1 /*
2  * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/uio.h>
19 #include <sys/stat.h>
20 #include <sys/socket.h>
21 #include <sys/wait.h>
22 
23 #include <dirent.h>
24 #include <errno.h>
25 #include <fcntl.h>
26 #include <stdint.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <limits.h>
31 #include <unistd.h>
32 
33 #include "got_error.h"
34 #include "got_cancel.h"
35 #include "got_object.h"
36 #include "got_reference.h"
37 #include "got_repository.h"
38 #include "got_repository_admin.h"
39 #include "got_opentemp.h"
40 #include "got_path.h"
41 
42 #include "got_lib_delta.h"
43 #include "got_lib_object.h"
44 #include "got_lib_object_idset.h"
45 #include "got_lib_object_cache.h"
46 #include "got_lib_pack.h"
47 #include "got_lib_privsep.h"
48 #include "got_lib_repository.h"
49 #include "got_lib_pack_create.h"
50 #include "got_lib_sha1.h"
51 #include "got_lib_lockfile.h"
52 #include "got_lib_ratelimit.h"
53 
54 #ifndef nitems
55 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
56 #endif
57 
58 static const struct got_error *
59 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
60     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
61     struct got_repository *repo,
62     got_cancel_cb cancel_cb, void *cancel_arg)
63 {
64 	const struct got_error *err = NULL;
65 	const size_t alloc_chunksz = 256;
66 	size_t nalloc;
67 	struct got_reflist_entry *re;
68 	int i;
69 
70 	*ids = NULL;
71 	*nobjects = 0;
72 
73 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
74 	if (*ids == NULL)
75 		return got_error_from_errno("reallocarray");
76 	nalloc = alloc_chunksz;
77 
78 	TAILQ_FOREACH(re, refs, entry) {
79 		struct got_object_id *id;
80 
81 		if (cancel_cb) {
82 			err = cancel_cb(cancel_arg);
83 			if (err)
84 				goto done;
85 		}
86 
87 		err = got_ref_resolve(&id, repo, re->ref);
88 		if (err)
89 			goto done;
90 
91 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
92 			int obj_type;
93 			err = got_object_get_type(&obj_type, repo, id);
94 			if (err)
95 				goto done;
96 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
97 				free(id);
98 				id = NULL;
99 				continue;
100 			}
101 		}
102 
103 		if (nalloc <= *nobjects) {
104 			struct got_object_id **new;
105 			new = recallocarray(*ids, nalloc,
106 			    nalloc + alloc_chunksz,
107 			    sizeof(struct got_object_id *));
108 			if (new == NULL) {
109 				err = got_error_from_errno(
110 				    "recallocarray");
111 				goto done;
112 			}
113 			*ids = new;
114 			nalloc += alloc_chunksz;
115 		}
116 		(*ids)[*nobjects] = id;
117 		if ((*ids)[*nobjects] == NULL) {
118 			err = got_error_from_errno("got_object_id_dup");
119 			goto done;
120 		}
121 		(*nobjects)++;
122 	}
123 done:
124 	if (err) {
125 		for (i = 0; i < *nobjects; i++)
126 			free((*ids)[i]);
127 		free(*ids);
128 		*ids = NULL;
129 		*nobjects = 0;
130 	}
131 	return err;
132 }
133 
134 const struct got_error *
135 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
136     struct got_reflist_head *include_refs,
137     struct got_reflist_head *exclude_refs, struct got_repository *repo,
138     int loose_obj_only,
139     got_pack_progress_cb progress_cb, void *progress_arg,
140     got_cancel_cb cancel_cb, void *cancel_arg)
141 {
142 	const struct got_error *err = NULL;
143 	struct got_object_id **ours = NULL, **theirs = NULL;
144 	int nours = 0, ntheirs = 0, packfd = -1, i;
145 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
146 	char *sha1_str = NULL;
147 
148 	*packfile = NULL;
149 	*pack_hash = NULL;
150 
151 	if (asprintf(&path, "%s/%s/packing.pack",
152 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
153 		err = got_error_from_errno("asprintf");
154 		goto done;
155 	}
156 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
157 	if (err)
158 		goto done;
159 
160 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
161 		err = got_error_from_errno2("fchmod", tmpfile_path);
162 		goto done;
163 	}
164 
165 	*packfile = fdopen(packfd, "w");
166 	if (*packfile == NULL) {
167 		err = got_error_from_errno2("fdopen", tmpfile_path);
168 		goto done;
169 	}
170 	packfd = -1;
171 
172 	err = get_reflist_object_ids(&ours, &nours,
173 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
174 	    include_refs, repo, cancel_cb, cancel_arg);
175 	if (err)
176 		goto done;
177 
178 	if (nours == 0) {
179 		err = got_error(GOT_ERR_CANNOT_PACK);
180 		goto done;
181 	}
182 
183 	if (!TAILQ_EMPTY(exclude_refs)) {
184 		err = get_reflist_object_ids(&theirs, &ntheirs,
185 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
186 		    exclude_refs, repo,
187 		    cancel_cb, cancel_arg);
188 		if (err)
189 			goto done;
190 	}
191 
192 	*pack_hash = calloc(1, sizeof(**pack_hash));
193 	if (*pack_hash == NULL) {
194 		err = got_error_from_errno("calloc");
195 		goto done;
196 	}
197 
198 	err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
199 	    ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
200 	    cancel_cb, cancel_arg);
201 	if (err)
202 		goto done;
203 
204 	err = got_object_id_str(&sha1_str, *pack_hash);
205 	if (err)
206 		goto done;
207 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
208 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
209 	    sha1_str) == -1) {
210 		err = got_error_from_errno("asprintf");
211 		goto done;
212 	}
213 
214 	if (fflush(*packfile) == -1) {
215 		err = got_error_from_errno("fflush");
216 		goto done;
217 	}
218 	if (fseek(*packfile, 0L, SEEK_SET) == -1) {
219 		err = got_error_from_errno("fseek");
220 		goto done;
221 	}
222 	if (rename(tmpfile_path, packfile_path) == -1) {
223 		err = got_error_from_errno3("rename", tmpfile_path,
224 		    packfile_path);
225 		goto done;
226 	}
227 	free(tmpfile_path);
228 	tmpfile_path = NULL;
229 done:
230 	for (i = 0; i < nours; i++)
231 		free(ours[i]);
232 	free(ours);
233 	for (i = 0; i < ntheirs; i++)
234 		free(theirs[i]);
235 	free(theirs);
236 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
237 		err = got_error_from_errno2("close", packfile_path);
238 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
239 		err = got_error_from_errno2("unlink", tmpfile_path);
240 	free(tmpfile_path);
241 	free(packfile_path);
242 	free(sha1_str);
243 	free(path);
244 	if (err) {
245 		free(*pack_hash);
246 		*pack_hash = NULL;
247 		if (*packfile)
248 			fclose(*packfile);
249 		*packfile = NULL;
250 	}
251 	return err;
252 }
253 
254 const struct got_error *
255 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
256     struct got_repository *repo,
257     got_pack_index_progress_cb progress_cb, void *progress_arg,
258     got_cancel_cb cancel_cb, void *cancel_arg)
259 {
260 	size_t i;
261 	char *path;
262 	int imsg_idxfds[2];
263 	int npackfd = -1, idxfd = -1, nidxfd = -1;
264 	int tmpfds[3];
265 	int idxstatus, done = 0;
266 	const struct got_error *err;
267 	struct imsgbuf idxibuf;
268 	pid_t idxpid;
269 	char *tmpidxpath = NULL;
270 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
271 	const char *repo_path = got_repo_get_path_git_dir(repo);
272 	struct stat sb;
273 
274 	for (i = 0; i < nitems(tmpfds); i++)
275 		tmpfds[i] = -1;
276 
277 	if (asprintf(&path, "%s/%s/indexing.idx",
278 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
279 		err = got_error_from_errno("asprintf");
280 		goto done;
281 	}
282 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
283 	free(path);
284 	if (err)
285 		goto done;
286 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
287 		err = got_error_from_errno2("fchmod", tmpidxpath);
288 		goto done;
289 	}
290 
291 	nidxfd = dup(idxfd);
292 	if (nidxfd == -1) {
293 		err = got_error_from_errno("dup");
294 		goto done;
295 	}
296 
297 	for (i = 0; i < nitems(tmpfds); i++) {
298 		tmpfds[i] = got_opentempfd();
299 		if (tmpfds[i] == -1) {
300 			err = got_error_from_errno("got_opentempfd");
301 			goto done;
302 		}
303 	}
304 
305 	err = got_object_id_str(&id_str, pack_hash);
306 	if (err)
307 		goto done;
308 
309 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
310 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
311 		err = got_error_from_errno("asprintf");
312 		goto done;
313 	}
314 
315 	if (fstat(fileno(packfile), &sb) == -1) {
316 		err = got_error_from_errno2("fstat", packfile_path);
317 		goto done;
318 	}
319 
320 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
321 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
322 		err = got_error_from_errno("asprintf");
323 		goto done;
324 	}
325 
326 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
327 		err = got_error_from_errno("socketpair");
328 		goto done;
329 	}
330 	idxpid = fork();
331 	if (idxpid == -1) {
332 		err= got_error_from_errno("fork");
333 		goto done;
334 	} else if (idxpid == 0)
335 		got_privsep_exec_child(imsg_idxfds,
336 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
337 	if (close(imsg_idxfds[1]) == -1) {
338 		err = got_error_from_errno("close");
339 		goto done;
340 	}
341 	imsg_init(&idxibuf, imsg_idxfds[0]);
342 
343 	npackfd = dup(fileno(packfile));
344 	if (npackfd == -1) {
345 		err = got_error_from_errno("dup");
346 		goto done;
347 	}
348 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
349 	    npackfd);
350 	if (err != NULL)
351 		goto done;
352 	npackfd = -1;
353 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
354 	if (err != NULL)
355 		goto done;
356 	nidxfd = -1;
357 	for (i = 0; i < nitems(tmpfds); i++) {
358 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
359 		if (err != NULL)
360 			goto done;
361 		tmpfds[i] = -1;
362 	}
363 	done = 0;
364 	while (!done) {
365 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
366 
367 		if (cancel_cb) {
368 			err = cancel_cb(cancel_arg);
369 			if (err)
370 				goto done;
371 		}
372 
373 		err = got_privsep_recv_index_progress(&done, &nobj_total,
374 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
375 		    &idxibuf);
376 		if (err != NULL)
377 			goto done;
378 		if (nobj_indexed != 0) {
379 			err = progress_cb(progress_arg, sb.st_size,
380 			    nobj_total, nobj_indexed, nobj_loose,
381 			    nobj_resolved);
382 			if (err)
383 				break;
384 		}
385 	}
386 	if (close(imsg_idxfds[0]) == -1) {
387 		err = got_error_from_errno("close");
388 		goto done;
389 	}
390 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
391 		err = got_error_from_errno("waitpid");
392 		goto done;
393 	}
394 
395 	if (rename(tmpidxpath, idxpath) == -1) {
396 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
397 		goto done;
398 	}
399 	free(tmpidxpath);
400 	tmpidxpath = NULL;
401 
402 done:
403 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
404 		err = got_error_from_errno2("unlink", tmpidxpath);
405 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
406 		err = got_error_from_errno("close");
407 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
408 		err = got_error_from_errno("close");
409 	for (i = 0; i < nitems(tmpfds); i++) {
410 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
411 			err = got_error_from_errno("close");
412 	}
413 	free(tmpidxpath);
414 	free(idxpath);
415 	free(packfile_path);
416 	return err;
417 }
418 
419 const struct got_error *
420 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
421     struct got_repository *repo, const char *packfile_path)
422 {
423 	const struct got_error *err = NULL;
424 	const char *packdir_path = NULL;
425 	char *packfile_name = NULL, *p, *dot;
426 	struct got_object_id id;
427 	int packfd = -1;
428 
429 	*packfile = NULL;
430 	*pack_hash = NULL;
431 
432 	packdir_path = got_repo_get_path_objects_pack(repo);
433 	if (packdir_path == NULL)
434 		return got_error_from_errno("got_repo_get_path_objects_pack");
435 
436 	if (!got_path_is_child(packfile_path, packdir_path,
437 	    strlen(packdir_path))) {
438 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
439 		goto done;
440 
441 	}
442 
443 	err = got_path_basename(&packfile_name, packfile_path);
444 	if (err)
445 		goto done;
446 	p = packfile_name;
447 
448 	if (strncmp(p, "pack-", 5) != 0) {
449 		err = got_error_fmt(GOT_ERR_BAD_PATH,
450 		    "'%s' is not a valid pack file name",
451 		    packfile_name);
452 		goto done;
453 	}
454 	p += 5;
455 	dot = strchr(p, '.');
456 	if (dot == NULL) {
457 		err = got_error_fmt(GOT_ERR_BAD_PATH,
458 		    "'%s' is not a valid pack file name",
459 		    packfile_name);
460 		goto done;
461 	}
462 	if (strcmp(dot + 1, "pack") != 0) {
463 		err = got_error_fmt(GOT_ERR_BAD_PATH,
464 		    "'%s' is not a valid pack file name",
465 		    packfile_name);
466 		goto done;
467 	}
468 	*dot = '\0';
469 	if (!got_parse_sha1_digest(id.sha1, p)) {
470 		err = got_error_fmt(GOT_ERR_BAD_PATH,
471 		    "'%s' is not a valid pack file name",
472 		    packfile_name);
473 		goto done;
474 	}
475 
476 	*pack_hash = got_object_id_dup(&id);
477 	if (*pack_hash == NULL) {
478 		err = got_error_from_errno("got_object_id_dup");
479 		goto done;
480 	}
481 
482 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
483 	if (packfd == -1) {
484 		err = got_error_from_errno2("open", packfile_path);
485 		goto done;
486 	}
487 
488 	*packfile = fdopen(packfd, "r");
489 	if (*packfile == NULL) {
490 		err = got_error_from_errno2("fdopen", packfile_path);
491 		goto done;
492 	}
493 	packfd = -1;
494 done:
495 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
496 		err = got_error_from_errno2("close", packfile_path);
497 	free(packfile_name);
498 	if (err) {
499 		free(*pack_hash);
500 		*pack_hash = NULL;
501 	}
502 	return err;
503 }
504 
505 const struct got_error *
506 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
507     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
508     got_cancel_cb cancel_cb, void *cancel_arg)
509 {
510 	const struct got_error *err = NULL;
511 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
512 	struct got_packidx *packidx = NULL;
513 	struct got_pack *pack = NULL;
514 	uint32_t nobj, i;
515 
516 	err = got_object_id_str(&id_str, pack_hash);
517 	if (err)
518 		goto done;
519 
520 	if (asprintf(&packpath, "%s/pack-%s.pack",
521 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
522 		err = got_error_from_errno("asprintf");
523 		goto done;
524 	}
525 	if (asprintf(&idxpath, "%s/pack-%s.idx",
526 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
527 		err = got_error_from_errno("asprintf");
528 		goto done;
529 	}
530 
531 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
532 	if (err)
533 		goto done;
534 
535 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
536 	if (err)
537 		goto done;
538 
539 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
540 	for (i = 0; i < nobj; i++) {
541 		struct got_packidx_object_id *oid;
542 		struct got_object_id id, base_id;
543 		off_t offset, base_offset = 0;
544 		uint8_t type;
545 		uint64_t size;
546 		size_t tslen, len;
547 
548 		if (cancel_cb) {
549 			err = cancel_cb(cancel_arg);
550 			if (err)
551 				break;
552 		}
553 		oid = &packidx->hdr.sorted_ids[i];
554 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
555 
556 		offset = got_packidx_get_object_offset(packidx, i);
557 		if (offset == -1) {
558 			err = got_error(GOT_ERR_BAD_PACKIDX);
559 			goto done;
560 		}
561 
562 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
563 		    pack, offset);
564 		if (err)
565 			goto done;
566 
567 		switch (type) {
568 		case GOT_OBJ_TYPE_OFFSET_DELTA:
569 			err = got_pack_parse_offset_delta(&base_offset, &len,
570 			    pack, offset, tslen);
571 			if (err)
572 				goto done;
573 			break;
574 		case GOT_OBJ_TYPE_REF_DELTA:
575 			err = got_pack_parse_ref_delta(&base_id,
576 			    pack, offset, tslen);
577 			if (err)
578 				goto done;
579 			break;
580 		}
581 		err = (*list_cb)(list_arg, &id, type, offset, size,
582 		    base_offset, &base_id);
583 		if (err)
584 			goto done;
585 	}
586 
587 done:
588 	free(id_str);
589 	free(idxpath);
590 	free(packpath);
591 	if (packidx)
592 		got_packidx_close(packidx);
593 	return err;
594 }
595 
596 static const struct got_error *
597 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
598     void *progress_arg, struct got_ratelimit *rl,
599     int nloose, int ncommits, int npurged)
600 {
601 	const struct got_error *err;
602 	int elapsed;
603 
604 	if (progress_cb == NULL)
605 		return NULL;
606 
607 	err = got_ratelimit_check(&elapsed, rl);
608 	if (err || !elapsed)
609 		return err;
610 
611 	return progress_cb(progress_arg, nloose, ncommits, npurged);
612 }
613 
614 static const struct got_error *
615 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
616     got_cleanup_progress_cb progress_cb, void *progress_arg,
617     struct got_ratelimit *rl, struct got_repository *repo)
618 {
619 	const struct got_error *err = NULL;
620 	char *path_objects = NULL, *path = NULL;
621 	DIR *dir = NULL;
622 	struct got_object *obj = NULL;
623 	struct got_object_id id;
624 	int i, fd = -1;
625 	struct stat sb;
626 
627 	*ondisk_size = 0;
628 	*loose_ids = got_object_idset_alloc();
629 	if (*loose_ids == NULL)
630 		return got_error_from_errno("got_object_idset_alloc");
631 
632 	path_objects = got_repo_get_path_objects(repo);
633 	if (path_objects == NULL) {
634 		err = got_error_from_errno("got_repo_get_path_objects");
635 		goto done;
636 	}
637 
638 	for (i = 0; i <= 0xff; i++) {
639 		struct dirent *dent;
640 
641 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
642 			err = got_error_from_errno("asprintf");
643 			break;
644 		}
645 
646 		dir = opendir(path);
647 		if (dir == NULL) {
648 			if (errno == ENOENT) {
649 				err = NULL;
650 				continue;
651 			}
652 			err = got_error_from_errno2("opendir", path);
653 			break;
654 		}
655 
656 		while ((dent = readdir(dir)) != NULL) {
657 			char *id_str;
658 
659 			if (strcmp(dent->d_name, ".") == 0 ||
660 			    strcmp(dent->d_name, "..") == 0)
661 				continue;
662 
663 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
664 				err = got_error_from_errno("asprintf");
665 				goto done;
666 			}
667 
668 			memset(&id, 0, sizeof(id));
669 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
670 				free(id_str);
671 				continue;
672 			}
673 			free(id_str);
674 
675 			err = got_object_open_loose_fd(&fd, &id, repo);
676 			if (err)
677 				goto done;
678 			if (fstat(fd, &sb) == -1) {
679 				err = got_error_from_errno("fstat");
680 				goto done;
681 			}
682 			err = got_object_read_header_privsep(&obj, &id, repo,
683 			    fd);
684 			if (err)
685 				goto done;
686 			fd = -1; /* already closed */
687 
688 			switch (obj->type) {
689 			case GOT_OBJ_TYPE_COMMIT:
690 			case GOT_OBJ_TYPE_TREE:
691 			case GOT_OBJ_TYPE_BLOB:
692 			case GOT_OBJ_TYPE_TAG:
693 				break;
694 			default:
695 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
696 				    "%d", obj->type);
697 				goto done;
698 			}
699 			got_object_close(obj);
700 			obj = NULL;
701 			(*ondisk_size) += sb.st_size;
702 			err = got_object_idset_add(*loose_ids, &id, NULL);
703 			if (err)
704 				goto done;
705 			err = report_cleanup_progress(progress_cb,
706 			    progress_arg, rl,
707 			    got_object_idset_num_elements(*loose_ids), -1, -1);
708 			if (err)
709 				goto done;
710 		}
711 
712 		if (closedir(dir) != 0) {
713 			err = got_error_from_errno("closedir");
714 			goto done;
715 		}
716 		dir = NULL;
717 
718 		free(path);
719 		path = NULL;
720 	}
721 done:
722 	if (dir && closedir(dir) != 0 && err == NULL)
723 		err = got_error_from_errno("closedir");
724 	if (fd != -1 && close(fd) == -1 && err == NULL)
725 		err = got_error_from_errno("close");
726 	if (err) {
727 		got_object_idset_free(*loose_ids);
728 		*loose_ids = NULL;
729 	}
730 	if (obj)
731 		got_object_close(obj);
732 	free(path_objects);
733 	free(path);
734 	return err;
735 }
736 
737 static const struct got_error *
738 preserve_loose_object(struct got_object_idset *loose_ids,
739     struct got_object_id *id, struct got_repository *repo, int *npacked)
740 {
741 	const struct got_error *err = NULL;
742 	struct got_object *obj;
743 
744 	if (!got_object_idset_contains(loose_ids, id))
745 		return NULL;
746 
747 	/*
748 	 * Try to open this object from a pack file. This ensures that
749 	 * we do in fact have a valid packed copy of the object. Otherwise
750 	 * we should not delete the loose representation of this object.
751 	 */
752 	err = got_object_open_packed(&obj, id, repo);
753 	if (err == NULL) {
754 		got_object_close(obj);
755 		/*
756 		 * The object is referenced and packed.
757 		 * We can purge the redundantly stored loose object.
758 		 */
759 		(*npacked)++;
760 		return NULL;
761 	} else if (err->code != GOT_ERR_NO_OBJ)
762 		return err;
763 
764 	/*
765 	 * This object is referenced and not packed.
766 	 * Remove it from our purge set.
767 	 */
768 	return got_object_idset_remove(NULL, loose_ids, id);
769 }
770 
771 static const struct got_error *
772 load_tree_entries(struct got_object_id_queue *ids,
773     struct got_object_idset *loose_ids,
774     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
775     const char *dpath, struct got_repository *repo, int *npacked,
776     got_cancel_cb cancel_cb, void *cancel_arg)
777 {
778 	const struct got_error *err;
779 	struct got_tree_object *tree;
780 	char *p = NULL;
781 	int i;
782 
783 	err = got_object_open_as_tree(&tree, repo, tree_id);
784 	if (err)
785 		return err;
786 
787 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
788 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
789 		struct got_object_id *id = got_tree_entry_get_id(e);
790 		mode_t mode = got_tree_entry_get_mode(e);
791 
792 		if (cancel_cb) {
793 			err = (*cancel_cb)(cancel_arg);
794 			if (err)
795 				break;
796 		}
797 
798 		if (got_object_tree_entry_is_symlink(e) ||
799 		    got_object_tree_entry_is_submodule(e) ||
800 		    got_object_idset_contains(traversed_ids, id))
801 			continue;
802 
803 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
804 		    got_tree_entry_get_name(e)) == -1) {
805 			err = got_error_from_errno("asprintf");
806 			break;
807 		}
808 
809 		if (S_ISDIR(mode)) {
810 			struct got_object_qid *qid;
811 			err = got_object_qid_alloc(&qid, id);
812 			if (err)
813 				break;
814 			STAILQ_INSERT_TAIL(ids, qid, entry);
815 		} else if (S_ISREG(mode)) {
816 			/* This blob is referenced. */
817 			err = preserve_loose_object(loose_ids, id, repo,
818 			    npacked);
819 			if (err)
820 				break;
821 			err = got_object_idset_add(traversed_ids, id, NULL);
822 			if (err)
823 				break;
824 			
825 		}
826 		free(p);
827 		p = NULL;
828 	}
829 
830 	got_object_tree_close(tree);
831 	free(p);
832 	return err;
833 }
834 
835 static const struct got_error *
836 load_tree(struct got_object_idset *loose_ids,
837     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
838     const char *dpath, struct got_repository *repo, int *npacked,
839     got_cancel_cb cancel_cb, void *cancel_arg)
840 {
841 	const struct got_error *err = NULL;
842 	struct got_object_id_queue tree_ids;
843 	struct got_object_qid *qid;
844 
845 	err = got_object_qid_alloc(&qid, tree_id);
846 	if (err)
847 		return err;
848 
849 	STAILQ_INIT(&tree_ids);
850 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
851 
852 	while (!STAILQ_EMPTY(&tree_ids)) {
853 		if (cancel_cb) {
854 			err = (*cancel_cb)(cancel_arg);
855 			if (err)
856 				break;
857 		}
858 
859 		qid = STAILQ_FIRST(&tree_ids);
860 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
861 
862 		if (got_object_idset_contains(traversed_ids, qid->id)) {
863 			got_object_qid_free(qid);
864 			continue;
865 		}
866 
867 		err = got_object_idset_add(traversed_ids, qid->id, NULL);
868 		if (err) {
869 			got_object_qid_free(qid);
870 			break;
871 		}
872 
873 		/* This tree is referenced. */
874 		err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
875 		if (err)
876 			break;
877 
878 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
879 		    qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
880 		got_object_qid_free(qid);
881 		if (err)
882 			break;
883 	}
884 
885 	got_object_id_queue_free(&tree_ids);
886 	return err;
887 }
888 
889 static const struct got_error *
890 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
891     int *npacked, struct got_object_idset *traversed_ids,
892     struct got_object_id *id, struct got_repository *repo,
893     got_cleanup_progress_cb progress_cb, void *progress_arg,
894     struct got_ratelimit *rl, int nloose,
895     got_cancel_cb cancel_cb, void *cancel_arg)
896 {
897 	const struct got_error *err;
898 	struct got_commit_object *commit = NULL;
899 	struct got_tag_object *tag = NULL;
900 	struct got_object_id *tree_id = NULL;
901 	struct got_object_id_queue ids;
902 	struct got_object_qid *qid;
903 	int obj_type;
904 
905 	err = got_object_qid_alloc(&qid, id);
906 	if (err)
907 		return err;
908 
909 	STAILQ_INIT(&ids);
910 	STAILQ_INSERT_TAIL(&ids, qid, entry);
911 
912 	while (!STAILQ_EMPTY(&ids)) {
913 		if (cancel_cb) {
914 			err = (*cancel_cb)(cancel_arg);
915 			if (err)
916 				break;
917 		}
918 
919 		qid = STAILQ_FIRST(&ids);
920 		STAILQ_REMOVE_HEAD(&ids, entry);
921 
922 		if (got_object_idset_contains(traversed_ids, qid->id)) {
923 			got_object_qid_free(qid);
924 			qid = NULL;
925 			continue;
926 		}
927 
928 		err = got_object_idset_add(traversed_ids, qid->id, NULL);
929 		if (err)
930 			break;
931 
932 		/* This commit or tag is referenced. */
933 		err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
934 		if (err)
935 			break;
936 
937 		err = got_object_get_type(&obj_type, repo, qid->id);
938 		if (err)
939 			break;
940 		switch (obj_type) {
941 		case GOT_OBJ_TYPE_COMMIT:
942 			err = got_object_open_as_commit(&commit, repo, qid->id);
943 			if (err)
944 				goto done;
945 			break;
946 		case GOT_OBJ_TYPE_TAG:
947 			err = got_object_open_as_tag(&tag, repo, qid->id);
948 			if (err)
949 				goto done;
950 			break;
951 		default:
952 			/* should not happen */
953 			err = got_error(GOT_ERR_OBJ_TYPE);
954 			goto done;
955 		}
956 
957 		/* Find a tree object to scan. */
958 		if (commit) {
959 			tree_id = got_object_commit_get_tree_id(commit);
960 		} else if (tag) {
961 			obj_type = got_object_tag_get_object_type(tag);
962 			switch (obj_type) {
963 			case GOT_OBJ_TYPE_COMMIT:
964 				err = got_object_open_as_commit(&commit, repo,
965 				    got_object_tag_get_object_id(tag));
966 				if (err)
967 					goto done;
968 				tree_id = got_object_commit_get_tree_id(commit);
969 				break;
970 			case GOT_OBJ_TYPE_TREE:
971 				tree_id = got_object_tag_get_object_id(tag);
972 				break;
973 			default:
974 				/*
975 				 * Tag points at something other than a
976 				 * commit or tree. Leave this weird tag object
977 				 * and the object it points to on disk.
978 				 */
979 				err = got_object_idset_remove(NULL, loose_ids,
980 				    qid->id);
981 				if (err && err->code != GOT_ERR_NO_OBJ)
982 					goto done;
983 				err = got_object_idset_remove(NULL, loose_ids,
984 				    got_object_tag_get_object_id(tag));
985 				if (err && err->code != GOT_ERR_NO_OBJ)
986 					goto done;
987 				err = NULL;
988 				break;
989 			}
990 		}
991 
992 		if (tree_id) {
993 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
994 			    repo, npacked, cancel_cb, cancel_arg);
995 			if (err)
996 				break;
997 		}
998 
999 		if (commit || tag)
1000 			(*ncommits)++; /* scanned tags are counted as commits */
1001 
1002 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1003 		    nloose, *ncommits, -1);
1004 		if (err)
1005 			break;
1006  
1007 		if (commit) {
1008 			/* Find parent commits to scan. */
1009 			const struct got_object_id_queue *parent_ids;
1010 			parent_ids = got_object_commit_get_parent_ids(commit);
1011 			err = got_object_id_queue_copy(parent_ids, &ids);
1012 			if (err)
1013 				break;
1014 			got_object_commit_close(commit);
1015 			commit = NULL;
1016 		}
1017 		if (tag) {
1018 			got_object_tag_close(tag);
1019 			tag = NULL;
1020 		}
1021 		got_object_qid_free(qid);
1022 		qid = NULL;
1023 	}
1024 done:
1025 	if (qid)
1026 		got_object_qid_free(qid);
1027 	if (commit)
1028 		got_object_commit_close(commit);
1029 	if (tag)
1030 		got_object_tag_close(tag);
1031 	got_object_id_queue_free(&ids);
1032 	return err;
1033 }
1034 
1035 struct purge_loose_object_arg {
1036 	struct got_repository *repo;
1037 	got_cleanup_progress_cb progress_cb;
1038 	void *progress_arg;
1039 	struct got_ratelimit *rl;
1040 	int nloose;
1041 	int ncommits;
1042 	int npurged;
1043 	off_t size_purged;
1044 	int dry_run;
1045 	time_t max_mtime;
1046 	int ignore_mtime;
1047 };
1048 
1049 static const struct got_error *
1050 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1051 {
1052 	struct purge_loose_object_arg *a = arg;
1053 	const struct got_error *err, *unlock_err = NULL;
1054 	char *path = NULL;
1055 	int fd = -1;
1056 	struct stat sb;
1057 	struct got_lockfile *lf = NULL;
1058 
1059 	err = got_object_get_path(&path, id, a->repo);
1060 	if (err)
1061 		return err;
1062 
1063 	err = got_object_open_loose_fd(&fd, id, a->repo);
1064 	if (err)
1065 		goto done;
1066 
1067 	if (fstat(fd, &sb) == -1) {
1068 		err = got_error_from_errno("fstat");
1069 		goto done;
1070 	}
1071 
1072 	/*
1073 	 * Do not delete objects which are younger than our maximum
1074 	 * modification time threshold. This prevents a race where
1075 	 * new objects which are being added to the repository
1076 	 * concurrently would be deleted.
1077 	 */
1078 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1079 		if (!a->dry_run) {
1080 			err = got_lockfile_lock(&lf, path, -1);
1081 			if (err)
1082 				goto done;
1083 			if (unlink(path) == -1) {
1084 				err = got_error_from_errno2("unlink", path);
1085 				goto done;
1086 			}
1087 		}
1088 
1089 		a->npurged++;
1090 		a->size_purged += sb.st_size;
1091 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1092 		    a->rl, a->nloose, a->ncommits, a->npurged);
1093 		if (err)
1094 			goto done;
1095 	}
1096 done:
1097 	if (fd != -1 && close(fd) == -1 && err == NULL)
1098 		err = got_error_from_errno("close");
1099 	free(path);
1100 	if (lf)
1101 		unlock_err = got_lockfile_unlock(lf, -1);
1102 	return err ? err : unlock_err;
1103 }
1104 
1105 const struct got_error *
1106 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1107     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1108     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1109     got_cancel_cb cancel_cb, void *cancel_arg)
1110 {
1111 	const struct got_error *err;
1112 	struct got_object_idset *loose_ids;
1113 	struct got_object_idset *traversed_ids;
1114 	struct got_object_id **referenced_ids;
1115 	int i, nreferenced, nloose, ncommits = 0;
1116 	struct got_reflist_head refs;
1117 	struct got_reflist_entry *re;
1118 	struct purge_loose_object_arg arg;
1119 	time_t max_mtime = 0;
1120 	struct got_ratelimit rl;
1121 
1122 	TAILQ_INIT(&refs);
1123 	got_ratelimit_init(&rl, 0, 500);
1124 
1125 	*size_before = 0;
1126 	*size_after = 0;
1127 	*npacked = 0;
1128 
1129 	err = get_loose_object_ids(&loose_ids, size_before,
1130 	    progress_cb, progress_arg, &rl, repo);
1131 	if (err)
1132 		return err;
1133 	nloose = got_object_idset_num_elements(loose_ids);
1134 	if (nloose == 0) {
1135 		got_object_idset_free(loose_ids);
1136 		if (progress_cb) {
1137 			err = progress_cb(progress_arg, 0, 0, 0);
1138 			if (err)
1139 				return err;
1140 		}
1141 		return NULL;
1142 	}
1143 
1144 	traversed_ids = got_object_idset_alloc();
1145 	if (traversed_ids == NULL) {
1146 		err = got_error_from_errno("got_object_idset_alloc");
1147 		goto done;
1148 	}
1149 
1150 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1151 	if (err)
1152 		goto done;
1153 	if (!ignore_mtime) {
1154 		TAILQ_FOREACH(re, &refs, entry) {
1155 			time_t mtime = got_ref_get_mtime(re->ref);
1156 			if (mtime > max_mtime)
1157 				max_mtime = mtime;
1158 		}
1159 		/*
1160 		 * For safety, keep objects created within 10 minutes
1161 		 * before the youngest reference was created.
1162 		 */
1163 		if (max_mtime >= 600)
1164 			max_mtime -= 600;
1165 	}
1166 
1167 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1168 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1169 	    &refs, repo, cancel_cb, cancel_arg);
1170 	if (err)
1171 		goto done;
1172 
1173 	for (i = 0; i < nreferenced; i++) {
1174 		struct got_object_id *id = referenced_ids[i];
1175 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1176 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1177 		    nloose, cancel_cb, cancel_arg);
1178 		if (err)
1179 			goto done;
1180 	}
1181 
1182 	/* Any remaining loose objects are unreferenced and can be purged. */
1183 	arg.repo = repo;
1184 	arg.progress_arg = progress_arg;
1185 	arg.progress_cb = progress_cb;
1186 	arg.rl = &rl;
1187 	arg.nloose = nloose;
1188 	arg.npurged = 0;
1189 	arg.size_purged = 0;
1190 	arg.ncommits = ncommits;
1191 	arg.dry_run = dry_run;
1192 	arg.max_mtime = max_mtime;
1193 	arg.ignore_mtime = ignore_mtime;
1194 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1195 	if (err)
1196 		goto done;
1197 	*size_after = *size_before - arg.size_purged;
1198 
1199 	/* Produce a final progress report. */
1200 	if (progress_cb) {
1201 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1202 		if (err)
1203 			goto done;
1204 	}
1205 done:
1206 	got_object_idset_free(loose_ids);
1207 	got_object_idset_free(traversed_ids);
1208 	return err;
1209 }
1210 
1211 static const struct got_error *
1212 remove_packidx(int dir_fd, const char *relpath)
1213 {
1214 	const struct got_error *err, *unlock_err;
1215 	struct got_lockfile *lf;
1216 
1217 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1218 	if (err)
1219 		return err;
1220 	if (unlinkat(dir_fd, relpath, 0) == -1)
1221 		err = got_error_from_errno("unlinkat");
1222 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1223 	return err ? err : unlock_err;
1224 }
1225 
1226 const struct got_error *
1227 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1228     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1229     got_cancel_cb cancel_cb, void *cancel_arg)
1230 {
1231 	const struct got_error *err;
1232 	DIR *packdir = NULL;
1233 	struct dirent *dent;
1234 	char *pack_relpath = NULL;
1235 	int packdir_fd;
1236 	struct stat sb;
1237 
1238 	packdir_fd = openat(got_repo_get_fd(repo),
1239 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1240 	if (packdir_fd == -1) {
1241 		if (errno == ENOENT)
1242 			return NULL;
1243 		return got_error_from_errno_fmt("openat: %s/%s",
1244 		    got_repo_get_path_git_dir(repo),
1245 		    GOT_OBJECTS_PACK_DIR);
1246 	}
1247 
1248 	packdir = fdopendir(packdir_fd);
1249 	if (packdir == NULL) {
1250 		err = got_error_from_errno("fdopendir");
1251 		goto done;
1252 	}
1253 
1254 	while ((dent = readdir(packdir)) != NULL) {
1255 		if (cancel_cb) {
1256 			err = cancel_cb(cancel_arg);
1257 			if (err)
1258 				goto done;
1259 		}
1260 
1261 		if (!got_repo_is_packidx_filename(dent->d_name,
1262 		    strlen(dent->d_name)))
1263 			continue;
1264 
1265 		err = got_packidx_get_packfile_path(&pack_relpath,
1266 		    dent->d_name);
1267 		if (err)
1268 			goto done;
1269 
1270 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1271 			free(pack_relpath);
1272 			pack_relpath = NULL;
1273 			continue;
1274 		}
1275 		if (errno != ENOENT) {
1276 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1277 			    got_repo_get_path_git_dir(repo),
1278 			    GOT_OBJECTS_PACK_DIR,
1279 			    pack_relpath);
1280 			goto done;
1281 		}
1282 
1283 		if (!dry_run) {
1284 			err = remove_packidx(packdir_fd, dent->d_name);
1285 			if (err)
1286 				goto done;
1287 		}
1288 		if (progress_cb) {
1289 			char *path;
1290 			if (asprintf(&path, "%s/%s/%s",
1291 			    got_repo_get_path_git_dir(repo),
1292 			    GOT_OBJECTS_PACK_DIR,
1293 			    dent->d_name) == -1) {
1294 				err = got_error_from_errno("asprintf");
1295 				goto done;
1296 			}
1297 			err = progress_cb(progress_arg, path);
1298 			free(path);
1299 			if (err)
1300 				goto done;
1301 		}
1302 		free(pack_relpath);
1303 		pack_relpath = NULL;
1304 	}
1305 done:
1306 	if (packdir && closedir(packdir) != 0 && err == NULL)
1307 		err = got_error_from_errno("closedir");
1308 	free(pack_relpath);
1309 	return err;
1310 }