * commit or tree. Leave this weird tag object
992 				 * and the object it points to on disk.
993 				 */
994 				err = got_object_idset_remove(NULL, loose_ids,
995 				    &qid->id);
996 				if (err && err->code != GOT_ERR_NO_OBJ)
997 					goto done;
998 				err = got_object_idset_remove(NULL, loose_ids,
999 				    got_object_tag_get_object_id(tag));
1000 				if (err && err->code != GOT_ERR_NO_OBJ)
1001 					goto done;
1002 				err = NULL;
1003 				break;
1004 			}
1005 		}
1006 
1007 		if (tree_id) {
1008 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1009 			    repo, npacked, cancel_cb, cancel_arg);
1010 			if (err)
1011 				break;
1012 		}
1013 
1014 		if (commit || tag)
1015 			(*ncommits)++; /* scanned tags are counted as commits */
1016 
1017 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1018 		    nloose, *ncommits, -1);
1019 		if (err)
1020 			break;
1021 
1022 		if (commit) {
1023 			/* Find parent commits to scan. */
1024 			const struct got_object_id_queue *parent_ids;
1025 			parent_ids = got_object_commit_get_parent_ids(commit);
1026 			err = got_object_id_queue_copy(parent_ids, &ids);
1027 			if (err)
1028 				break;
1029 			got_object_commit_close(commit);
1030 			commit = NULL;
1031 		}
1032 		if (tag) {
1033 			got_object_tag_close(tag);
1034 			tag = NULL;
1035 		}
1036 		got_object_qid_free(qid);
1037 		qid = NULL;
1038 	}
1039 done:
1040 	if (qid)
1041 		got_object_qid_free(qid);
1042 	if (commit)
1043 		got_object_commit_close(commit);
1044 	if (tag)
1045 		got_object_tag_close(tag);
1046 	got_object_id_queue_free(&ids);
1047 	return err;
1048 }
1049 
1050 struct purge_loose_object_arg {
1051 	struct got_repository *repo;
1052 	got_cleanup_progress_cb progress_cb;
1053 	void *progress_arg;
1054 	struct got_ratelimit *rl;
1055 	int nloose;
1056 	int ncommits;
1057 	int npurged;
1058 	off_t size_purged;
1059 	int dry_run;
1060 	time_t max_mtime;
1061 	int ignore_mtime;
1062 };
1063 
1064 static const struct got_error *
1065 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1066 {
1067 	struct purge_loose_object_arg *a = arg;
1068 	const struct got_error *err, *unlock_err = NULL;
1069 	char *path = NULL;
1070 	int fd = -1;
1071 	struct stat sb;
1072 	struct got_lockfile *lf = NULL;
1073 
1074 	err = got_object_get_path(&path, id, a->repo);
1075 	if (err)
1076 		return err;
1077 
1078 	err = got_object_open_loose_fd(&fd, id, a->repo);
1079 	if (err)
1080 		goto done;
1081 
1082 	if (fstat(fd, &sb) == -1) {
1083 		err = got_error_from_errno("fstat");
1084 		goto done;
1085 	}
1086 
1087 	/*
1088 	 * Do not delete objects which are younger than our maximum
1089 	 * modification time threshold. This prevents a race where
1090 	 * new objects which are being added to the repository
1091 	 * concurrently would be deleted.
1092 	 */
1093 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1094 		if (!a->dry_run) {
1095 			err = got_lockfile_lock(&lf, path, -1);
1096 			if (err)
1097 				goto done;
1098 			if (unlink(path) == -1) {
1099 				err = got_error_from_errno2("unlink", path);
1100 				goto done;
1101 			}
1102 		}
1103 
1104 		a->npurged++;
1105 		a->size_purged += sb.st_size;
1106 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1107 		    a->rl, a->nloose, a->ncommits, a->npurged);
1108 		if (err)
1109 			goto done;
1110 	}
1111 done:
1112 	if (fd != -1 && close(fd) == -1 && err == NULL)
1113 		err = got_error_from_errno("close");
1114 	free(path);
1115 	if (lf)
1116 		unlock_err = got_lockfile_unlock(lf, -1);
1117 	return err ? err : unlock_err;
1118 }
1119 
1120 const struct got_error *
1121 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1122     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1123     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1124     got_cancel_cb cancel_cb, void *cancel_arg)
1125 {
1126 	const struct got_error *err;
1127 	struct got_object_idset *loose_ids;
1128 	struct got_object_idset *traversed_ids;
1129 	struct got_object_id **referenced_ids;
1130 	int i, nreferenced, nloose, ncommits = 0;
1131 	struct got_reflist_head refs;
1132 	struct got_reflist_entry *re;
1133 	struct purge_loose_object_arg arg;
1134 	time_t max_mtime = 0;
1135 	struct got_ratelimit rl;
1136 
1137 	TAILQ_INIT(&refs);
1138 	got_ratelimit_init(&rl, 0, 500);
1139 
1140 	*size_before = 0;
1141 	*size_after = 0;
1142 	*npacked = 0;
1143 
1144 	err = get_loose_object_ids(&loose_ids, size_before,
1145 	    progress_cb, progress_arg, &rl, repo);
1146 	if (err)
1147 		return err;
1148 	nloose = got_object_idset_num_elements(loose_ids);
1149 	if (nloose == 0) {
1150 		got_object_idset_free(loose_ids);
1151 		if (progress_cb) {
1152 			err = progress_cb(progress_arg, 0, 0, 0);
1153 			if (err)
1154 				return err;
1155 		}
1156 		return NULL;
1157 	}
1158 
1159 	traversed_ids = got_object_idset_alloc();
1160 	if (traversed_ids == NULL) {
1161 		err = got_error_from_errno("got_object_idset_alloc");
1162 		goto done;
1163 	}
1164 
1165 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1166 	if (err)
1167 		goto done;
1168 	if (!ignore_mtime) {
1169 		TAILQ_FOREACH(re, &refs, entry) {
1170 			time_t mtime = got_ref_get_mtime(re->ref);
1171 			if (mtime > max_mtime)
1172 				max_mtime = mtime;
1173 		}
1174 		/*
1175 		 * For safety, keep objects created within 10 minutes
1176 		 * before the youngest reference was created.
1177 		 */
1178 		if (max_mtime >= 600)
1179 			max_mtime -= 600;
1180 	}
1181 
1182 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1183 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1184 	    &refs, repo, cancel_cb, cancel_arg);
1185 	if (err)
1186 		goto done;
1187 
1188 	for (i = 0; i < nreferenced; i++) {
1189 		struct got_object_id *id = referenced_ids[i];
1190 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1191 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1192 		    nloose, cancel_cb, cancel_arg);
1193 		if (err)
1194 			goto done;
1195 	}
1196 
1197 	/* Any remaining loose objects are unreferenced and can be purged. */
1198 	arg.repo = repo;
1199 	arg.progress_arg = progress_arg;
1200 	arg.progress_cb = progress_cb;
1201 	arg.rl = &rl;
1202 	arg.nloose = nloose;
1203 	arg.npurged = 0;
1204 	arg.size_purged = 0;
1205 	arg.ncommits = ncommits;
1206 	arg.dry_run = dry_run;
1207 	arg.max_mtime = max_mtime;
1208 	arg.ignore_mtime = ignore_mtime;
1209 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1210 	if (err)
1211 		goto done;
1212 	*size_after = *size_before - arg.size_purged;
1213 
1214 	/* Produce a final progress report. */
1215 	if (progress_cb) {
1216 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1217 		if (err)
1218 			goto done;
1219 	}
1220 done:
1221 	got_object_idset_free(loose_ids);
1222 	got_object_idset_free(traversed_ids);
1223 	return err;
1224 }
1225 
1226 static const struct got_error *
1227 remove_packidx(int dir_fd, const char *relpath)
1228 {
1229 	const struct got_error *err, *unlock_err;
1230 	struct got_lockfile *lf;
1231 
1232 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1233 	if (err)
1234 		return err;
1235 	if (unlinkat(dir_fd, relpath, 0) == -1)
1236 		err = got_error_from_errno("unlinkat");
1237 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1238 	return err ? err : unlock_err;
1239 }
1240 
1241 const struct got_error *
1242 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1243     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1244     got_cancel_cb cancel_cb, void *cancel_arg)
1245 {
1246 	const struct got_error *err = NULL;
1247 	DIR *packdir = NULL;
1248 	struct dirent *dent;
1249 	char *pack_relpath = NULL;
1250 	int packdir_fd;
1251 	struct stat sb;
1252 
1253 	packdir_fd = openat(got_repo_get_fd(repo),
1254 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1255 	if (packdir_fd == -1) {
1256 		if (errno == ENOENT)
1257 			return NULL;
1258 		return got_error_from_errno_fmt("openat: %s/%s",
1259 		    got_repo_get_path_git_dir(repo),
1260 		    GOT_OBJECTS_PACK_DIR);
1261 	}
1262 
1263 	packdir = fdopendir(packdir_fd);
1264 	if (packdir == NULL) {
1265 		err = got_error_from_errno("fdopendir");
1266 		goto done;
1267 	}
1268 
1269 	while ((dent = readdir(packdir)) != NULL) {
1270 		if (cancel_cb) {
1271 			err = cancel_cb(cancel_arg);
1272 			if (err)
1273 				goto done;
1274 		}
1275 
1276 		if (!got_repo_is_packidx_filename(dent->d_name,
1277 		    strlen(dent->d_name)))
1278 			continue;
1279 
1280 		err = got_packidx_get_packfile_path(&pack_relpath,
1281 		    dent->d_name);
1282 		if (err)
1283 			goto done;
1284 
1285 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1286 			free(pack_relpath);
1287 			pack_relpath = NULL;
1288 			continue;
1289 		}
1290 		if (errno != ENOENT) {
1291 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1292 			    got_repo_get_path_git_dir(repo),
1293 			    GOT_OBJECTS_PACK_DIR,
1294 			    pack_relpath);
1295 			goto done;
1296 		}
1297 
1298 		if (!dry_run) {
1299 			err = remove_packidx(packdir_fd, dent->d_name);
1300 			if (err)
1301 				goto done;
1302 		}
1303 		if (progress_cb) {
1304 			char *path;
1305 			if (asprintf(&path, "%s/%s/%s",
1306 			    got_repo_get_path_git_dir(repo),
1307 			    GOT_OBJECTS_PACK_DIR,
1308 			    dent->d_name) == -1) {
1309 				err = got_error_from_errno("asprintf");
1310 				goto done;
1311 			}
1312 			err = progress_cb(progress_arg, path);
1313 			free(path);
1314 			if (err)
1315 				goto done;
1316 		}
1317 		free(pack_relpath);
1318 		pack_relpath = NULL;
1319 	}
1320 done:
1321 	if (packdir && closedir(packdir) != 0 && err == NULL)
1322 		err = got_error_from_errno("closedir");
1323 	free(pack_relpath);
1324 	return err;
1325 }