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 
489 	memset(&id, 0, sizeof(id));
490 	id.algo = repo->algo;
491 	if (!got_parse_hash_digest(id.hash, p, repo->algo)) {
492 		err = got_error_fmt(GOT_ERR_BAD_PATH,
493 		    "'%s' is not a valid pack file name",
494 		    packfile_name);
495 		goto done;
496 	}
497 
498 	*pack_hash = got_object_id_dup(&id);
499 	if (*pack_hash == NULL) {
500 		err = got_error_from_errno("got_object_id_dup");
501 		goto done;
502 	}
503 
504 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
505 	if (packfd == -1) {
506 		err = got_error_from_errno2("open", packfile_path);
507 		goto done;
508 	}
509 
510 	*packfile = fdopen(packfd, "r");
511 	if (*packfile == NULL) {
512 		err = got_error_from_errno2("fdopen", packfile_path);
513 		goto done;
514 	}
515 	packfd = -1;
516 done:
517 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
518 		err = got_error_from_errno2("close", packfile_path);
519 	free(packfile_name);
520 	if (err) {
521 		free(*pack_hash);
522 		*pack_hash = NULL;
523 	}
524 	return err;
525 }
526 
527 const struct got_error *
528 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
529     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
530     got_cancel_cb cancel_cb, void *cancel_arg)
531 {
532 	const struct got_error *err = NULL;
533 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
534 	struct got_packidx *packidx = NULL;
535 	struct got_pack *pack = NULL;
536 	size_t idlen;
537 	uint32_t nobj, i;
538 
539 	idlen = got_hash_digest_length(repo->algo);
540 
541 	err = got_object_id_str(&id_str, pack_hash);
542 	if (err)
543 		goto done;
544 
545 	if (asprintf(&packpath, "%s/pack-%s.pack",
546 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
547 		err = got_error_from_errno("asprintf");
548 		goto done;
549 	}
550 	if (asprintf(&idxpath, "%s/pack-%s.idx",
551 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
552 		err = got_error_from_errno("asprintf");
553 		goto done;
554 	}
555 
556 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1,
557 	    repo->algo);
558 	if (err)
559 		goto done;
560 
561 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
562 	if (err)
563 		goto done;
564 
565 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
566 	for (i = 0; i < nobj; i++) {
567 		struct got_packidx_object_id *oid;
568 		struct got_object_id id, base_id;
569 		off_t offset, base_offset = 0;
570 		uint8_t type;
571 		uint64_t size;
572 		size_t tslen, len;
573 
574 		if (cancel_cb) {
575 			err = cancel_cb(cancel_arg);
576 			if (err)
577 				break;
578 		}
579 		oid = packidx->hdr.sorted_ids + i * idlen;
580 		memset(&id, 0, sizeof(id));
581 		id.algo = repo->algo;
582 		memcpy(id.hash, oid->hash, idlen);
583 
584 		offset = got_packidx_get_object_offset(packidx, i);
585 		if (offset == -1) {
586 			err = got_error(GOT_ERR_BAD_PACKIDX);
587 			goto done;
588 		}
589 
590 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
591 		    pack, offset);
592 		if (err)
593 			goto done;
594 
595 		switch (type) {
596 		case GOT_OBJ_TYPE_OFFSET_DELTA:
597 			err = got_pack_parse_offset_delta(&base_offset, &len,
598 			    pack, offset, tslen);
599 			if (err)
600 				goto done;
601 			break;
602 		case GOT_OBJ_TYPE_REF_DELTA:
603 			err = got_pack_parse_ref_delta(&base_id,
604 			    pack, offset, tslen);
605 			if (err)
606 				goto done;
607 			break;
608 		}
609 		err = (*list_cb)(list_arg, &id, type, offset, size,
610 		    base_offset, &base_id);
611 		if (err)
612 			goto done;
613 	}
614 
615 done:
616 	free(id_str);
617 	free(idxpath);
618 	free(packpath);
619 	if (packidx)
620 		got_packidx_close(packidx);
621 	return err;
622 }
623 
624 static const struct got_error *
625 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
626     void *progress_arg, struct got_ratelimit *rl,
627     int nloose, int ncommits, int npurged)
628 {
629 	const struct got_error *err;
630 	int elapsed;
631 
632 	if (progress_cb == NULL)
633 		return NULL;
634 
635 	err = got_ratelimit_check(&elapsed, rl);
636 	if (err || !elapsed)
637 		return err;
638 
639 	return progress_cb(progress_arg, nloose, ncommits, npurged);
640 }
641 
642 static const struct got_error *
643 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
644     got_cleanup_progress_cb progress_cb, void *progress_arg,
645     struct got_ratelimit *rl, struct got_repository *repo)
646 {
647 	const struct got_error *err = NULL;
648 	char *path_objects = NULL, *path = NULL;
649 	DIR *dir = NULL;
650 	struct got_object *obj = NULL;
651 	struct got_object_id id;
652 	int i, fd = -1;
653 	struct stat sb;
654 
655 	*ondisk_size = 0;
656 	*loose_ids = got_object_idset_alloc();
657 	if (*loose_ids == NULL)
658 		return got_error_from_errno("got_object_idset_alloc");
659 
660 	path_objects = got_repo_get_path_objects(repo);
661 	if (path_objects == NULL) {
662 		err = got_error_from_errno("got_repo_get_path_objects");
663 		goto done;
664 	}
665 
666 	for (i = 0; i <= 0xff; i++) {
667 		struct dirent *dent;
668 
669 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
670 			err = got_error_from_errno("asprintf");
671 			break;
672 		}
673 
674 		dir = opendir(path);
675 		if (dir == NULL) {
676 			if (errno == ENOENT) {
677 				err = NULL;
678 				continue;
679 			}
680 			err = got_error_from_errno2("opendir", path);
681 			break;
682 		}
683 
684 		while ((dent = readdir(dir)) != NULL) {
685 			char *id_str;
686 
687 			if (strcmp(dent->d_name, ".") == 0 ||
688 			    strcmp(dent->d_name, "..") == 0)
689 				continue;
690 
691 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
692 				err = got_error_from_errno("asprintf");
693 				goto done;
694 			}
695 
696 			memset(&id, 0, sizeof(id));
697 			id.algo = repo->algo;
698 			if (!got_parse_hash_digest(id.hash, id_str,
699 			    repo->algo)) {
700 				free(id_str);
701 				continue;
702 			}
703 			free(id_str);
704 
705 			err = got_object_open_loose_fd(&fd, &id, repo);
706 			if (err)
707 				goto done;
708 			if (fstat(fd, &sb) == -1) {
709 				err = got_error_from_errno("fstat");
710 				goto done;
711 			}
712 			err = got_object_read_header_privsep(&obj, &id, repo,
713 			    fd);
714 			if (err)
715 				goto done;
716 			fd = -1; /* already closed */
717 
718 			switch (obj->type) {
719 			case GOT_OBJ_TYPE_COMMIT:
720 			case GOT_OBJ_TYPE_TREE:
721 			case GOT_OBJ_TYPE_BLOB:
722 			case GOT_OBJ_TYPE_TAG:
723 				break;
724 			default:
725 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
726 				    "%d", obj->type);
727 				goto done;
728 			}
729 			got_object_close(obj);
730 			obj = NULL;
731 			(*ondisk_size) += sb.st_size;
732 			err = got_object_idset_add(*loose_ids, &id, NULL);
733 			if (err)
734 				goto done;
735 			err = report_cleanup_progress(progress_cb,
736 			    progress_arg, rl,
737 			    got_object_idset_num_elements(*loose_ids), -1, -1);
738 			if (err)
739 				goto done;
740 		}
741 
742 		if (closedir(dir) != 0) {
743 			err = got_error_from_errno("closedir");
744 			goto done;
745 		}
746 		dir = NULL;
747 
748 		free(path);
749 		path = NULL;
750 	}
751 done:
752 	if (dir && closedir(dir) != 0 && err == NULL)
753 		err = got_error_from_errno("closedir");
754 	if (fd != -1 && close(fd) == -1 && err == NULL)
755 		err = got_error_from_errno("close");
756 	if (err) {
757 		got_object_idset_free(*loose_ids);
758 		*loose_ids = NULL;
759 	}
760 	if (obj)
761 		got_object_close(obj);
762 	free(path_objects);
763 	free(path);
764 	return err;
765 }
766 
767 static const struct got_error *
768 preserve_loose_object(struct got_object_idset *loose_ids,
769     struct got_object_id *id, struct got_repository *repo, int *npacked)
770 {
771 	const struct got_error *err = NULL;
772 	struct got_object *obj;
773 
774 	if (!got_object_idset_contains(loose_ids, id))
775 		return NULL;
776 
777 	/*
778 	 * Try to open this object from a pack file. This ensures that
779 	 * we do in fact have a valid packed copy of the object. Otherwise
780 	 * we should not delete the loose representation of this object.
781 	 */
782 	err = got_object_open_packed(&obj, id, repo);
783 	if (err == NULL) {
784 		got_object_close(obj);
785 		/*
786 		 * The object is referenced and packed.
787 		 * We can purge the redundantly stored loose object.
788 		 */
789 		(*npacked)++;
790 		return NULL;
791 	} else if (err->code != GOT_ERR_NO_OBJ)
792 		return err;
793 
794 	/*
795 	 * This object is referenced and not packed.
796 	 * Remove it from our purge set.
797 	 */
798 	return got_object_idset_remove(NULL, loose_ids, id);
799 }
800 
801 static const struct got_error *
802 load_tree_entries(struct got_object_id_queue *ids,
803     struct got_object_idset *loose_ids,
804     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
805     const char *dpath, struct got_repository *repo, int *npacked,
806     got_cancel_cb cancel_cb, void *cancel_arg)
807 {
808 	const struct got_error *err;
809 	struct got_tree_object *tree;
810 	char *p = NULL;
811 	int i;
812 
813 	err = got_object_open_as_tree(&tree, repo, tree_id);
814 	if (err)
815 		return err;
816 
817 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
818 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
819 		struct got_object_id *id = got_tree_entry_get_id(e);
820 		mode_t mode = got_tree_entry_get_mode(e);
821 
822 		if (cancel_cb) {
823 			err = (*cancel_cb)(cancel_arg);
824 			if (err)
825 				break;
826 		}
827 
828 		if (got_object_tree_entry_is_symlink(e) ||
829 		    got_object_tree_entry_is_submodule(e) ||
830 		    got_object_idset_contains(traversed_ids, id))
831 			continue;
832 
833 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
834 		    got_tree_entry_get_name(e)) == -1) {
835 			err = got_error_from_errno("asprintf");
836 			break;
837 		}
838 
839 		if (S_ISDIR(mode)) {
840 			struct got_object_qid *qid;
841 			err = got_object_qid_alloc(&qid, id);
842 			if (err)
843 				break;
844 			STAILQ_INSERT_TAIL(ids, qid, entry);
845 		} else if (S_ISREG(mode)) {
846 			/* This blob is referenced. */
847 			err = preserve_loose_object(loose_ids, id, repo,
848 			    npacked);
849 			if (err)
850 				break;
851 			err = got_object_idset_add(traversed_ids, id, NULL);
852 			if (err)
853 				break;
854 		}
855 		free(p);
856 		p = NULL;
857 	}
858 
859 	got_object_tree_close(tree);
860 	free(p);
861 	return err;
862 }
863 
864 static const struct got_error *
865 load_tree(struct got_object_idset *loose_ids,
866     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
867     const char *dpath, struct got_repository *repo, int *npacked,
868     got_cancel_cb cancel_cb, void *cancel_arg)
869 {
870 	const struct got_error *err = NULL;
871 	struct got_object_id_queue tree_ids;
872 	struct got_object_qid *qid;
873 
874 	err = got_object_qid_alloc(&qid, tree_id);
875 	if (err)
876 		return err;
877 
878 	STAILQ_INIT(&tree_ids);
879 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
880 
881 	while (!STAILQ_EMPTY(&tree_ids)) {
882 		if (cancel_cb) {
883 			err = (*cancel_cb)(cancel_arg);
884 			if (err)
885 				break;
886 		}
887 
888 		qid = STAILQ_FIRST(&tree_ids);
889 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
890 
891 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
892 			got_object_qid_free(qid);
893 			continue;
894 		}
895 
896 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
897 		if (err) {
898 			got_object_qid_free(qid);
899 			break;
900 		}
901 
902 		/* This tree is referenced. */
903 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
904 		if (err)
905 			break;
906 
907 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
908 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
909 		got_object_qid_free(qid);
910 		if (err)
911 			break;
912 	}
913 
914 	got_object_id_queue_free(&tree_ids);
915 	return err;
916 }
917 
918 static const struct got_error *
919 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
920     int *npacked, struct got_object_idset *traversed_ids,
921     struct got_object_id *id, struct got_repository *repo,
922     got_cleanup_progress_cb progress_cb, void *progress_arg,
923     struct got_ratelimit *rl, int nloose,
924     got_cancel_cb cancel_cb, void *cancel_arg)
925 {
926 	const struct got_error *err;
927 	struct got_commit_object *commit = NULL;
928 	struct got_tag_object *tag = NULL;
929 	struct got_object_id *tree_id = NULL;
930 	struct got_object_id_queue ids;
931 	struct got_object_qid *qid;
932 	int obj_type;
933 
934 	err = got_object_qid_alloc(&qid, id);
935 	if (err)
936 		return err;
937 
938 	STAILQ_INIT(&ids);
939 	STAILQ_INSERT_TAIL(&ids, qid, entry);
940 
941 	while (!STAILQ_EMPTY(&ids)) {
942 		if (cancel_cb) {
943 			err = (*cancel_cb)(cancel_arg);
944 			if (err)
945 				break;
946 		}
947 
948 		qid = STAILQ_FIRST(&ids);
949 		STAILQ_REMOVE_HEAD(&ids, entry);
950 
951 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
952 			got_object_qid_free(qid);
953 			qid = NULL;
954 			continue;
955 		}
956 
957 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
958 		if (err)
959 			break;
960 
961 		/* This commit or tag is referenced. */
962 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
963 		if (err)
964 			break;
965 
966 		err = got_object_get_type(&obj_type, repo, &qid->id);
967 		if (err)
968 			break;
969 		switch (obj_type) {
970 		case GOT_OBJ_TYPE_COMMIT:
971 			err = got_object_open_as_commit(&commit, repo,
972 			    &qid->id);
973 			if (err)
974 				goto done;
975 			break;
976 		case GOT_OBJ_TYPE_TAG:
977 			err = got_object_open_as_tag(&tag, repo, &qid->id);
978 			if (err)
979 				goto done;
980 			break;
981 		default:
982 			/* should not happen */
983 			err = got_error(GOT_ERR_OBJ_TYPE);
984 			goto done;
985 		}
986 
987 		/* Find a tree object to scan. */
988 		if (commit) {
989 			tree_id = got_object_commit_get_tree_id(commit);
990 		} else if (tag) {
991 			obj_type = got_object_tag_get_object_type(tag);
992 			switch (obj_type) {
993 			case GOT_OBJ_TYPE_COMMIT:
994 				err = got_object_open_as_commit(&commit, repo,
995 				    got_object_tag_get_object_id(tag));
996 				if (err)
997 					goto done;
998 				tree_id = got_object_commit_get_tree_id(commit);
999 				break;
1000 			case GOT_OBJ_TYPE_TREE:
1001 				tree_id = got_object_tag_get_object_id(tag);
1002 				break;
1003 			default:
1004 				/*
1005 				 * Tag points at something other than a
1006 				 * commit or tree. Leave this weird tag object
1007 				 * and the object it points to on disk.
1008 				 */
1009 				err = got_object_idset_remove(NULL, loose_ids,
1010 				    &qid->id);
1011 				if (err && err->code != GOT_ERR_NO_OBJ)
1012 					goto done;
1013 				err = got_object_idset_remove(NULL, loose_ids,
1014 				    got_object_tag_get_object_id(tag));
1015 				if (err && err->code != GOT_ERR_NO_OBJ)
1016 					goto done;
1017 				err = NULL;
1018 				break;
1019 			}
1020 		}
1021 
1022 		if (tree_id) {
1023 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1024 			    repo, npacked, cancel_cb, cancel_arg);
1025 			if (err)
1026 				break;
1027 		}
1028 
1029 		if (commit || tag)
1030 			(*ncommits)++; /* scanned tags are counted as commits */
1031 
1032 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1033 		    nloose, *ncommits, -1);
1034 		if (err)
1035 			break;
1036 
1037 		if (commit) {
1038 			/* Find parent commits to scan. */
1039 			const struct got_object_id_queue *parent_ids;
1040 			parent_ids = got_object_commit_get_parent_ids(commit);
1041 			err = got_object_id_queue_copy(parent_ids, &ids);
1042 			if (err)
1043 				break;
1044 			got_object_commit_close(commit);
1045 			commit = NULL;
1046 		}
1047 		if (tag) {
1048 			got_object_tag_close(tag);
1049 			tag = NULL;
1050 		}
1051 		got_object_qid_free(qid);
1052 		qid = NULL;
1053 	}
1054 done:
1055 	if (qid)
1056 		got_object_qid_free(qid);
1057 	if (commit)
1058 		got_object_commit_close(commit);
1059 	if (tag)
1060 		got_object_tag_close(tag);
1061 	got_object_id_queue_free(&ids);
1062 	return err;
1063 }
1064 
1065 struct purge_loose_object_arg {
1066 	struct got_repository *repo;
1067 	got_cleanup_progress_cb progress_cb;
1068 	void *progress_arg;
1069 	struct got_ratelimit *rl;
1070 	int nloose;
1071 	int ncommits;
1072 	int npurged;
1073 	off_t size_purged;
1074 	int dry_run;
1075 	time_t max_mtime;
1076 	int ignore_mtime;
1077 };
1078 
1079 static const struct got_error *
1080 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1081 {
1082 	struct purge_loose_object_arg *a = arg;
1083 	const struct got_error *err, *unlock_err = NULL;
1084 	char *path = NULL;
1085 	int fd = -1;
1086 	struct stat sb;
1087 	struct got_lockfile *lf = NULL;
1088 
1089 	err = got_object_get_path(&path, id, a->repo);
1090 	if (err)
1091 		return err;
1092 
1093 	err = got_object_open_loose_fd(&fd, id, a->repo);
1094 	if (err)
1095 		goto done;
1096 
1097 	if (fstat(fd, &sb) == -1) {
1098 		err = got_error_from_errno("fstat");
1099 		goto done;
1100 	}
1101 
1102 	/*
1103 	 * Do not delete objects which are younger than our maximum
1104 	 * modification time threshold. This prevents a race where
1105 	 * new objects which are being added to the repository
1106 	 * concurrently would be deleted.
1107 	 */
1108 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1109 		if (!a->dry_run) {
1110 			err = got_lockfile_lock(&lf, path, -1);
1111 			if (err)
1112 				goto done;
1113 			if (unlink(path) == -1) {
1114 				err = got_error_from_errno2("unlink", path);
1115 				goto done;
1116 			}
1117 		}
1118 
1119 		a->npurged++;
1120 		a->size_purged += sb.st_size;
1121 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1122 		    a->rl, a->nloose, a->ncommits, a->npurged);
1123 		if (err)
1124 			goto done;
1125 	}
1126 done:
1127 	if (fd != -1 && close(fd) == -1 && err == NULL)
1128 		err = got_error_from_errno("close");
1129 	free(path);
1130 	if (lf)
1131 		unlock_err = got_lockfile_unlock(lf, -1);
1132 	return err ? err : unlock_err;
1133 }
1134 
1135 const struct got_error *
1136 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1137     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1138     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1139     got_cancel_cb cancel_cb, void *cancel_arg)
1140 {
1141 	const struct got_error *err;
1142 	struct got_object_idset *loose_ids;
1143 	struct got_object_idset *traversed_ids;
1144 	struct got_object_id **referenced_ids;
1145 	int i, nreferenced, nloose, ncommits = 0;
1146 	struct got_reflist_head refs;
1147 	struct got_reflist_entry *re;
1148 	struct purge_loose_object_arg arg;
1149 	time_t max_mtime = 0;
1150 	struct got_ratelimit rl;
1151 
1152 	TAILQ_INIT(&refs);
1153 	got_ratelimit_init(&rl, 0, 500);
1154 
1155 	*size_before = 0;
1156 	*size_after = 0;
1157 	*npacked = 0;
1158 
1159 	err = get_loose_object_ids(&loose_ids, size_before,
1160 	    progress_cb, progress_arg, &rl, repo);
1161 	if (err)
1162 		return err;
1163 	nloose = got_object_idset_num_elements(loose_ids);
1164 	if (nloose == 0) {
1165 		got_object_idset_free(loose_ids);
1166 		if (progress_cb) {
1167 			err = progress_cb(progress_arg, 0, 0, 0);
1168 			if (err)
1169 				return err;
1170 		}
1171 		return NULL;
1172 	}
1173 
1174 	traversed_ids = got_object_idset_alloc();
1175 	if (traversed_ids == NULL) {
1176 		err = got_error_from_errno("got_object_idset_alloc");
1177 		goto done;
1178 	}
1179 
1180 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1181 	if (err)
1182 		goto done;
1183 	if (!ignore_mtime) {
1184 		TAILQ_FOREACH(re, &refs, entry) {
1185 			time_t mtime = got_ref_get_mtime(re->ref);
1186 			if (mtime > max_mtime)
1187 				max_mtime = mtime;
1188 		}
1189 		/*
1190 		 * For safety, keep objects created within 10 minutes
1191 		 * before the youngest reference was created.
1192 		 */
1193 		if (max_mtime >= 600)
1194 			max_mtime -= 600;
1195 	}
1196 
1197 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1198 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1199 	    &refs, repo, cancel_cb, cancel_arg);
1200 	if (err)
1201 		goto done;
1202 
1203 	for (i = 0; i < nreferenced; i++) {
1204 		struct got_object_id *id = referenced_ids[i];
1205 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1206 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1207 		    nloose, cancel_cb, cancel_arg);
1208 		if (err)
1209 			goto done;
1210 	}
1211 
1212 	/* Any remaining loose objects are unreferenced and can be purged. */
1213 	arg.repo = repo;
1214 	arg.progress_arg = progress_arg;
1215 	arg.progress_cb = progress_cb;
1216 	arg.rl = &rl;
1217 	arg.nloose = nloose;
1218 	arg.npurged = 0;
1219 	arg.size_purged = 0;
1220 	arg.ncommits = ncommits;
1221 	arg.dry_run = dry_run;
1222 	arg.max_mtime = max_mtime;
1223 	arg.ignore_mtime = ignore_mtime;
1224 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1225 	if (err)
1226 		goto done;
1227 	*size_after = *size_before - arg.size_purged;
1228 
1229 	/* Produce a final progress report. */
1230 	if (progress_cb) {
1231 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1232 		if (err)
1233 			goto done;
1234 	}
1235 done:
1236 	got_object_idset_free(loose_ids);
1237 	got_object_idset_free(traversed_ids);
1238 	return err;
1239 }
1240 
1241 static const struct got_error *
1242 remove_packidx(int dir_fd, const char *relpath)
1243 {
1244 	const struct got_error *err, *unlock_err;
1245 	struct got_lockfile *lf;
1246 
1247 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1248 	if (err)
1249 		return err;
1250 	if (unlinkat(dir_fd, relpath, 0) == -1)
1251 		err = got_error_from_errno("unlinkat");
1252 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1253 	return err ? err : unlock_err;
1254 }
1255 
1256 const struct got_error *
1257 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1258     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1259     got_cancel_cb cancel_cb, void *cancel_arg)
1260 {
1261 	const struct got_error *err = NULL;
1262 	DIR *packdir = NULL;
1263 	struct dirent *dent;
1264 	char *pack_relpath = NULL;
1265 	int packdir_fd;
1266 	struct stat sb;
1267 
1268 	packdir_fd = openat(got_repo_get_fd(repo),
1269 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1270 	if (packdir_fd == -1) {
1271 		if (errno == ENOENT)
1272 			return NULL;
1273 		return got_error_from_errno_fmt("openat: %s/%s",
1274 		    got_repo_get_path_git_dir(repo),
1275 		    GOT_OBJECTS_PACK_DIR);
1276 	}
1277 
1278 	packdir = fdopendir(packdir_fd);
1279 	if (packdir == NULL) {
1280 		err = got_error_from_errno("fdopendir");
1281 		goto done;
1282 	}
1283 
1284 	while ((dent = readdir(packdir)) != NULL) {
1285 		if (cancel_cb) {
1286 			err = cancel_cb(cancel_arg);
1287 			if (err)
1288 				goto done;
1289 		}
1290 
1291 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen,
1292 		    got_repo_get_object_format(repo)))
1293 			continue;
1294 
1295 		err = got_packidx_get_packfile_path(&pack_relpath,
1296 		    dent->d_name);
1297 		if (err)
1298 			goto done;
1299 
1300 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1301 			free(pack_relpath);
1302 			pack_relpath = NULL;
1303 			continue;
1304 		}
1305 		if (errno != ENOENT) {
1306 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1307 			    got_repo_get_path_git_dir(repo),
1308 			    GOT_OBJECTS_PACK_DIR,
1309 			    pack_relpath);
1310 			goto done;
1311 		}
1312 
1313 		if (!dry_run) {
1314 			err = remove_packidx(packdir_fd, dent->d_name);
1315 			if (err)
1316 				goto done;
1317 		}
1318 		if (progress_cb) {
1319 			char *path;
1320 			if (asprintf(&path, "%s/%s/%s",
1321 			    got_repo_get_path_git_dir(repo),
1322 			    GOT_OBJECTS_PACK_DIR,
1323 			    dent->d_name) == -1) {
1324 				err = got_error_from_errno("asprintf");
1325 				goto done;
1326 			}
1327 			err = progress_cb(progress_arg, path);
1328 			free(path);
1329 			if (err)
1330 				goto done;
1331 		}
1332 		free(pack_relpath);
1333 		pack_relpath = NULL;
1334 	}
1335 done:
1336 	if (packdir && closedir(packdir) != 0 && err == NULL)
1337 		err = got_error_from_errno("closedir");
1338 	free(pack_relpath);
1339 	return err;
1340 }