1 /*
2  * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/stat.h>
22 #include <sys/socket.h>
23 #include <sys/wait.h>
24 
25 #include <dirent.h>
26 #include <endian.h>
27 #include <errno.h>
28 #include <fcntl.h>
29 #include <stdint.h>
30 #include <sha1.h>
31 #include <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <limits.h>
35 #include <unistd.h>
36 #include <imsg.h>
37 
38 #include "got_error.h"
39 #include "got_cancel.h"
40 #include "got_object.h"
41 #include "got_reference.h"
42 #include "got_repository.h"
43 #include "got_repository_admin.h"
44 #include "got_opentemp.h"
45 #include "got_path.h"
46 
47 #include "got_lib_delta.h"
48 #include "got_lib_object.h"
49 #include "got_lib_object_idset.h"
50 #include "got_lib_object_cache.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_repository.h"
54 #include "got_lib_pack_create.h"
55 #include "got_lib_sha1.h"
56 #include "got_lib_lockfile.h"
57 #include "got_lib_ratelimit.h"
58 
59 #ifndef nitems
60 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
61 #endif
62 
63 static const struct got_error *
64 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
65     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
66     struct got_repository *repo,
67     got_cancel_cb cancel_cb, void *cancel_arg)
68 {
69 	const struct got_error *err = NULL;
70 	const size_t alloc_chunksz = 256;
71 	size_t nalloc;
72 	struct got_reflist_entry *re;
73 	int i;
74 
75 	*ids = NULL;
76 	*nobjects = 0;
77 
78 	err = got_reflist_sort(refs,
79 	    got_ref_cmp_by_commit_timestamp_descending, repo);
80 	if (err)
81 		return err;
82 
83 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
84 	if (*ids == NULL)
85 		return got_error_from_errno("reallocarray");
86 	nalloc = alloc_chunksz;
87 
88 	TAILQ_FOREACH(re, refs, entry) {
89 		struct got_object_id *id;
90 
91 		if (cancel_cb) {
92 			err = cancel_cb(cancel_arg);
93 			if (err)
94 				goto done;
95 		}
96 
97 		err = got_ref_resolve(&id, repo, re->ref);
98 		if (err)
99 			goto done;
100 
101 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
102 			int obj_type;
103 			err = got_object_get_type(&obj_type, repo, id);
104 			if (err)
105 				goto done;
106 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
107 				free(id);
108 				id = NULL;
109 				continue;
110 			}
111 		}
112 
113 		if (nalloc <= *nobjects) {
114 			struct got_object_id **new;
115 			new = recallocarray(*ids, nalloc,
116 			    nalloc + alloc_chunksz,
117 			    sizeof(struct got_object_id *));
118 			if (new == NULL) {
119 				err = got_error_from_errno(
120 				    "recallocarray");
121 				goto done;
122 			}
123 			*ids = new;
124 			nalloc += alloc_chunksz;
125 		}
126 		(*ids)[*nobjects] = id;
127 		if ((*ids)[*nobjects] == NULL) {
128 			err = got_error_from_errno("got_object_id_dup");
129 			goto done;
130 		}
131 		(*nobjects)++;
132 	}
133 done:
134 	if (err) {
135 		for (i = 0; i < *nobjects; i++)
136 			free((*ids)[i]);
137 		free(*ids);
138 		*ids = NULL;
139 		*nobjects = 0;
140 	}
141 	return err;
142 }
143 
144 const struct got_error *
145 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
146     struct got_reflist_head *include_refs,
147     struct got_reflist_head *exclude_refs, struct got_repository *repo,
148     int loose_obj_only,
149     got_pack_progress_cb progress_cb, void *progress_arg,
150     got_cancel_cb cancel_cb, void *cancel_arg)
151 {
152 	const struct got_error *err = NULL;
153 	struct got_object_id **ours = NULL, **theirs = NULL;
154 	int nours = 0, ntheirs = 0, packfd = -1, i;
155 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
156 	char *sha1_str = NULL;
157 
158 	*packfile = NULL;
159 	*pack_hash = NULL;
160 
161 	if (asprintf(&path, "%s/%s/packing.pack",
162 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
163 		err = got_error_from_errno("asprintf");
164 		goto done;
165 	}
166 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
167 	if (err)
168 		goto done;
169 
170 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
171 		err = got_error_from_errno2("fchmod", tmpfile_path);
172 		goto done;
173 	}
174 
175 	*packfile = fdopen(packfd, "w");
176 	if (*packfile == NULL) {
177 		err = got_error_from_errno2("fdopen", tmpfile_path);
178 		goto done;
179 	}
180 	packfd = -1;
181 
182 	err = get_reflist_object_ids(&ours, &nours,
183 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
184 	    include_refs, repo, cancel_cb, cancel_arg);
185 	if (err)
186 		goto done;
187 
188 	if (nours == 0) {
189 		err = got_error(GOT_ERR_CANNOT_PACK);
190 		goto done;
191 	}
192 
193 	if (!TAILQ_EMPTY(exclude_refs)) {
194 		err = get_reflist_object_ids(&theirs, &ntheirs,
195 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
196 		    exclude_refs, repo,
197 		    cancel_cb, cancel_arg);
198 		if (err)
199 			goto done;
200 	}
201 
202 	*pack_hash = calloc(1, sizeof(**pack_hash));
203 	if (*pack_hash == NULL) {
204 		err = got_error_from_errno("calloc");
205 		goto done;
206 	}
207 
208 	err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
209 	    ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
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 (fflush(*packfile) == -1) {
225 		err = got_error_from_errno("fflush");
226 		goto done;
227 	}
228 	if (fseek(*packfile, 0L, SEEK_SET) == -1) {
229 		err = got_error_from_errno("fseek");
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 done:
240 	for (i = 0; i < nours; i++)
241 		free(ours[i]);
242 	free(ours);
243 	for (i = 0; i < ntheirs; i++)
244 		free(theirs[i]);
245 	free(theirs);
246 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
247 		err = got_error_from_errno2("close", packfile_path);
248 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
249 		err = got_error_from_errno2("unlink", tmpfile_path);
250 	free(tmpfile_path);
251 	free(packfile_path);
252 	free(sha1_str);
253 	free(path);
254 	if (err) {
255 		free(*pack_hash);
256 		*pack_hash = NULL;
257 		if (*packfile)
258 			fclose(*packfile);
259 		*packfile = NULL;
260 	}
261 	return err;
262 }
263 
264 const struct got_error *
265 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
266     struct got_repository *repo,
267     got_pack_index_progress_cb progress_cb, void *progress_arg,
268     got_cancel_cb cancel_cb, void *cancel_arg)
269 {
270 	size_t i;
271 	char *path;
272 	int imsg_idxfds[2];
273 	int npackfd = -1, idxfd = -1, nidxfd = -1;
274 	int tmpfds[3];
275 	int idxstatus, done = 0;
276 	const struct got_error *err;
277 	struct imsgbuf idxibuf;
278 	pid_t idxpid;
279 	char *tmpidxpath = NULL;
280 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
281 	const char *repo_path = got_repo_get_path_git_dir(repo);
282 	struct stat sb;
283 
284 	for (i = 0; i < nitems(tmpfds); i++)
285 		tmpfds[i] = -1;
286 
287 	if (asprintf(&path, "%s/%s/indexing.idx",
288 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
289 		err = got_error_from_errno("asprintf");
290 		goto done;
291 	}
292 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
293 	free(path);
294 	if (err)
295 		goto done;
296 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
297 		err = got_error_from_errno2("fchmod", tmpidxpath);
298 		goto done;
299 	}
300 
301 	nidxfd = dup(idxfd);
302 	if (nidxfd == -1) {
303 		err = got_error_from_errno("dup");
304 		goto done;
305 	}
306 
307 	for (i = 0; i < nitems(tmpfds); i++) {
308 		tmpfds[i] = got_opentempfd();
309 		if (tmpfds[i] == -1) {
310 			err = got_error_from_errno("got_opentempfd");
311 			goto done;
312 		}
313 	}
314 
315 	err = got_object_id_str(&id_str, pack_hash);
316 	if (err)
317 		goto done;
318 
319 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
320 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
321 		err = got_error_from_errno("asprintf");
322 		goto done;
323 	}
324 
325 	if (fstat(fileno(packfile), &sb) == -1) {
326 		err = got_error_from_errno2("fstat", packfile_path);
327 		goto done;
328 	}
329 
330 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
331 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
332 		err = got_error_from_errno("asprintf");
333 		goto done;
334 	}
335 
336 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
337 		err = got_error_from_errno("socketpair");
338 		goto done;
339 	}
340 	idxpid = fork();
341 	if (idxpid == -1) {
342 		err= got_error_from_errno("fork");
343 		goto done;
344 	} else if (idxpid == 0)
345 		got_privsep_exec_child(imsg_idxfds,
346 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
347 	if (close(imsg_idxfds[1]) == -1) {
348 		err = got_error_from_errno("close");
349 		goto done;
350 	}
351 	imsg_init(&idxibuf, imsg_idxfds[0]);
352 
353 	npackfd = dup(fileno(packfile));
354 	if (npackfd == -1) {
355 		err = got_error_from_errno("dup");
356 		goto done;
357 	}
358 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
359 	    npackfd);
360 	if (err != NULL)
361 		goto done;
362 	npackfd = -1;
363 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
364 	if (err != NULL)
365 		goto done;
366 	nidxfd = -1;
367 	for (i = 0; i < nitems(tmpfds); i++) {
368 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
369 		if (err != NULL)
370 			goto done;
371 		tmpfds[i] = -1;
372 	}
373 	done = 0;
374 	while (!done) {
375 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
376 
377 		if (cancel_cb) {
378 			err = cancel_cb(cancel_arg);
379 			if (err)
380 				goto done;
381 		}
382 
383 		err = got_privsep_recv_index_progress(&done, &nobj_total,
384 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
385 		    &idxibuf);
386 		if (err != NULL)
387 			goto done;
388 		if (nobj_indexed != 0) {
389 			err = progress_cb(progress_arg, sb.st_size,
390 			    nobj_total, nobj_indexed, nobj_loose,
391 			    nobj_resolved);
392 			if (err)
393 				break;
394 		}
395 	}
396 	if (close(imsg_idxfds[0]) == -1) {
397 		err = got_error_from_errno("close");
398 		goto done;
399 	}
400 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
401 		err = got_error_from_errno("waitpid");
402 		goto done;
403 	}
404 
405 	if (rename(tmpidxpath, idxpath) == -1) {
406 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
407 		goto done;
408 	}
409 	free(tmpidxpath);
410 	tmpidxpath = NULL;
411 
412 done:
413 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
414 		err = got_error_from_errno2("unlink", tmpidxpath);
415 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
416 		err = got_error_from_errno("close");
417 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
418 		err = got_error_from_errno("close");
419 	for (i = 0; i < nitems(tmpfds); i++) {
420 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
421 			err = got_error_from_errno("close");
422 	}
423 	free(tmpidxpath);
424 	free(idxpath);
425 	free(packfile_path);
426 	return err;
427 }
428 
429 const struct got_error *
430 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
431     struct got_repository *repo, const char *packfile_path)
432 {
433 	const struct got_error *err = NULL;
434 	const char *packdir_path = NULL;
435 	char *packfile_name = NULL, *p, *dot;
436 	struct got_object_id id;
437 	int packfd = -1;
438 
439 	*packfile = NULL;
440 	*pack_hash = NULL;
441 
442 	packdir_path = got_repo_get_path_objects_pack(repo);
443 	if (packdir_path == NULL)
444 		return got_error_from_errno("got_repo_get_path_objects_pack");
445 
446 	if (!got_path_is_child(packfile_path, packdir_path,
447 	    strlen(packdir_path))) {
448 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
449 		goto done;
450 
451 	}
452 
453 	err = got_path_basename(&packfile_name, packfile_path);
454 	if (err)
455 		goto done;
456 	p = packfile_name;
457 
458 	if (strncmp(p, "pack-", 5) != 0) {
459 		err = got_error_fmt(GOT_ERR_BAD_PATH,
460 		    "'%s' is not a valid pack file name",
461 		    packfile_name);
462 		goto done;
463 	}
464 	p += 5;
465 	dot = strchr(p, '.');
466 	if (dot == NULL) {
467 		err = got_error_fmt(GOT_ERR_BAD_PATH,
468 		    "'%s' is not a valid pack file name",
469 		    packfile_name);
470 		goto done;
471 	}
472 	if (strcmp(dot + 1, "pack") != 0) {
473 		err = got_error_fmt(GOT_ERR_BAD_PATH,
474 		    "'%s' is not a valid pack file name",
475 		    packfile_name);
476 		goto done;
477 	}
478 	*dot = '\0';
479 	if (!got_parse_sha1_digest(id.sha1, p)) {
480 		err = got_error_fmt(GOT_ERR_BAD_PATH,
481 		    "'%s' is not a valid pack file name",
482 		    packfile_name);
483 		goto done;
484 	}
485 
486 	*pack_hash = got_object_id_dup(&id);
487 	if (*pack_hash == NULL) {
488 		err = got_error_from_errno("got_object_id_dup");
489 		goto done;
490 	}
491 
492 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
493 	if (packfd == -1) {
494 		err = got_error_from_errno2("open", packfile_path);
495 		goto done;
496 	}
497 
498 	*packfile = fdopen(packfd, "r");
499 	if (*packfile == NULL) {
500 		err = got_error_from_errno2("fdopen", packfile_path);
501 		goto done;
502 	}
503 	packfd = -1;
504 done:
505 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
506 		err = got_error_from_errno2("close", packfile_path);
507 	free(packfile_name);
508 	if (err) {
509 		free(*pack_hash);
510 		*pack_hash = NULL;
511 	}
512 	return err;
513 }
514 
515 const struct got_error *
516 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
517     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
518     got_cancel_cb cancel_cb, void *cancel_arg)
519 {
520 	const struct got_error *err = NULL;
521 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
522 	struct got_packidx *packidx = NULL;
523 	struct got_pack *pack = NULL;
524 	uint32_t nobj, i;
525 
526 	err = got_object_id_str(&id_str, pack_hash);
527 	if (err)
528 		goto done;
529 
530 	if (asprintf(&packpath, "%s/pack-%s.pack",
531 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
532 		err = got_error_from_errno("asprintf");
533 		goto done;
534 	}
535 	if (asprintf(&idxpath, "%s/pack-%s.idx",
536 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
537 		err = got_error_from_errno("asprintf");
538 		goto done;
539 	}
540 
541 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
542 	if (err)
543 		goto done;
544 
545 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
546 	if (err)
547 		goto done;
548 
549 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
550 	for (i = 0; i < nobj; i++) {
551 		struct got_packidx_object_id *oid;
552 		struct got_object_id id, base_id;
553 		off_t offset, base_offset = 0;
554 		uint8_t type;
555 		uint64_t size;
556 		size_t tslen, len;
557 
558 		if (cancel_cb) {
559 			err = cancel_cb(cancel_arg);
560 			if (err)
561 				break;
562 		}
563 		oid = &packidx->hdr.sorted_ids[i];
564 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
565 
566 		offset = got_packidx_get_object_offset(packidx, i);
567 		if (offset == -1) {
568 			err = got_error(GOT_ERR_BAD_PACKIDX);
569 			goto done;
570 		}
571 
572 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
573 		    pack, offset);
574 		if (err)
575 			goto done;
576 
577 		switch (type) {
578 		case GOT_OBJ_TYPE_OFFSET_DELTA:
579 			err = got_pack_parse_offset_delta(&base_offset, &len,
580 			    pack, offset, tslen);
581 			if (err)
582 				goto done;
583 			break;
584 		case GOT_OBJ_TYPE_REF_DELTA:
585 			err = got_pack_parse_ref_delta(&base_id,
586 			    pack, offset, tslen);
587 			if (err)
588 				goto done;
589 			break;
590 		}
591 		err = (*list_cb)(list_arg, &id, type, offset, size,
592 		    base_offset, &base_id);
593 		if (err)
594 			goto done;
595 	}
596 
597 done:
598 	free(id_str);
599 	free(idxpath);
600 	free(packpath);
601 	if (packidx)
602 		got_packidx_close(packidx);
603 	return err;
604 }
605 
606 static const struct got_error *
607 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
608     void *progress_arg, struct got_ratelimit *rl,
609     int nloose, int ncommits, int npurged)
610 {
611 	const struct got_error *err;
612 	int elapsed;
613 
614 	if (progress_cb == NULL)
615 		return NULL;
616 
617 	err = got_ratelimit_check(&elapsed, rl);
618 	if (err || !elapsed)
619 		return err;
620 
621 	return progress_cb(progress_arg, nloose, ncommits, npurged);
622 }
623 
624 static const struct got_error *
625 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
626     got_cleanup_progress_cb progress_cb, void *progress_arg,
627     struct got_ratelimit *rl, struct got_repository *repo)
628 {
629 	const struct got_error *err = NULL;
630 	char *path_objects = NULL, *path = NULL;
631 	DIR *dir = NULL;
632 	struct got_object *obj = NULL;
633 	struct got_object_id id;
634 	int i, fd = -1;
635 	struct stat sb;
636 
637 	*ondisk_size = 0;
638 	*loose_ids = got_object_idset_alloc();
639 	if (*loose_ids == NULL)
640 		return got_error_from_errno("got_object_idset_alloc");
641 
642 	path_objects = got_repo_get_path_objects(repo);
643 	if (path_objects == NULL) {
644 		err = got_error_from_errno("got_repo_get_path_objects");
645 		goto done;
646 	}
647 
648 	for (i = 0; i <= 0xff; i++) {
649 		struct dirent *dent;
650 
651 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
652 			err = got_error_from_errno("asprintf");
653 			break;
654 		}
655 
656 		dir = opendir(path);
657 		if (dir == NULL) {
658 			if (errno == ENOENT) {
659 				err = NULL;
660 				continue;
661 			}
662 			err = got_error_from_errno2("opendir", path);
663 			break;
664 		}
665 
666 		while ((dent = readdir(dir)) != NULL) {
667 			char *id_str;
668 
669 			if (strcmp(dent->d_name, ".") == 0 ||
670 			    strcmp(dent->d_name, "..") == 0)
671 				continue;
672 
673 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
674 				err = got_error_from_errno("asprintf");
675 				goto done;
676 			}
677 
678 			memset(&id, 0, sizeof(id));
679 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
680 				free(id_str);
681 				continue;
682 			}
683 			free(id_str);
684 
685 			err = got_object_open_loose_fd(&fd, &id, repo);
686 			if (err)
687 				goto done;
688 			if (fstat(fd, &sb) == -1) {
689 				err = got_error_from_errno("fstat");
690 				goto done;
691 			}
692 			err = got_object_read_header_privsep(&obj, &id, repo,
693 			    fd);
694 			if (err)
695 				goto done;
696 			fd = -1; /* already closed */
697 
698 			switch (obj->type) {
699 			case GOT_OBJ_TYPE_COMMIT:
700 			case GOT_OBJ_TYPE_TREE:
701 			case GOT_OBJ_TYPE_BLOB:
702 			case GOT_OBJ_TYPE_TAG:
703 				break;
704 			default:
705 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
706 				    "%d", obj->type);
707 				goto done;
708 			}
709 			got_object_close(obj);
710 			obj = NULL;
711 			(*ondisk_size) += sb.st_size;
712 			err = got_object_idset_add(*loose_ids, &id, NULL);
713 			if (err)
714 				goto done;
715 			err = report_cleanup_progress(progress_cb,
716 			    progress_arg, rl,
717 			    got_object_idset_num_elements(*loose_ids), -1, -1);
718 			if (err)
719 				goto done;
720 		}
721 
722 		if (closedir(dir) != 0) {
723 			err = got_error_from_errno("closedir");
724 			goto done;
725 		}
726 		dir = NULL;
727 
728 		free(path);
729 		path = NULL;
730 	}
731 done:
732 	if (dir && closedir(dir) != 0 && err == NULL)
733 		err = got_error_from_errno("closedir");
734 	if (fd != -1 && close(fd) == -1 && err == NULL)
735 		err = got_error_from_errno("close");
736 	if (err) {
737 		got_object_idset_free(*loose_ids);
738 		*loose_ids = NULL;
739 	}
740 	if (obj)
741 		got_object_close(obj);
742 	free(path_objects);
743 	free(path);
744 	return err;
745 }
746 
747 static const struct got_error *
748 preserve_loose_object(struct got_object_idset *loose_ids,
749     struct got_object_id *id, struct got_repository *repo, int *npacked)
750 {
751 	const struct got_error *err = NULL;
752 	struct got_object *obj;
753 
754 	if (!got_object_idset_contains(loose_ids, id))
755 		return NULL;
756 
757 	/*
758 	 * Try to open this object from a pack file. This ensures that
759 	 * we do in fact have a valid packed copy of the object. Otherwise
760 	 * we should not delete the loose representation of this object.
761 	 */
762 	err = got_object_open_packed(&obj, id, repo);
763 	if (err == NULL) {
764 		got_object_close(obj);
765 		/*
766 		 * The object is referenced and packed.
767 		 * We can purge the redundantly stored loose object.
768 		 */
769 		(*npacked)++;
770 		return NULL;
771 	} else if (err->code != GOT_ERR_NO_OBJ)
772 		return err;
773 
774 	/*
775 	 * This object is referenced and not packed.
776 	 * Remove it from our purge set.
777 	 */
778 	return got_object_idset_remove(NULL, loose_ids, id);
779 }
780 
781 static const struct got_error *
782 load_tree_entries(struct got_object_id_queue *ids,
783     struct got_object_idset *loose_ids,
784     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
785     const char *dpath, struct got_repository *repo, int *npacked,
786     got_cancel_cb cancel_cb, void *cancel_arg)
787 {
788 	const struct got_error *err;
789 	struct got_tree_object *tree;
790 	char *p = NULL;
791 	int i;
792 
793 	err = got_object_open_as_tree(&tree, repo, tree_id);
794 	if (err)
795 		return err;
796 
797 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
798 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
799 		struct got_object_id *id = got_tree_entry_get_id(e);
800 		mode_t mode = got_tree_entry_get_mode(e);
801 
802 		if (cancel_cb) {
803 			err = (*cancel_cb)(cancel_arg);
804 			if (err)
805 				break;
806 		}
807 
808 		if (got_object_tree_entry_is_symlink(e) ||
809 		    got_object_tree_entry_is_submodule(e) ||
810 		    got_object_idset_contains(traversed_ids, id))
811 			continue;
812 
813 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
814 		    got_tree_entry_get_name(e)) == -1) {
815 			err = got_error_from_errno("asprintf");
816 			break;
817 		}
818 
819 		if (S_ISDIR(mode)) {
820 			struct got_object_qid *qid;
821 			err = got_object_qid_alloc(&qid, id);
822 			if (err)
823 				break;
824 			STAILQ_INSERT_TAIL(ids, qid, entry);
825 		} else if (S_ISREG(mode)) {
826 			/* This blob is referenced. */
827 			err = preserve_loose_object(loose_ids, id, repo,
828 			    npacked);
829 			if (err)
830 				break;
831 			err = got_object_idset_add(traversed_ids, id, NULL);
832 			if (err)
833 				break;
834 			
835 		}
836 		free(p);
837 		p = NULL;
838 	}
839 
840 	got_object_tree_close(tree);
841 	free(p);
842 	return err;
843 }
844 
845 static const struct got_error *
846 load_tree(struct got_object_idset *loose_ids,
847     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
848     const char *dpath, struct got_repository *repo, int *npacked,
849     got_cancel_cb cancel_cb, void *cancel_arg)
850 {
851 	const struct got_error *err = NULL;
852 	struct got_object_id_queue tree_ids;
853 	struct got_object_qid *qid;
854 
855 	err = got_object_qid_alloc(&qid, tree_id);
856 	if (err)
857 		return err;
858 
859 	STAILQ_INIT(&tree_ids);
860 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
861 
862 	while (!STAILQ_EMPTY(&tree_ids)) {
863 		if (cancel_cb) {
864 			err = (*cancel_cb)(cancel_arg);
865 			if (err)
866 				break;
867 		}
868 
869 		qid = STAILQ_FIRST(&tree_ids);
870 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
871 
872 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
873 			got_object_qid_free(qid);
874 			continue;
875 		}
876 
877 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
878 		if (err) {
879 			got_object_qid_free(qid);
880 			break;
881 		}
882 
883 		/* This tree is referenced. */
884 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
885 		if (err)
886 			break;
887 
888 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
889 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
890 		got_object_qid_free(qid);
891 		if (err)
892 			break;
893 	}
894 
895 	got_object_id_queue_free(&tree_ids);
896 	return err;
897 }
898 
899 static const struct got_error *
900 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
901     int *npacked, struct got_object_idset *traversed_ids,
902     struct got_object_id *id, struct got_repository *repo,
903     got_cleanup_progress_cb progress_cb, void *progress_arg,
904     struct got_ratelimit *rl, int nloose,
905     got_cancel_cb cancel_cb, void *cancel_arg)
906 {
907 	const struct got_error *err;
908 	struct got_commit_object *commit = NULL;
909 	struct got_tag_object *tag = NULL;
910 	struct got_object_id *tree_id = NULL;
911 	struct got_object_id_queue ids;
912 	struct got_object_qid *qid;
913 	int obj_type;
914 
915 	err = got_object_qid_alloc(&qid, id);
916 	if (err)
917 		return err;
918 
919 	STAILQ_INIT(&ids);
920 	STAILQ_INSERT_TAIL(&ids, qid, entry);
921 
922 	while (!STAILQ_EMPTY(&ids)) {
923 		if (cancel_cb) {
924 			err = (*cancel_cb)(cancel_arg);
925 			if (err)
926 				break;
927 		}
928 
929 		qid = STAILQ_FIRST(&ids);
930 		STAILQ_REMOVE_HEAD(&ids, entry);
931 
932 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
933 			got_object_qid_free(qid);
934 			qid = NULL;
935 			continue;
936 		}
937 
938 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
939 		if (err)
940 			break;
941 
942 		/* This commit or tag is referenced. */
943 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
944 		if (err)
945 			break;
946 
947 		err = got_object_get_type(&obj_type, repo, &qid->id);
948 		if (err)
949 			break;
950 		switch (obj_type) {
951 		case GOT_OBJ_TYPE_COMMIT:
952 			err = got_object_open_as_commit(&commit, repo,
953 			    &qid->id);
954 			if (err)
955 				goto done;
956 			break;
957 		case GOT_OBJ_TYPE_TAG:
958 			err = got_object_open_as_tag(&tag, repo, &qid->id);
959 			if (err)
960 				goto done;
961 			break;
962 		default:
963 			/* should not happen */
964 			err = got_error(GOT_ERR_OBJ_TYPE);
965 			goto done;
966 		}
967 
968 		/* Find a tree object to scan. */
969 		if (commit) {
970 			tree_id = got_object_commit_get_tree_id(commit);
971 		} else if (tag) {
972 			obj_type = got_object_tag_get_object_type(tag);
973 			switch (obj_type) {
974 			case GOT_OBJ_TYPE_COMMIT:
975 				err = got_object_open_as_commit(&commit, repo,
976 				    got_object_tag_get_object_id(tag));
977 				if (err)
978 					goto done;
979 				tree_id = got_object_commit_get_tree_id(commit);
980 				break;
981 			case GOT_OBJ_TYPE_TREE:
982 				tree_id = got_object_tag_get_object_id(tag);
983 				break;
984 			default:
985 				/*
986 				 * Tag points at something other than a
987 				 * commit or tree. Leave this weird tag object
988 				 * and the object it points to on disk.
989 				 */
990 				err = got_object_idset_remove(NULL, loose_ids,
991 				    &qid->id);
992 				if (err && err->code != GOT_ERR_NO_OBJ)
993 					goto done;
994 				err = got_object_idset_remove(NULL, loose_ids,
995 				    got_object_tag_get_object_id(tag));
996 				if (err && err->code != GOT_ERR_NO_OBJ)
997 					goto done;
998 				err = NULL;
999 				break;
1000 			}
1001 		}
1002 
1003 		if (tree_id) {
1004 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1005 			    repo, npacked, cancel_cb, cancel_arg);
1006 			if (err)
1007 				break;
1008 		}
1009 
1010 		if (commit || tag)
1011 			(*ncommits)++; /* scanned tags are counted as commits */
1012 
1013 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1014 		    nloose, *ncommits, -1);
1015 		if (err)
1016 			break;
1017  
1018 		if (commit) {
1019 			/* Find parent commits to scan. */
1020 			const struct got_object_id_queue *parent_ids;
1021 			parent_ids = got_object_commit_get_parent_ids(commit);
1022 			err = got_object_id_queue_copy(parent_ids, &ids);
1023 			if (err)
1024 				break;
1025 			got_object_commit_close(commit);
1026 			commit = NULL;
1027 		}
1028 		if (tag) {
1029 			got_object_tag_close(tag);
1030 			tag = NULL;
1031 		}
1032 		got_object_qid_free(qid);
1033 		qid = NULL;
1034 	}
1035 done:
1036 	if (qid)
1037 		got_object_qid_free(qid);
1038 	if (commit)
1039 		got_object_commit_close(commit);
1040 	if (tag)
1041 		got_object_tag_close(tag);
1042 	got_object_id_queue_free(&ids);
1043 	return err;
1044 }
1045 
1046 struct purge_loose_object_arg {
1047 	struct got_repository *repo;
1048 	got_cleanup_progress_cb progress_cb;
1049 	void *progress_arg;
1050 	struct got_ratelimit *rl;
1051 	int nloose;
1052 	int ncommits;
1053 	int npurged;
1054 	off_t size_purged;
1055 	int dry_run;
1056 	time_t max_mtime;
1057 	int ignore_mtime;
1058 };
1059 
1060 static const struct got_error *
1061 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1062 {
1063 	struct purge_loose_object_arg *a = arg;
1064 	const struct got_error *err, *unlock_err = NULL;
1065 	char *path = NULL;
1066 	int fd = -1;
1067 	struct stat sb;
1068 	struct got_lockfile *lf = NULL;
1069 
1070 	err = got_object_get_path(&path, id, a->repo);
1071 	if (err)
1072 		return err;
1073 
1074 	err = got_object_open_loose_fd(&fd, id, a->repo);
1075 	if (err)
1076 		goto done;
1077 
1078 	if (fstat(fd, &sb) == -1) {
1079 		err = got_error_from_errno("fstat");
1080 		goto done;
1081 	}
1082 
1083 	/*
1084 	 * Do not delete objects which are younger than our maximum
1085 	 * modification time threshold. This prevents a race where
1086 	 * new objects which are being added to the repository
1087 	 * concurrently would be deleted.
1088 	 */
1089 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1090 		if (!a->dry_run) {
1091 			err = got_lockfile_lock(&lf, path, -1);
1092 			if (err)
1093 				goto done;
1094 			if (unlink(path) == -1) {
1095 				err = got_error_from_errno2("unlink", path);
1096 				goto done;
1097 			}
1098 		}
1099 
1100 		a->npurged++;
1101 		a->size_purged += sb.st_size;
1102 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1103 		    a->rl, a->nloose, a->ncommits, a->npurged);
1104 		if (err)
1105 			goto done;
1106 	}
1107 done:
1108 	if (fd != -1 && close(fd) == -1 && err == NULL)
1109 		err = got_error_from_errno("close");
1110 	free(path);
1111 	if (lf)
1112 		unlock_err = got_lockfile_unlock(lf, -1);
1113 	return err ? err : unlock_err;
1114 }
1115 
1116 const struct got_error *
1117 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1118     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1119     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1120     got_cancel_cb cancel_cb, void *cancel_arg)
1121 {
1122 	const struct got_error *err;
1123 	struct got_object_idset *loose_ids;
1124 	struct got_object_idset *traversed_ids;
1125 	struct got_object_id **referenced_ids;
1126 	int i, nreferenced, nloose, ncommits = 0;
1127 	struct got_reflist_head refs;
1128 	struct got_reflist_entry *re;
1129 	struct purge_loose_object_arg arg;
1130 	time_t max_mtime = 0;
1131 	struct got_ratelimit rl;
1132 
1133 	TAILQ_INIT(&refs);
1134 	got_ratelimit_init(&rl, 0, 500);
1135 
1136 	*size_before = 0;
1137 	*size_after = 0;
1138 	*npacked = 0;
1139 
1140 	err = get_loose_object_ids(&loose_ids, size_before,
1141 	    progress_cb, progress_arg, &rl, repo);
1142 	if (err)
1143 		return err;
1144 	nloose = got_object_idset_num_elements(loose_ids);
1145 	if (nloose == 0) {
1146 		got_object_idset_free(loose_ids);
1147 		if (progress_cb) {
1148 			err = progress_cb(progress_arg, 0, 0, 0);
1149 			if (err)
1150 				return err;
1151 		}
1152 		return NULL;
1153 	}
1154 
1155 	traversed_ids = got_object_idset_alloc();
1156 	if (traversed_ids == NULL) {
1157 		err = got_error_from_errno("got_object_idset_alloc");
1158 		goto done;
1159 	}
1160 
1161 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1162 	if (err)
1163 		goto done;
1164 	if (!ignore_mtime) {
1165 		TAILQ_FOREACH(re, &refs, entry) {
1166 			time_t mtime = got_ref_get_mtime(re->ref);
1167 			if (mtime > max_mtime)
1168 				max_mtime = mtime;
1169 		}
1170 		/*
1171 		 * For safety, keep objects created within 10 minutes
1172 		 * before the youngest reference was created.
1173 		 */
1174 		if (max_mtime >= 600)
1175 			max_mtime -= 600;
1176 	}
1177 
1178 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1179 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1180 	    &refs, repo, cancel_cb, cancel_arg);
1181 	if (err)
1182 		goto done;
1183 
1184 	for (i = 0; i < nreferenced; i++) {
1185 		struct got_object_id *id = referenced_ids[i];
1186 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1187 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1188 		    nloose, cancel_cb, cancel_arg);
1189 		if (err)
1190 			goto done;
1191 	}
1192 
1193 	/* Any remaining loose objects are unreferenced and can be purged. */
1194 	arg.repo = repo;
1195 	arg.progress_arg = progress_arg;
1196 	arg.progress_cb = progress_cb;
1197 	arg.rl = &rl;
1198 	arg.nloose = nloose;
1199 	arg.npurged = 0;
1200 	arg.size_purged = 0;
1201 	arg.ncommits = ncommits;
1202 	arg.dry_run = dry_run;
1203 	arg.max_mtime = max_mtime;
1204 	arg.ignore_mtime = ignore_mtime;
1205 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1206 	if (err)
1207 		goto done;
1208 	*size_after = *size_before - arg.size_purged;
1209 
1210 	/* Produce a final progress report. */
1211 	if (progress_cb) {
1212 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1213 		if (err)
1214 			goto done;
1215 	}
1216 done:
1217 	got_object_idset_free(loose_ids);
1218 	got_object_idset_free(traversed_ids);
1219 	return err;
1220 }
1221 
1222 static const struct got_error *
1223 remove_packidx(int dir_fd, const char *relpath)
1224 {
1225 	const struct got_error *err, *unlock_err;
1226 	struct got_lockfile *lf;
1227 
1228 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1229 	if (err)
1230 		return err;
1231 	if (unlinkat(dir_fd, relpath, 0) == -1)
1232 		err = got_error_from_errno("unlinkat");
1233 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1234 	return err ? err : unlock_err;
1235 }
1236 
1237 const struct got_error *
1238 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1239     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1240     got_cancel_cb cancel_cb, void *cancel_arg)
1241 {
1242 	const struct got_error *err;
1243 	DIR *packdir = NULL;
1244 	struct dirent *dent;
1245 	char *pack_relpath = NULL;
1246 	int packdir_fd;
1247 	struct stat sb;
1248 
1249 	packdir_fd = openat(got_repo_get_fd(repo),
1250 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1251 	if (packdir_fd == -1) {
1252 		if (errno == ENOENT)
1253 			return NULL;
1254 		return got_error_from_errno_fmt("openat: %s/%s",
1255 		    got_repo_get_path_git_dir(repo),
1256 		    GOT_OBJECTS_PACK_DIR);
1257 	}
1258 
1259 	packdir = fdopendir(packdir_fd);
1260 	if (packdir == NULL) {
1261 		err = got_error_from_errno("fdopendir");
1262 		goto done;
1263 	}
1264 
1265 	while ((dent = readdir(packdir)) != NULL) {
1266 		if (cancel_cb) {
1267 			err = cancel_cb(cancel_arg);
1268 			if (err)
1269 				goto done;
1270 		}
1271 
1272 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1273 			continue;
1274 
1275 		err = got_packidx_get_packfile_path(&pack_relpath,
1276 		    dent->d_name);
1277 		if (err)
1278 			goto done;
1279 
1280 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1281 			free(pack_relpath);
1282 			pack_relpath = NULL;
1283 			continue;
1284 		}
1285 		if (errno != ENOENT) {
1286 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1287 			    got_repo_get_path_git_dir(repo),
1288 			    GOT_OBJECTS_PACK_DIR,
1289 			    pack_relpath);
1290 			goto done;
1291 		}
1292 
1293 		if (!dry_run) {
1294 			err = remove_packidx(packdir_fd, dent->d_name);
1295 			if (err)
1296 				goto done;
1297 		}
1298 		if (progress_cb) {
1299 			char *path;
1300 			if (asprintf(&path, "%s/%s/%s",
1301 			    got_repo_get_path_git_dir(repo),
1302 			    GOT_OBJECTS_PACK_DIR,
1303 			    dent->d_name) == -1) {
1304 				err = got_error_from_errno("asprintf");
1305 				goto done;
1306 			}
1307 			err = progress_cb(progress_arg, path);
1308 			free(path);
1309 			if (err)
1310 				goto done;
1311 		}
1312 		free(pack_relpath);
1313 		pack_relpath = NULL;
1314 	}
1315 done:
1316 	if (packdir && closedir(packdir) != 0 && err == NULL)
1317 		err = got_error_from_errno("closedir");
1318 	free(pack_relpath);
1319 	return err;
1320 }