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 <sha2.h>
32 #include <stdio.h>
33 #include <stdlib.h>
34 #include <string.h>
35 #include <limits.h>
36 #include <unistd.h>
37 #include <imsg.h>
38 
39 #include "got_error.h"
40 #include "got_cancel.h"
41 #include "got_object.h"
42 #include "got_reference.h"
43 #include "got_repository.h"
44 #include "got_repository_admin.h"
45 #include "got_opentemp.h"
46 #include "got_path.h"
47 
48 #include "got_lib_delta.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_idset.h"
51 #include "got_lib_object_cache.h"
52 #include "got_lib_pack.h"
53 #include "got_lib_privsep.h"
54 #include "got_lib_repository.h"
55 #include "got_lib_ratelimit.h"
56 #include "got_lib_pack_create.h"
57 #include "got_lib_hash.h"
58 #include "got_lib_lockfile.h"
59 
60 #ifndef nitems
61 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
62 #endif
63 
64 static const struct got_error *
65 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
66     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
67     struct got_repository *repo,
68     got_cancel_cb cancel_cb, void *cancel_arg)
69 {
70 	const struct got_error *err = NULL;
71 	const size_t alloc_chunksz = 256;
72 	size_t nalloc;
73 	struct got_reflist_entry *re;
74 	int i;
75 
76 	*ids = NULL;
77 	*nobjects = 0;
78 
79 	err = got_reflist_sort(refs,
80 	    got_ref_cmp_by_commit_timestamp_descending, repo);
81 	if (err)
82 		return err;
83 
84 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
85 	if (*ids == NULL)
86 		return got_error_from_errno("reallocarray");
87 	nalloc = alloc_chunksz;
88 
89 	TAILQ_FOREACH(re, refs, entry) {
90 		struct got_object_id *id;
91 
92 		if (cancel_cb) {
93 			err = cancel_cb(cancel_arg);
94 			if (err)
95 				goto done;
96 		}
97 
98 		err = got_ref_resolve(&id, repo, re->ref);
99 		if (err)
100 			goto done;
101 
102 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
103 			int obj_type;
104 			err = got_object_get_type(&obj_type, repo, id);
105 			if (err)
106 				goto done;
107 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
108 				free(id);
109 				id = NULL;
110 				continue;
111 			}
112 		}
113 
114 		if (nalloc <= *nobjects) {
115 			struct got_object_id **new;
116 			new = recallocarray(*ids, nalloc,
117 			    nalloc + alloc_chunksz,
118 			    sizeof(struct got_object_id *));
119 			if (new == NULL) {
120 				err = got_error_from_errno(
121 				    "recallocarray");
122 				goto done;
123 			}
124 			*ids = new;
125 			nalloc += alloc_chunksz;
126 		}
127 		(*ids)[*nobjects] = id;
128 		if ((*ids)[*nobjects] == NULL) {
129 			err = got_error_from_errno("got_object_id_dup");
130 			goto done;
131 		}
132 		(*nobjects)++;
133 	}
134 done:
135 	if (err) {
136 		for (i = 0; i < *nobjects; i++)
137 			free((*ids)[i]);
138 		free(*ids);
139 		*ids = NULL;
140 		*nobjects = 0;
141 	}
142 	return err;
143 }
144 
145 const struct got_error *
146 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
147     struct got_reflist_head *include_refs,
148     struct got_reflist_head *exclude_refs, struct got_repository *repo,
149     int loose_obj_only,
150     got_pack_progress_cb progress_cb, void *progress_arg,
151     got_cancel_cb cancel_cb, void *cancel_arg)
152 {
153 	const struct got_error *err = NULL;
154 	struct got_object_id **ours = NULL, **theirs = NULL;
155 	int nours = 0, ntheirs = 0, packfd = -1, i;
156 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
157 	char *hash_str = NULL;
158 	FILE *delta_cache = NULL;
159 	struct got_ratelimit rl;
160 
161 	*packfile = NULL;
162 	*pack_hash = NULL;
163 
164 	got_ratelimit_init(&rl, 0, 500);
165 
166 	if (asprintf(&path, "%s/%s/packing.pack",
167 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
168 		err = got_error_from_errno("asprintf");
169 		goto done;
170 	}
171 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path, "");
172 	if (err)
173 		goto done;
174 
175 	if (fchmod(packfd, GOT_DEFAULT_PACK_MODE) == -1) {
176 		err = got_error_from_errno2("fchmod", tmpfile_path);
177 		goto done;
178 	}
179 
180 	delta_cache = got_opentemp();
181 	if (delta_cache == NULL) {
182 		err = got_error_from_errno("got_opentemp");
183 		goto done;
184 	}
185 
186 	err = get_reflist_object_ids(&ours, &nours,
187 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
188 	    include_refs, repo, cancel_cb, cancel_arg);
189 	if (err)
190 		goto done;
191 
192 	if (nours == 0) {
193 		err = got_error(GOT_ERR_CANNOT_PACK);
194 		goto done;
195 	}
196 
197 	if (!TAILQ_EMPTY(exclude_refs)) {
198 		err = get_reflist_object_ids(&theirs, &ntheirs,
199 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
200 		    exclude_refs, repo,
201 		    cancel_cb, cancel_arg);
202 		if (err)
203 			goto done;
204 	}
205 
206 	*pack_hash = calloc(1, sizeof(**pack_hash));
207 	if (*pack_hash == NULL) {
208 		err = got_error_from_errno("calloc");
209 		goto done;
210 	}
211 
212 	err = got_pack_create((*pack_hash)->hash, packfd, delta_cache,
213 	    theirs, ntheirs, ours, nours, repo, loose_obj_only, 0,
214 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
215 	if (err)
216 		goto done;
217 
218 	err = got_object_id_str(&hash_str, *pack_hash);
219 	if (err)
220 		goto done;
221 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
222 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
223 	    hash_str) == -1) {
224 		err = got_error_from_errno("asprintf");
225 		goto done;
226 	}
227 
228 	if (lseek(packfd, 0L, SEEK_SET) == -1) {
229 		err = got_error_from_errno("lseek");
230 		goto done;
231 	}
232 	if (rename(tmpfile_path, packfile_path) == -1) {
233 		err = got_error_from_errno3("rename", tmpfile_path,
234 		    packfile_path);
235 		goto done;
236 	}
237 	free(tmpfile_path);
238 	tmpfile_path = NULL;
239 
240 	*packfile = fdopen(packfd, "w");
241 	if (*packfile == NULL) {
242 		err = got_error_from_errno2("fdopen", tmpfile_path);
243 		goto done;
244 	}
245 	packfd = -1;
246 done:
247 	for (i = 0; i < nours; i++)
248 		free(ours[i]);
249 	free(ours);
250 	for (i = 0; i < ntheirs; i++)
251 		free(theirs[i]);
252 	free(theirs);
253 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
254 		err = got_error_from_errno2("close", packfile_path);
255 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
256 		err = got_error_from_errno("fclose");
257 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
258 		err = got_error_from_errno2("unlink", tmpfile_path);
259 	free(tmpfile_path);
260 	free(packfile_path);
261 	free(hash_str);
262 	free(path);
263 	if (err) {
264 		free(*pack_hash);
265 		*pack_hash = NULL;
266 		if (*packfile)
267 			fclose(*packfile);
268 		*packfile = NULL;
269 	}
270 	return err;
271 }
272 
273 const struct got_error *
274 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
275     struct got_repository *repo,
276     got_pack_index_progress_cb progress_cb, void *progress_arg,
277     got_cancel_cb cancel_cb, void *cancel_arg)
278 {
279 	size_t i;
280 	char *path;
281 	int imsg_idxfds[2];
282 	int npackfd = -1, idxfd = -1, nidxfd = -1;
283 	int tmpfds[3];
284 	int idxstatus, done = 0;
285 	const struct got_error *err;
286 	struct imsgbuf idxibuf;
287 	pid_t idxpid;
288 	char *tmpidxpath = NULL;
289 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
290 	const char *repo_path = got_repo_get_path_git_dir(repo);
291 	struct stat sb;
292 
293 	for (i = 0; i < nitems(tmpfds); i++)
294 		tmpfds[i] = -1;
295 
296 	if (asprintf(&path, "%s/%s/indexing.idx",
297 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
298 		err = got_error_from_errno("asprintf");
299 		goto done;
300 	}
301 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
302 	free(path);
303 	if (err)
304 		goto done;
305 	if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
306 		err = got_error_from_errno2("fchmod", tmpidxpath);
307 		goto done;
308 	}
309 
310 	nidxfd = dup(idxfd);
311 	if (nidxfd == -1) {
312 		err = got_error_from_errno("dup");
313 		goto done;
314 	}
315 
316 	for (i = 0; i < nitems(tmpfds); i++) {
317 		tmpfds[i] = got_opentempfd();
318 		if (tmpfds[i] == -1) {
319 			err = got_error_from_errno("got_opentempfd");
320 			goto done;
321 		}
322 	}
323 
324 	err = got_object_id_str(&id_str, pack_hash);
325 	if (err)
326 		goto done;
327 
328 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
329 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
330 		err = got_error_from_errno("asprintf");
331 		goto done;
332 	}
333 
334 	if (fstat(fileno(packfile), &sb) == -1) {
335 		err = got_error_from_errno2("fstat", packfile_path);
336 		goto done;
337 	}
338 
339 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
340 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
341 		err = got_error_from_errno("asprintf");
342 		goto done;
343 	}
344 
345 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
346 		err = got_error_from_errno("socketpair");
347 		goto done;
348 	}
349 	idxpid = fork();
350 	if (idxpid == -1) {
351 		err= got_error_from_errno("fork");
352 		goto done;
353 	} else if (idxpid == 0)
354 		got_privsep_exec_child(imsg_idxfds,
355 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
356 	if (close(imsg_idxfds[1]) == -1) {
357 		err = got_error_from_errno("close");
358 		goto done;
359 	}
360 	imsg_init(&idxibuf, imsg_idxfds[0]);
361 
362 	npackfd = dup(fileno(packfile));
363 	if (npackfd == -1) {
364 		err = got_error_from_errno("dup");
365 		goto done;
366 	}
367 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->hash,
368 	    npackfd);
369 	if (err != NULL)
370 		goto done;
371 	npackfd = -1;
372 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
373 	if (err != NULL)
374 		goto done;
375 	nidxfd = -1;
376 	for (i = 0; i < nitems(tmpfds); i++) {
377 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
378 		if (err != NULL)
379 			goto done;
380 		tmpfds[i] = -1;
381 	}
382 	done = 0;
383 	while (!done) {
384 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
385 
386 		if (cancel_cb) {
387 			err = cancel_cb(cancel_arg);
388 			if (err)
389 				goto done;
390 		}
391 
392 		err = got_privsep_recv_index_progress(&done, &nobj_total,
393 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
394 		    &idxibuf);
395 		if (err != NULL)
396 			goto done;
397 		if (nobj_indexed != 0) {
398 			err = progress_cb(progress_arg, sb.st_size,
399 			    nobj_total, nobj_indexed, nobj_loose,
400 			    nobj_resolved);
401 			if (err)
402 				break;
403 		}
404 	}
405 	if (close(imsg_idxfds[0]) == -1) {
406 		err = got_error_from_errno("close");
407 		goto done;
408 	}
409 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
410 		err = got_error_from_errno("waitpid");
411 		goto done;
412 	}
413 
414 	if (rename(tmpidxpath, idxpath) == -1) {
415 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
416 		goto done;
417 	}
418 	free(tmpidxpath);
419 	tmpidxpath = NULL;
420 
421 done:
422 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
423 		err = got_error_from_errno2("unlink", tmpidxpath);
424 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
425 		err = got_error_from_errno("close");
426 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
427 		err = got_error_from_errno("close");
428 	for (i = 0; i < nitems(tmpfds); i++) {
429 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
430 			err = got_error_from_errno("close");
431 	}
432 	free(tmpidxpath);
433 	free(idxpath);
434 	free(packfile_path);
435 	return err;
436 }
437 
438 const struct got_error *
439 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
440     struct got_repository *repo, const char *packfile_path)
441 {
442 	const struct got_error *err = NULL;
443 	const char *packdir_path = NULL;
444 	char *packfile_name = NULL, *p, *dot;
445 	struct got_object_id id;
446 	int packfd = -1;
447 
448 	*packfile = NULL;
449 	*pack_hash = NULL;
450 
451 	packdir_path = got_repo_get_path_objects_pack(repo);
452 	if (packdir_path == NULL)
453 		return got_error_from_errno("got_repo_get_path_objects_pack");
454 
455 	if (!got_path_is_child(packfile_path, packdir_path,
456 	    strlen(packdir_path))) {
457 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
458 		goto done;
459 
460 	}
461 
462 	err = got_path_basename(&packfile_name, packfile_path);
463 	if (err)
464 		goto done;
465 	p = packfile_name;
466 
467 	if (strncmp(p, "pack-", 5) != 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 	p += 5;
474 	dot = strchr(p, '.');
475 	if (dot == NULL) {
476 		err = got_error_fmt(GOT_ERR_BAD_PATH,
477 		    "'%s' is not a valid pack file name",
478 		    packfile_name);
479 		goto done;
480 	}
481 	if (strcmp(dot + 1, "pack") != 0) {
482 		err = got_error_fmt(GOT_ERR_BAD_PATH,
483 		    "'%s' is not a valid pack file name",
484 		    packfile_name);
485 		goto done;
486 	}
487 	*dot = '\0';
488 	if (!got_parse_sha1_digest(id.hash, p)) {
489 		err = got_error_fmt(GOT_ERR_BAD_PATH,
490 		    "'%s' is not a valid pack file name",
491 		    packfile_name);
492 		goto done;
493 	}
494 
495 	*pack_hash = got_object_id_dup(&id);
496 	if (*pack_hash == NULL) {
497 		err = got_error_from_errno("got_object_id_dup");
498 		goto done;
499 	}
500 
501 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
502 	if (packfd == -1) {
503 		err = got_error_from_errno2("open", packfile_path);
504 		goto done;
505 	}
506 
507 	*packfile = fdopen(packfd, "r");
508 	if (*packfile == NULL) {
509 		err = got_error_from_errno2("fdopen", packfile_path);
510 		goto done;
511 	}
512 	packfd = -1;
513 done:
514 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
515 		err = got_error_from_errno2("close", packfile_path);
516 	free(packfile_name);
517 	if (err) {
518 		free(*pack_hash);
519 		*pack_hash = NULL;
520 	}
521 	return err;
522 }
523 
524 const struct got_error *
525 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
526     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
527     got_cancel_cb cancel_cb, void *cancel_arg)
528 {
529 	const struct got_error *err = NULL;
530 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
531 	struct got_packidx *packidx = NULL;
532 	struct got_pack *pack = NULL;
533 	uint32_t nobj, i;
534 
535 	err = got_object_id_str(&id_str, pack_hash);
536 	if (err)
537 		goto done;
538 
539 	if (asprintf(&packpath, "%s/pack-%s.pack",
540 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
541 		err = got_error_from_errno("asprintf");
542 		goto done;
543 	}
544 	if (asprintf(&idxpath, "%s/pack-%s.idx",
545 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
546 		err = got_error_from_errno("asprintf");
547 		goto done;
548 	}
549 
550 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1,
551 	    repo->algo);
552 	if (err)
553 		goto done;
554 
555 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
556 	if (err)
557 		goto done;
558 
559 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
560 	for (i = 0; i < nobj; i++) {
561 		struct got_packidx_object_id *oid;
562 		struct got_object_id id, base_id;
563 		off_t offset, base_offset = 0;
564 		uint8_t type;
565 		uint64_t size;
566 		size_t tslen, len;
567 
568 		if (cancel_cb) {
569 			err = cancel_cb(cancel_arg);
570 			if (err)
571 				break;
572 		}
573 		oid = &packidx->hdr.sorted_ids[i];
574 		memcpy(id.hash, oid->hash, SHA1_DIGEST_LENGTH);
575 
576 		offset = got_packidx_get_object_offset(packidx, i);
577 		if (offset == -1) {
578 			err = got_error(GOT_ERR_BAD_PACKIDX);
579 			goto done;
580 		}
581 
582 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
583 		    pack, offset);
584 		if (err)
585 			goto done;
586 
587 		switch (type) {
588 		case GOT_OBJ_TYPE_OFFSET_DELTA:
589 			err = got_pack_parse_offset_delta(&base_offset, &len,
590 			    pack, offset, tslen);
591 			if (err)
592 				goto done;
593 			break;
594 		case GOT_OBJ_TYPE_REF_DELTA:
595 			err = got_pack_parse_ref_delta(&base_id,
596 			    pack, offset, tslen);
597 			if (err)
598 				goto done;
599 			break;
600 		}
601 		err = (*list_cb)(list_arg, &id, type, offset, size,
602 		    base_offset, &base_id);
603 		if (err)
604 			goto done;
605 	}
606 
607 done:
608 	free(id_str);
609 	free(idxpath);
610 	free(packpath);
611 	if (packidx)
612 		got_packidx_close(packidx);
613 	return err;
614 }
615 
616 static const struct got_error *
617 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
618     void *progress_arg, struct got_ratelimit *rl,
619     int nloose, int ncommits, int npurged)
620 {
621 	const struct got_error *err;
622 	int elapsed;
623 
624 	if (progress_cb == NULL)
625 		return NULL;
626 
627 	err = got_ratelimit_check(&elapsed, rl);
628 	if (err || !elapsed)
629 		return err;
630 
631 	return progress_cb(progress_arg, nloose, ncommits, npurged);
632 }
633 
634 static const struct got_error *
635 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
636     got_cleanup_progress_cb progress_cb, void *progress_arg,
637     struct got_ratelimit *rl, struct got_repository *repo)
638 {
639 	const struct got_error *err = NULL;
640 	char *path_objects = NULL, *path = NULL;
641 	DIR *dir = NULL;
642 	struct got_object *obj = NULL;
643 	struct got_object_id id;
644 	int i, fd = -1;
645 	struct stat sb;
646 
647 	*ondisk_size = 0;
648 	*loose_ids = got_object_idset_alloc();
649 	if (*loose_ids == NULL)
650 		return got_error_from_errno("got_object_idset_alloc");
651 
652 	path_objects = got_repo_get_path_objects(repo);
653 	if (path_objects == NULL) {
654 		err = got_error_from_errno("got_repo_get_path_objects");
655 		goto done;
656 	}
657 
658 	for (i = 0; i <= 0xff; i++) {
659 		struct dirent *dent;
660 
661 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
662 			err = got_error_from_errno("asprintf");
663 			break;
664 		}
665 
666 		dir = opendir(path);
667 		if (dir == NULL) {
668 			if (errno == ENOENT) {
669 				err = NULL;
670 				continue;
671 			}
672 			err = got_error_from_errno2("opendir", path);
673 			break;
674 		}
675 
676 		while ((dent = readdir(dir)) != NULL) {
677 			char *id_str;
678 
679 			if (strcmp(dent->d_name, ".") == 0 ||
680 			    strcmp(dent->d_name, "..") == 0)
681 				continue;
682 
683 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
684 				err = got_error_from_errno("asprintf");
685 				goto done;
686 			}
687 
688 			memset(&id, 0, sizeof(id));
689 			if (!got_parse_sha1_digest(id.hash, id_str)) {
690 				free(id_str);
691 				continue;
692 			}
693 			free(id_str);
694 
695 			err = got_object_open_loose_fd(&fd, &id, repo);
696 			if (err)
697 				goto done;
698 			if (fstat(fd, &sb) == -1) {
699 				err = got_error_from_errno("fstat");
700 				goto done;
701 			}
702 			err = got_object_read_header_privsep(&obj, &id, repo,
703 			    fd);
704 			if (err)
705 				goto done;
706 			fd = -1; /* already closed */
707 
708 			switch (obj->type) {
709 			case GOT_OBJ_TYPE_COMMIT:
710 			case GOT_OBJ_TYPE_TREE:
711 			case GOT_OBJ_TYPE_BLOB:
712 			case GOT_OBJ_TYPE_TAG:
713 				break;
714 			default:
715 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
716 				    "%d", obj->type);
717 				goto done;
718 			}
719 			got_object_close(obj);
720 			obj = NULL;
721 			(*ondisk_size) += sb.st_size;
722 			err = got_object_idset_add(*loose_ids, &id, NULL);
723 			if (err)
724 				goto done;
725 			err = report_cleanup_progress(progress_cb,
726 			    progress_arg, rl,
727 			    got_object_idset_num_elements(*loose_ids), -1, -1);
728 			if (err)
729 				goto done;
730 		}
731 
732 		if (closedir(dir) != 0) {
733 			err = got_error_from_errno("closedir");
734 			goto done;
735 		}
736 		dir = NULL;
737 
738 		free(path);
739 		path = NULL;
740 	}
741 done:
742 	if (dir && closedir(dir) != 0 && err == NULL)
743 		err = got_error_from_errno("closedir");
744 	if (fd != -1 && close(fd) == -1 && err == NULL)
745 		err = got_error_from_errno("close");
746 	if (err) {
747 		got_object_idset_free(*loose_ids);
748 		*loose_ids = NULL;
749 	}
750 	if (obj)
751 		got_object_close(obj);
752 	free(path_objects);
753 	free(path);
754 	return err;
755 }
756 
757 static const struct got_error *
758 preserve_loose_object(struct got_object_idset *loose_ids,
759     struct got_object_id *id, struct got_repository *repo, int *npacked)
760 {
761 	const struct got_error *err = NULL;
762 	struct got_object *obj;
763 
764 	if (!got_object_idset_contains(loose_ids, id))
765 		return NULL;
766 
767 	/*
768 	 * Try to open this object from a pack file. This ensures that
769 	 * we do in fact have a valid packed copy of the object. Otherwise
770 	 * we should not delete the loose representation of this object.
771 	 */
772 	err = got_object_open_packed(&obj, id, repo);
773 	if (err == NULL) {
774 		got_object_close(obj);
775 		/*
776 		 * The object is referenced and packed.
777 		 * We can purge the redundantly stored loose object.
778 		 */
779 		(*npacked)++;
780 		return NULL;
781 	} else if (err->code != GOT_ERR_NO_OBJ)
782 		return err;
783 
784 	/*
785 	 * This object is referenced and not packed.
786 	 * Remove it from our purge set.
787 	 */
788 	return got_object_idset_remove(NULL, loose_ids, id);
789 }
790 
791 static const struct got_error *
792 load_tree_entries(struct got_object_id_queue *ids,
793     struct got_object_idset *loose_ids,
794     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
795     const char *dpath, struct got_repository *repo, int *npacked,
796     got_cancel_cb cancel_cb, void *cancel_arg)
797 {
798 	const struct got_error *err;
799 	struct got_tree_object *tree;
800 	char *p = NULL;
801 	int i;
802 
803 	err = got_object_open_as_tree(&tree, repo, tree_id);
804 	if (err)
805 		return err;
806 
807 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
808 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
809 		struct got_object_id *id = got_tree_entry_get_id(e);
810 		mode_t mode = got_tree_entry_get_mode(e);
811 
812 		if (cancel_cb) {
813 			err = (*cancel_cb)(cancel_arg);
814 			if (err)
815 				break;
816 		}
817 
818 		if (got_object_tree_entry_is_symlink(e) ||
819 		    got_object_tree_entry_is_submodule(e) ||
820 		    got_object_idset_contains(traversed_ids, id))
821 			continue;
822 
823 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
824 		    got_tree_entry_get_name(e)) == -1) {
825 			err = got_error_from_errno("asprintf");
826 			break;
827 		}
828 
829 		if (S_ISDIR(mode)) {
830 			struct got_object_qid *qid;
831 			err = got_object_qid_alloc(&qid, id);
832 			if (err)
833 				break;
834 			STAILQ_INSERT_TAIL(ids, qid, entry);
835 		} else if (S_ISREG(mode)) {
836 			/* This blob is referenced. */
837 			err = preserve_loose_object(loose_ids, id, repo,
838 			    npacked);
839 			if (err)
840 				break;
841 			err = got_object_idset_add(traversed_ids, id, NULL);
842 			if (err)
843 				break;
844 		}
845 		free(p);
846 		p = NULL;
847 	}
848 
849 	got_object_tree_close(tree);
850 	free(p);
851 	return err;
852 }
853 
854 static const struct got_error *
855 load_tree(struct got_object_idset *loose_ids,
856     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
857     const char *dpath, struct got_repository *repo, int *npacked,
858     got_cancel_cb cancel_cb, void *cancel_arg)
859 {
860 	const struct got_error *err = NULL;
861 	struct got_object_id_queue tree_ids;
862 	struct got_object_qid *qid;
863 
864 	err = got_object_qid_alloc(&qid, tree_id);
865 	if (err)
866 		return err;
867 
868 	STAILQ_INIT(&tree_ids);
869 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
870 
871 	while (!STAILQ_EMPTY(&tree_ids)) {
872 		if (cancel_cb) {
873 			err = (*cancel_cb)(cancel_arg);
874 			if (err)
875 				break;
876 		}
877 
878 		qid = STAILQ_FIRST(&tree_ids);
879 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
880 
881 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
882 			got_object_qid_free(qid);
883 			continue;
884 		}
885 
886 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
887 		if (err) {
888 			got_object_qid_free(qid);
889 			break;
890 		}
891 
892 		/* This tree is referenced. */
893 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
894 		if (err)
895 			break;
896 
897 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
898 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
899 		got_object_qid_free(qid);
900 		if (err)
901 			break;
902 	}
903 
904 	got_object_id_queue_free(&tree_ids);
905 	return err;
906 }
907 
908 static const struct got_error *
909 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
910     int *npacked, struct got_object_idset *traversed_ids,
911     struct got_object_id *id, struct got_repository *repo,
912     got_cleanup_progress_cb progress_cb, void *progress_arg,
913     struct got_ratelimit *rl, int nloose,
914     got_cancel_cb cancel_cb, void *cancel_arg)
915 {
916 	const struct got_error *err;
917 	struct got_commit_object *commit = NULL;
918 	struct got_tag_object *tag = NULL;
919 	struct got_object_id *tree_id = NULL;
920 	struct got_object_id_queue ids;
921 	struct got_object_qid *qid;
922 	int obj_type;
923 
924 	err = got_object_qid_alloc(&qid, id);
925 	if (err)
926 		return err;
927 
928 	STAILQ_INIT(&ids);
929 	STAILQ_INSERT_TAIL(&ids, qid, entry);
930 
931 	while (!STAILQ_EMPTY(&ids)) {
932 		if (cancel_cb) {
933 			err = (*cancel_cb)(cancel_arg);
934 			if (err)
935 				break;
936 		}
937 
938 		qid = STAILQ_FIRST(&ids);
939 		STAILQ_REMOVE_HEAD(&ids, entry);
940 
941 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
942 			got_object_qid_free(qid);
943 			qid = NULL;
944 			continue;
945 		}
946 
947 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
948 		if (err)
949 			break;
950 
951 		/* This commit or tag is referenced. */
952 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
953 		if (err)
954 			break;
955 
956 		err = got_object_get_type(&obj_type, repo, &qid->id);
957 		if (err)
958 			break;
959 		switch (obj_type) {
960 		case GOT_OBJ_TYPE_COMMIT:
961 			err = got_object_open_as_commit(&commit, repo,
962 			    &qid->id);
963 			if (err)
964 				goto done;
965 			break;
966 		case GOT_OBJ_TYPE_TAG:
967 			err = got_object_open_as_tag(&tag, repo, &qid->id);
968 			if (err)
969 				goto done;
970 			break;
971 		default:
972 			/* should not happen */
973 			err = got_error(GOT_ERR_OBJ_TYPE);
974 			goto done;
975 		}
976 
977 		/* Find a tree object to scan. */
978 		if (commit) {
979 			tree_id = got_object_commit_get_tree_id(commit);
980 		} else if (tag) {
981 			obj_type = got_object_tag_get_object_type(tag);
982 			switch (obj_type) {
983 			case GOT_OBJ_TYPE_COMMIT:
984 				err = got_object_open_as_commit(&commit, repo,
985 				    got_object_tag_get_object_id(tag));
986 				if (err)
987 					goto done;
988 				tree_id = got_object_commit_get_tree_id(commit);
989 				break;
990 			case GOT_OBJ_TYPE_TREE:
991 				tree_id = got_object_tag_get_object_id(tag);
992 				break;
993 			default:
994 				/*
995 				 * Tag points at something other than a
996 				 * commit or tree. Leave this weird tag object
997 				 * and the object it points to on disk.
998 				 */
999 				err = got_object_idset_remove(NULL, loose_ids,
1000 				    &qid->id);
1001 				if (err && err->code != GOT_ERR_NO_OBJ)
1002 					goto done;
1003 				err = got_object_idset_remove(NULL, loose_ids,
1004 				    got_object_tag_get_object_id(tag));
1005 				if (err && err->code != GOT_ERR_NO_OBJ)
1006 					goto done;
1007 				err = NULL;
1008 				break;
1009 			}
1010 		}
1011 
1012 		if (tree_id) {
1013 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1014 			    repo, npacked, cancel_cb, cancel_arg);
1015 			if (err)
1016 				break;
1017 		}
1018 
1019 		if (commit || tag)
1020 			(*ncommits)++; /* scanned tags are counted as commits */
1021 
1022 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1023 		    nloose, *ncommits, -1);
1024 		if (err)
1025 			break;
1026 
1027 		if (commit) {
1028 			/* Find parent commits to scan. */
1029 			const struct got_object_id_queue *parent_ids;
1030 			parent_ids = got_object_commit_get_parent_ids(commit);
1031 			err = got_object_id_queue_copy(parent_ids, &ids);
1032 			if (err)
1033 				break;
1034 			got_object_commit_close(commit);
1035 			commit = NULL;
1036 		}
1037 		if (tag) {
1038 			got_object_tag_close(tag);
1039 			tag = NULL;
1040 		}
1041 		got_object_qid_free(qid);
1042 		qid = NULL;
1043 	}
1044 done:
1045 	if (qid)
1046 		got_object_qid_free(qid);
1047 	if (commit)
1048 		got_object_commit_close(commit);
1049 	if (tag)
1050 		got_object_tag_close(tag);
1051 	got_object_id_queue_free(&ids);
1052 	return err;
1053 }
1054 
1055 struct purge_loose_object_arg {
1056 	struct got_repository *repo;
1057 	got_cleanup_progress_cb progress_cb;
1058 	void *progress_arg;
1059 	struct got_ratelimit *rl;
1060 	int nloose;
1061 	int ncommits;
1062 	int npurged;
1063 	off_t size_purged;
1064 	int dry_run;
1065 	time_t max_mtime;
1066 	int ignore_mtime;
1067 };
1068 
1069 static const struct got_error *
1070 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1071 {
1072 	struct purge_loose_object_arg *a = arg;
1073 	const struct got_error *err, *unlock_err = NULL;
1074 	char *path = NULL;
1075 	int fd = -1;
1076 	struct stat sb;
1077 	struct got_lockfile *lf = NULL;
1078 
1079 	err = got_object_get_path(&path, id, a->repo);
1080 	if (err)
1081 		return err;
1082 
1083 	err = got_object_open_loose_fd(&fd, id, a->repo);
1084 	if (err)
1085 		goto done;
1086 
1087 	if (fstat(fd, &sb) == -1) {
1088 		err = got_error_from_errno("fstat");
1089 		goto done;
1090 	}
1091 
1092 	/*
1093 	 * Do not delete objects which are younger than our maximum
1094 	 * modification time threshold. This prevents a race where
1095 	 * new objects which are being added to the repository
1096 	 * concurrently would be deleted.
1097 	 */
1098 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1099 		if (!a->dry_run) {
1100 			err = got_lockfile_lock(&lf, path, -1);
1101 			if (err)
1102 				goto done;
1103 			if (unlink(path) == -1) {
1104 				err = got_error_from_errno2("unlink", path);
1105 				goto done;
1106 			}
1107 		}
1108 
1109 		a->npurged++;
1110 		a->size_purged += sb.st_size;
1111 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1112 		    a->rl, a->nloose, a->ncommits, a->npurged);
1113 		if (err)
1114 			goto done;
1115 	}
1116 done:
1117 	if (fd != -1 && close(fd) == -1 && err == NULL)
1118 		err = got_error_from_errno("close");
1119 	free(path);
1120 	if (lf)
1121 		unlock_err = got_lockfile_unlock(lf, -1);
1122 	return err ? err : unlock_err;
1123 }
1124 
1125 const struct got_error *
1126 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1127     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1128     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1129     got_cancel_cb cancel_cb, void *cancel_arg)
1130 {
1131 	const struct got_error *err;
1132 	struct got_object_idset *loose_ids;
1133 	struct got_object_idset *traversed_ids;
1134 	struct got_object_id **referenced_ids;
1135 	int i, nreferenced, nloose, ncommits = 0;
1136 	struct got_reflist_head refs;
1137 	struct got_reflist_entry *re;
1138 	struct purge_loose_object_arg arg;
1139 	time_t max_mtime = 0;
1140 	struct got_ratelimit rl;
1141 
1142 	TAILQ_INIT(&refs);
1143 	got_ratelimit_init(&rl, 0, 500);
1144 
1145 	*size_before = 0;
1146 	*size_after = 0;
1147 	*npacked = 0;
1148 
1149 	err = get_loose_object_ids(&loose_ids, size_before,
1150 	    progress_cb, progress_arg, &rl, repo);
1151 	if (err)
1152 		return err;
1153 	nloose = got_object_idset_num_elements(loose_ids);
1154 	if (nloose == 0) {
1155 		got_object_idset_free(loose_ids);
1156 		if (progress_cb) {
1157 			err = progress_cb(progress_arg, 0, 0, 0);
1158 			if (err)
1159 				return err;
1160 		}
1161 		return NULL;
1162 	}
1163 
1164 	traversed_ids = got_object_idset_alloc();
1165 	if (traversed_ids == NULL) {
1166 		err = got_error_from_errno("got_object_idset_alloc");
1167 		goto done;
1168 	}
1169 
1170 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1171 	if (err)
1172 		goto done;
1173 	if (!ignore_mtime) {
1174 		TAILQ_FOREACH(re, &refs, entry) {
1175 			time_t mtime = got_ref_get_mtime(re->ref);
1176 			if (mtime > max_mtime)
1177 				max_mtime = mtime;
1178 		}
1179 		/*
1180 		 * For safety, keep objects created within 10 minutes
1181 		 * before the youngest reference was created.
1182 		 */
1183 		if (max_mtime >= 600)
1184 			max_mtime -= 600;
1185 	}
1186 
1187 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1188 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1189 	    &refs, repo, cancel_cb, cancel_arg);
1190 	if (err)
1191 		goto done;
1192 
1193 	for (i = 0; i < nreferenced; i++) {
1194 		struct got_object_id *id = referenced_ids[i];
1195 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1196 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1197 		    nloose, cancel_cb, cancel_arg);
1198 		if (err)
1199 			goto done;
1200 	}
1201 
1202 	/* Any remaining loose objects are unreferenced and can be purged. */
1203 	arg.repo = repo;
1204 	arg.progress_arg = progress_arg;
1205 	arg.progress_cb = progress_cb;
1206 	arg.rl = &rl;
1207 	arg.nloose = nloose;
1208 	arg.npurged = 0;
1209 	arg.size_purged = 0;
1210 	arg.ncommits = ncommits;
1211 	arg.dry_run = dry_run;
1212 	arg.max_mtime = max_mtime;
1213 	arg.ignore_mtime = ignore_mtime;
1214 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1215 	if (err)
1216 		goto done;
1217 	*size_after = *size_before - arg.size_purged;
1218 
1219 	/* Produce a final progress report. */
1220 	if (progress_cb) {
1221 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1222 		if (err)
1223 			goto done;
1224 	}
1225 done:
1226 	got_object_idset_free(loose_ids);
1227 	got_object_idset_free(traversed_ids);
1228 	return err;
1229 }
1230 
1231 static const struct got_error *
1232 remove_packidx(int dir_fd, const char *relpath)
1233 {
1234 	const struct got_error *err, *unlock_err;
1235 	struct got_lockfile *lf;
1236 
1237 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1238 	if (err)
1239 		return err;
1240 	if (unlinkat(dir_fd, relpath, 0) == -1)
1241 		err = got_error_from_errno("unlinkat");
1242 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1243 	return err ? err : unlock_err;
1244 }
1245 
1246 const struct got_error *
1247 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1248     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1249     got_cancel_cb cancel_cb, void *cancel_arg)
1250 {
1251 	const struct got_error *err = NULL;
1252 	DIR *packdir = NULL;
1253 	struct dirent *dent;
1254 	char *pack_relpath = NULL;
1255 	int packdir_fd;
1256 	struct stat sb;
1257 
1258 	packdir_fd = openat(got_repo_get_fd(repo),
1259 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1260 	if (packdir_fd == -1) {
1261 		if (errno == ENOENT)
1262 			return NULL;
1263 		return got_error_from_errno_fmt("openat: %s/%s",
1264 		    got_repo_get_path_git_dir(repo),
1265 		    GOT_OBJECTS_PACK_DIR);
1266 	}
1267 
1268 	packdir = fdopendir(packdir_fd);
1269 	if (packdir == NULL) {
1270 		err = got_error_from_errno("fdopendir");
1271 		goto done;
1272 	}
1273 
1274 	while ((dent = readdir(packdir)) != NULL) {
1275 		if (cancel_cb) {
1276 			err = cancel_cb(cancel_arg);
1277 			if (err)
1278 				goto done;
1279 		}
1280 
1281 		if (!got_repo_is_packidx_filename(repo, dent->d_name,
1282 		    dent->d_namlen))
1283 			continue;
1284 
1285 		err = got_packidx_get_packfile_path(&pack_relpath,
1286 		    dent->d_name);
1287 		if (err)
1288 			goto done;
1289 
1290 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1291 			free(pack_relpath);
1292 			pack_relpath = NULL;
1293 			continue;
1294 		}
1295 		if (errno != ENOENT) {
1296 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1297 			    got_repo_get_path_git_dir(repo),
1298 			    GOT_OBJECTS_PACK_DIR,
1299 			    pack_relpath);
1300 			goto done;
1301 		}
1302 
1303 		if (!dry_run) {
1304 			err = remove_packidx(packdir_fd, dent->d_name);
1305 			if (err)
1306 				goto done;
1307 		}
1308 		if (progress_cb) {
1309 			char *path;
1310 			if (asprintf(&path, "%s/%s/%s",
1311 			    got_repo_get_path_git_dir(repo),
1312 			    GOT_OBJECTS_PACK_DIR,
1313 			    dent->d_name) == -1) {
1314 				err = got_error_from_errno("asprintf");
1315 				goto done;
1316 			}
1317 			err = progress_cb(progress_arg, path);
1318 			free(path);
1319 			if (err)
1320 				goto done;
1321 		}
1322 		free(pack_relpath);
1323 		pack_relpath = NULL;
1324 	}
1325 done:
1326 	if (packdir && closedir(packdir) != 0 && err == NULL)
1327 		err = got_error_from_errno("closedir");
1328 	free(pack_relpath);
1329 	return err;
1330 }