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 *sha1_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(&sha1_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 	    sha1_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(sha1_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 * SHA1_DIGEST_LENGTH;
574 		memset(&id, 0, sizeof(id));
575 		memcpy(id.hash, oid->hash, SHA1_DIGEST_LENGTH);
576 
577 		offset = got_packidx_get_object_offset(packidx, i);
578 		if (offset == -1) {
579 			err = got_error(GOT_ERR_BAD_PACKIDX);
580 			goto done;
581 		}
582 
583 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
584 		    pack, offset);
585 		if (err)
586 			goto done;
587 
588 		switch (type) {
589 		case GOT_OBJ_TYPE_OFFSET_DELTA:
590 			err = got_pack_parse_offset_delta(&base_offset, &len,
591 			    pack, offset, tslen);
592 			if (err)
593 				goto done;
594 			break;
595 		case GOT_OBJ_TYPE_REF_DELTA:
596 			err = got_pack_parse_ref_delta(&base_id,
597 			    pack, offset, tslen);
598 			if (err)
599 				goto done;
600 			break;
601 		}
602 		err = (*list_cb)(list_arg, &id, type, offset, size,
603 		    base_offset, &base_id);
604 		if (err)
605 			goto done;
606 	}
607 
608 done:
609 	free(id_str);
610 	free(idxpath);
611 	free(packpath);
612 	if (packidx)
613 		got_packidx_close(packidx);
614 	return err;
615 }
616 
617 static const struct got_error *
618 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
619     void *progress_arg, struct got_ratelimit *rl,
620     int nloose, int ncommits, int npurged)
621 {
622 	const struct got_error *err;
623 	int elapsed;
624 
625 	if (progress_cb == NULL)
626 		return NULL;
627 
628 	err = got_ratelimit_check(&elapsed, rl);
629 	if (err || !elapsed)
630 		return err;
631 
632 	return progress_cb(progress_arg, nloose, ncommits, npurged);
633 }
634 
635 static const struct got_error *
636 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
637     got_cleanup_progress_cb progress_cb, void *progress_arg,
638     struct got_ratelimit *rl, struct got_repository *repo)
639 {
640 	const struct got_error *err = NULL;
641 	char *path_objects = NULL, *path = NULL;
642 	DIR *dir = NULL;
643 	struct got_object *obj = NULL;
644 	struct got_object_id id;
645 	int i, fd = -1;
646 	struct stat sb;
647 
648 	*ondisk_size = 0;
649 	*loose_ids = got_object_idset_alloc();
650 	if (*loose_ids == NULL)
651 		return got_error_from_errno("got_object_idset_alloc");
652 
653 	path_objects = got_repo_get_path_objects(repo);
654 	if (path_objects == NULL) {
655 		err = got_error_from_errno("got_repo_get_path_objects");
656 		goto done;
657 	}
658 
659 	for (i = 0; i <= 0xff; i++) {
660 		struct dirent *dent;
661 
662 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
663 			err = got_error_from_errno("asprintf");
664 			break;
665 		}
666 
667 		dir = opendir(path);
668 		if (dir == NULL) {
669 			if (errno == ENOENT) {
670 				err = NULL;
671 				continue;
672 			}
673 			err = got_error_from_errno2("opendir", path);
674 			break;
675 		}
676 
677 		while ((dent = readdir(dir)) != NULL) {
678 			char *id_str;
679 
680 			if (strcmp(dent->d_name, ".") == 0 ||
681 			    strcmp(dent->d_name, "..") == 0)
682 				continue;
683 
684 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
685 				err = got_error_from_errno("asprintf");
686 				goto done;
687 			}
688 
689 			memset(&id, 0, sizeof(id));
690 			if (!got_parse_sha1_digest(id.hash, id_str)) {
691 				free(id_str);
692 				continue;
693 			}
694 			free(id_str);
695 
696 			err = got_object_open_loose_fd(&fd, &id, repo);
697 			if (err)
698 				goto done;
699 			if (fstat(fd, &sb) == -1) {
700 				err = got_error_from_errno("fstat");
701 				goto done;
702 			}
703 			err = got_object_read_header_privsep(&obj, &id, repo,
704 			    fd);
705 			if (err)
706 				goto done;
707 			fd = -1; /* already closed */
708 
709 			switch (obj->type) {
710 			case GOT_OBJ_TYPE_COMMIT:
711 			case GOT_OBJ_TYPE_TREE:
712 			case GOT_OBJ_TYPE_BLOB:
713 			case GOT_OBJ_TYPE_TAG:
714 				break;
715 			default:
716 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
717 				    "%d", obj->type);
718 				goto done;
719 			}
720 			got_object_close(obj);
721 			obj = NULL;
722 			(*ondisk_size) += sb.st_size;
723 			err = got_object_idset_add(*loose_ids, &id, NULL);
724 			if (err)
725 				goto done;
726 			err = report_cleanup_progress(progress_cb,
727 			    progress_arg, rl,
728 			    got_object_idset_num_elements(*loose_ids), -1, -1);
729 			if (err)
730 				goto done;
731 		}
732 
733 		if (closedir(dir) != 0) {
734 			err = got_error_from_errno("closedir");
735 			goto done;
736 		}
737 		dir = NULL;
738 
739 		free(path);
740 		path = NULL;
741 	}
742 done:
743 	if (dir && closedir(dir) != 0 && err == NULL)
744 		err = got_error_from_errno("closedir");
745 	if (fd != -1 && close(fd) == -1 && err == NULL)
746 		err = got_error_from_errno("close");
747 	if (err) {
748 		got_object_idset_free(*loose_ids);
749 		*loose_ids = NULL;
750 	}
751 	if (obj)
752 		got_object_close(obj);
753 	free(path_objects);
754 	free(path);
755 	return err;
756 }
757 
758 static const struct got_error *
759 preserve_loose_object(struct got_object_idset *loose_ids,
760     struct got_object_id *id, struct got_repository *repo, int *npacked)
761 {
762 	const struct got_error *err = NULL;
763 	struct got_object *obj;
764 
765 	if (!got_object_idset_contains(loose_ids, id))
766 		return NULL;
767 
768 	/*
769 	 * Try to open this object from a pack file. This ensures that
770 	 * we do in fact have a valid packed copy of the object. Otherwise
771 	 * we should not delete the loose representation of this object.
772 	 */
773 	err = got_object_open_packed(&obj, id, repo);
774 	if (err == NULL) {
775 		got_object_close(obj);
776 		/*
777 		 * The object is referenced and packed.
778 		 * We can purge the redundantly stored loose object.
779 		 */
780 		(*npacked)++;
781 		return NULL;
782 	} else if (err->code != GOT_ERR_NO_OBJ)
783 		return err;
784 
785 	/*
786 	 * This object is referenced and not packed.
787 	 * Remove it from our purge set.
788 	 */
789 	return got_object_idset_remove(NULL, loose_ids, id);
790 }
791 
792 static const struct got_error *
793 load_tree_entries(struct got_object_id_queue *ids,
794     struct got_object_idset *loose_ids,
795     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
796     const char *dpath, struct got_repository *repo, int *npacked,
797     got_cancel_cb cancel_cb, void *cancel_arg)
798 {
799 	const struct got_error *err;
800 	struct got_tree_object *tree;
801 	char *p = NULL;
802 	int i;
803 
804 	err = got_object_open_as_tree(&tree, repo, tree_id);
805 	if (err)
806 		return err;
807 
808 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
809 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
810 		struct got_object_id *id = got_tree_entry_get_id(e);
811 		mode_t mode = got_tree_entry_get_mode(e);
812 
813 		if (cancel_cb) {
814 			err = (*cancel_cb)(cancel_arg);
815 			if (err)
816 				break;
817 		}
818 
819 		if (got_object_tree_entry_is_symlink(e) ||
820 		    got_object_tree_entry_is_submodule(e) ||
821 		    got_object_idset_contains(traversed_ids, id))
822 			continue;
823 
824 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
825 		    got_tree_entry_get_name(e)) == -1) {
826 			err = got_error_from_errno("asprintf");
827 			break;
828 		}
829 
830 		if (S_ISDIR(mode)) {
831 			struct got_object_qid *qid;
832 			err = got_object_qid_alloc(&qid, id);
833 			if (err)
834 				break;
835 			STAILQ_INSERT_TAIL(ids, qid, entry);
836 		} else if (S_ISREG(mode)) {
837 			/* This blob is referenced. */
838 			err = preserve_loose_object(loose_ids, id, repo,
839 			    npacked);
840 			if (err)
841 				break;
842 			err = got_object_idset_add(traversed_ids, id, NULL);
843 			if (err)
844 				break;
845 		}
846 		free(p);
847 		p = NULL;
848 	}
849 
850 	got_object_tree_close(tree);
851 	free(p);
852 	return err;
853 }
854 
855 static const struct got_error *
856 load_tree(struct got_object_idset *loose_ids,
857     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
858     const char *dpath, struct got_repository *repo, int *npacked,
859     got_cancel_cb cancel_cb, void *cancel_arg)
860 {
861 	const struct got_error *err = NULL;
862 	struct got_object_id_queue tree_ids;
863 	struct got_object_qid *qid;
864 
865 	err = got_object_qid_alloc(&qid, tree_id);
866 	if (err)
867 		return err;
868 
869 	STAILQ_INIT(&tree_ids);
870 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
871 
872 	while (!STAILQ_EMPTY(&tree_ids)) {
873 		if (cancel_cb) {
874 			err = (*cancel_cb)(cancel_arg);
875 			if (err)
876 				break;
877 		}
878 
879 		qid = STAILQ_FIRST(&tree_ids);
880 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
881 
882 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
883 			got_object_qid_free(qid);
884 			continue;
885 		}
886 
887 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
888 		if (err) {
889 			got_object_qid_free(qid);
890 			break;
891 		}
892 
893 		/* This tree is referenced. */
894 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
895 		if (err)
896 			break;
897 
898 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
899 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
900 		got_object_qid_free(qid);
901 		if (err)
902 			break;
903 	}
904 
905 	got_object_id_queue_free(&tree_ids);
906 	return err;
907 }
908 
909 static const struct got_error *
910 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
911     int *npacked, struct got_object_idset *traversed_ids,
912     struct got_object_id *id, struct got_repository *repo,
913     got_cleanup_progress_cb progress_cb, void *progress_arg,
914     struct got_ratelimit *rl, int nloose,
915     got_cancel_cb cancel_cb, void *cancel_arg)
916 {
917 	const struct got_error *err;
918 	struct got_commit_object *commit = NULL;
919 	struct got_tag_object *tag = NULL;
920 	struct got_object_id *tree_id = NULL;
921 	struct got_object_id_queue ids;
922 	struct got_object_qid *qid;
923 	int obj_type;
924 
925 	err = got_object_qid_alloc(&qid, id);
926 	if (err)
927 		return err;
928 
929 	STAILQ_INIT(&ids);
930 	STAILQ_INSERT_TAIL(&ids, qid, entry);
931 
932 	while (!STAILQ_EMPTY(&ids)) {
933 		if (cancel_cb) {
934 			err = (*cancel_cb)(cancel_arg);
935 			if (err)
936 				break;
937 		}
938 
939 		qid = STAILQ_FIRST(&ids);
940 		STAILQ_REMOVE_HEAD(&ids, entry);
941 
942 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
943 			got_object_qid_free(qid);
944 			qid = NULL;
945 			continue;
946 		}
947 
948 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
949 		if (err)
950 			break;
951 
952 		/* This commit or tag is referenced. */
953 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
954 		if (err)
955 			break;
956 
957 		err = got_object_get_type(&obj_type, repo, &qid->id);
958 		if (err)
959 			break;
960 		switch (obj_type) {
961 		case GOT_OBJ_TYPE_COMMIT:
962 			err = got_object_open_as_commit(&commit, repo,
963 			    &qid->id);
964 			if (err)
965 				goto done;
966 			break;
967 		case GOT_OBJ_TYPE_TAG:
968 			err = got_object_open_as_tag(&tag, repo, &qid->id);
969 			if (err)
970 				goto done;
971 			break;
972 		default:
973 			/* should not happen */
974 			err = got_error(GOT_ERR_OBJ_TYPE);
975 			goto done;
976 		}
977 
978 		/* Find a tree object to scan. */
979 		if (commit) {
980 			tree_id = got_object_commit_get_tree_id(commit);
981 		} else if (tag) {
982 			obj_type = got_object_tag_get_object_type(tag);
983 			switch (obj_type) {
984 			case GOT_OBJ_TYPE_COMMIT:
985 				err = got_object_open_as_commit(&commit, repo,
986 				    got_object_tag_get_object_id(tag));
987 				if (err)
988 					goto done;
989 				tree_id = got_object_commit_get_tree_id(commit);
990 				break;
991 			case GOT_OBJ_TYPE_TREE:
992 				tree_id = got_object_tag_get_object_id(tag);
993 				break;
994 			default:
995 				/*
996 				 * Tag points at something other than a
997 				 * commit or tree. Leave this weird tag object
998 				 * and the object it points to on disk.
999 				 */
1000 				err = got_object_idset_remove(NULL, loose_ids,
1001 				    &qid->id);
1002 				if (err && err->code != GOT_ERR_NO_OBJ)
1003 					goto done;
1004 				err = got_object_idset_remove(NULL, loose_ids,
1005 				    got_object_tag_get_object_id(tag));
1006 				if (err && err->code != GOT_ERR_NO_OBJ)
1007 					goto done;
1008 				err = NULL;
1009 				break;
1010 			}
1011 		}
1012 
1013 		if (tree_id) {
1014 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1015 			    repo, npacked, cancel_cb, cancel_arg);
1016 			if (err)
1017 				break;
1018 		}
1019 
1020 		if (commit || tag)
1021 			(*ncommits)++; /* scanned tags are counted as commits */
1022 
1023 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1024 		    nloose, *ncommits, -1);
1025 		if (err)
1026 			break;
1027 
1028 		if (commit) {
1029 			/* Find parent commits to scan. */
1030 			const struct got_object_id_queue *parent_ids;
1031 			parent_ids = got_object_commit_get_parent_ids(commit);
1032 			err = got_object_id_queue_copy(parent_ids, &ids);
1033 			if (err)
1034 				break;
1035 			got_object_commit_close(commit);
1036 			commit = NULL;
1037 		}
1038 		if (tag) {
1039 			got_object_tag_close(tag);
1040 			tag = NULL;
1041 		}
1042 		got_object_qid_free(qid);
1043 		qid = NULL;
1044 	}
1045 done:
1046 	if (qid)
1047 		got_object_qid_free(qid);
1048 	if (commit)
1049 		got_object_commit_close(commit);
1050 	if (tag)
1051 		got_object_tag_close(tag);
1052 	got_object_id_queue_free(&ids);
1053 	return err;
1054 }
1055 
1056 struct purge_loose_object_arg {
1057 	struct got_repository *repo;
1058 	got_cleanup_progress_cb progress_cb;
1059 	void *progress_arg;
1060 	struct got_ratelimit *rl;
1061 	int nloose;
1062 	int ncommits;
1063 	int npurged;
1064 	off_t size_purged;
1065 	int dry_run;
1066 	time_t max_mtime;
1067 	int ignore_mtime;
1068 };
1069 
1070 static const struct got_error *
1071 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1072 {
1073 	struct purge_loose_object_arg *a = arg;
1074 	const struct got_error *err, *unlock_err = NULL;
1075 	char *path = NULL;
1076 	int fd = -1;
1077 	struct stat sb;
1078 	struct got_lockfile *lf = NULL;
1079 
1080 	err = got_object_get_path(&path, id, a->repo);
1081 	if (err)
1082 		return err;
1083 
1084 	err = got_object_open_loose_fd(&fd, id, a->repo);
1085 	if (err)
1086 		goto done;
1087 
1088 	if (fstat(fd, &sb) == -1) {
1089 		err = got_error_from_errno("fstat");
1090 		goto done;
1091 	}
1092 
1093 	/*
1094 	 * Do not delete objects which are younger than our maximum
1095 	 * modification time threshold. This prevents a race where
1096 	 * new objects which are being added to the repository
1097 	 * concurrently would be deleted.
1098 	 */
1099 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1100 		if (!a->dry_run) {
1101 			err = got_lockfile_lock(&lf, path, -1);
1102 			if (err)
1103 				goto done;
1104 			if (unlink(path) == -1) {
1105 				err = got_error_from_errno2("unlink", path);
1106 				goto done;
1107 			}
1108 		}
1109 
1110 		a->npurged++;
1111 		a->size_purged += sb.st_size;
1112 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1113 		    a->rl, a->nloose, a->ncommits, a->npurged);
1114 		if (err)
1115 			goto done;
1116 	}
1117 done:
1118 	if (fd != -1 && close(fd) == -1 && err == NULL)
1119 		err = got_error_from_errno("close");
1120 	free(path);
1121 	if (lf)
1122 		unlock_err = got_lockfile_unlock(lf, -1);
1123 	return err ? err : unlock_err;
1124 }
1125 
1126 const struct got_error *
1127 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1128     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1129     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1130     got_cancel_cb cancel_cb, void *cancel_arg)
1131 {
1132 	const struct got_error *err;
1133 	struct got_object_idset *loose_ids;
1134 	struct got_object_idset *traversed_ids;
1135 	struct got_object_id **referenced_ids;
1136 	int i, nreferenced, nloose, ncommits = 0;
1137 	struct got_reflist_head refs;
1138 	struct got_reflist_entry *re;
1139 	struct purge_loose_object_arg arg;
1140 	time_t max_mtime = 0;
1141 	struct got_ratelimit rl;
1142 
1143 	TAILQ_INIT(&refs);
1144 	got_ratelimit_init(&rl, 0, 500);
1145 
1146 	*size_before = 0;
1147 	*size_after = 0;
1148 	*npacked = 0;
1149 
1150 	err = get_loose_object_ids(&loose_ids, size_before,
1151 	    progress_cb, progress_arg, &rl, repo);
1152 	if (err)
1153 		return err;
1154 	nloose = got_object_idset_num_elements(loose_ids);
1155 	if (nloose == 0) {
1156 		got_object_idset_free(loose_ids);
1157 		if (progress_cb) {
1158 			err = progress_cb(progress_arg, 0, 0, 0);
1159 			if (err)
1160 				return err;
1161 		}
1162 		return NULL;
1163 	}
1164 
1165 	traversed_ids = got_object_idset_alloc();
1166 	if (traversed_ids == NULL) {
1167 		err = got_error_from_errno("got_object_idset_alloc");
1168 		goto done;
1169 	}
1170 
1171 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1172 	if (err)
1173 		goto done;
1174 	if (!ignore_mtime) {
1175 		TAILQ_FOREACH(re, &refs, entry) {
1176 			time_t mtime = got_ref_get_mtime(re->ref);
1177 			if (mtime > max_mtime)
1178 				max_mtime = mtime;
1179 		}
1180 		/*
1181 		 * For safety, keep objects created within 10 minutes
1182 		 * before the youngest reference was created.
1183 		 */
1184 		if (max_mtime >= 600)
1185 			max_mtime -= 600;
1186 	}
1187 
1188 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1189 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1190 	    &refs, repo, cancel_cb, cancel_arg);
1191 	if (err)
1192 		goto done;
1193 
1194 	for (i = 0; i < nreferenced; i++) {
1195 		struct got_object_id *id = referenced_ids[i];
1196 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1197 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1198 		    nloose, cancel_cb, cancel_arg);
1199 		if (err)
1200 			goto done;
1201 	}
1202 
1203 	/* Any remaining loose objects are unreferenced and can be purged. */
1204 	arg.repo = repo;
1205 	arg.progress_arg = progress_arg;
1206 	arg.progress_cb = progress_cb;
1207 	arg.rl = &rl;
1208 	arg.nloose = nloose;
1209 	arg.npurged = 0;
1210 	arg.size_purged = 0;
1211 	arg.ncommits = ncommits;
1212 	arg.dry_run = dry_run;
1213 	arg.max_mtime = max_mtime;
1214 	arg.ignore_mtime = ignore_mtime;
1215 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1216 	if (err)
1217 		goto done;
1218 	*size_after = *size_before - arg.size_purged;
1219 
1220 	/* Produce a final progress report. */
1221 	if (progress_cb) {
1222 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1223 		if (err)
1224 			goto done;
1225 	}
1226 done:
1227 	got_object_idset_free(loose_ids);
1228 	got_object_idset_free(traversed_ids);
1229 	return err;
1230 }
1231 
1232 static const struct got_error *
1233 remove_packidx(int dir_fd, const char *relpath)
1234 {
1235 	const struct got_error *err, *unlock_err;
1236 	struct got_lockfile *lf;
1237 
1238 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1239 	if (err)
1240 		return err;
1241 	if (unlinkat(dir_fd, relpath, 0) == -1)
1242 		err = got_error_from_errno("unlinkat");
1243 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1244 	return err ? err : unlock_err;
1245 }
1246 
1247 const struct got_error *
1248 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1249     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1250     got_cancel_cb cancel_cb, void *cancel_arg)
1251 {
1252 	const struct got_error *err = NULL;
1253 	DIR *packdir = NULL;
1254 	struct dirent *dent;
1255 	char *pack_relpath = NULL;
1256 	int packdir_fd;
1257 	struct stat sb;
1258 
1259 	packdir_fd = openat(got_repo_get_fd(repo),
1260 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1261 	if (packdir_fd == -1) {
1262 		if (errno == ENOENT)
1263 			return NULL;
1264 		return got_error_from_errno_fmt("openat: %s/%s",
1265 		    got_repo_get_path_git_dir(repo),
1266 		    GOT_OBJECTS_PACK_DIR);
1267 	}
1268 
1269 	packdir = fdopendir(packdir_fd);
1270 	if (packdir == NULL) {
1271 		err = got_error_from_errno("fdopendir");
1272 		goto done;
1273 	}
1274 
1275 	while ((dent = readdir(packdir)) != NULL) {
1276 		if (cancel_cb) {
1277 			err = cancel_cb(cancel_arg);
1278 			if (err)
1279 				goto done;
1280 		}
1281 
1282 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen,
1283 		    got_repo_get_object_format(repo)))
1284 			continue;
1285 
1286 		err = got_packidx_get_packfile_path(&pack_relpath,
1287 		    dent->d_name);
1288 		if (err)
1289 			goto done;
1290 
1291 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1292 			free(pack_relpath);
1293 			pack_relpath = NULL;
1294 			continue;
1295 		}
1296 		if (errno != ENOENT) {
1297 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1298 			    got_repo_get_path_git_dir(repo),
1299 			    GOT_OBJECTS_PACK_DIR,
1300 			    pack_relpath);
1301 			goto done;
1302 		}
1303 
1304 		if (!dry_run) {
1305 			err = remove_packidx(packdir_fd, dent->d_name);
1306 			if (err)
1307 				goto done;
1308 		}
1309 		if (progress_cb) {
1310 			char *path;
1311 			if (asprintf(&path, "%s/%s/%s",
1312 			    got_repo_get_path_git_dir(repo),
1313 			    GOT_OBJECTS_PACK_DIR,
1314 			    dent->d_name) == -1) {
1315 				err = got_error_from_errno("asprintf");
1316 				goto done;
1317 			}
1318 			err = progress_cb(progress_arg, path);
1319 			free(path);
1320 			if (err)
1321 				goto done;
1322 		}
1323 		free(pack_relpath);
1324 		pack_relpath = NULL;
1325 	}
1326 done:
1327 	if (packdir && closedir(packdir) != 0 && err == NULL)
1328 		err = got_error_from_errno("closedir");
1329 	free(pack_relpath);
1330 	return err;
1331 }