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/uio.h>
20 #include <sys/stat.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 
24 #include <dirent.h>
25 #include <errno.h>
26 #include <fcntl.h>
27 #include <stdint.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <limits.h>
32 #include <unistd.h>
33 
34 #include "got_error.h"
35 #include "got_cancel.h"
36 #include "got_object.h"
37 #include "got_reference.h"
38 #include "got_repository.h"
39 #include "got_repository_admin.h"
40 #include "got_opentemp.h"
41 #include "got_path.h"
42 
43 #include "got_lib_delta.h"
44 #include "got_lib_object.h"
45 #include "got_lib_object_idset.h"
46 #include "got_lib_object_cache.h"
47 #include "got_lib_pack.h"
48 #include "got_lib_privsep.h"
49 #include "got_lib_repository.h"
50 #include "got_lib_pack_create.h"
51 #include "got_lib_sha1.h"
52 #include "got_lib_lockfile.h"
53 #include "got_lib_ratelimit.h"
54 
55 #ifndef nitems
56 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
57 #endif
58 
59 static const struct got_error *
60 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
61     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
62     struct got_repository *repo,
63     got_cancel_cb cancel_cb, void *cancel_arg)
64 {
65 	const struct got_error *err = NULL;
66 	const size_t alloc_chunksz = 256;
67 	size_t nalloc;
68 	struct got_reflist_entry *re;
69 	int i;
70 
71 	*ids = NULL;
72 	*nobjects = 0;
73 
74 	err = got_reflist_sort(refs,
75 	    got_ref_cmp_by_commit_timestamp_descending, repo);
76 	if (err)
77 		return err;
78 
79 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
80 	if (*ids == NULL)
81 		return got_error_from_errno("reallocarray");
82 	nalloc = alloc_chunksz;
83 
84 	TAILQ_FOREACH(re, refs, entry) {
85 		struct got_object_id *id;
86 
87 		if (cancel_cb) {
88 			err = cancel_cb(cancel_arg);
89 			if (err)
90 				goto done;
91 		}
92 
93 		err = got_ref_resolve(&id, repo, re->ref);
94 		if (err)
95 			goto done;
96 
97 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
98 			int obj_type;
99 			err = got_object_get_type(&obj_type, repo, id);
100 			if (err)
101 				goto done;
102 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
103 				free(id);
104 				id = NULL;
105 				continue;
106 			}
107 		}
108 
109 		if (nalloc <= *nobjects) {
110 			struct got_object_id **new;
111 			new = recallocarray(*ids, nalloc,
112 			    nalloc + alloc_chunksz,
113 			    sizeof(struct got_object_id *));
114 			if (new == NULL) {
115 				err = got_error_from_errno(
116 				    "recallocarray");
117 				goto done;
118 			}
119 			*ids = new;
120 			nalloc += alloc_chunksz;
121 		}
122 		(*ids)[*nobjects] = id;
123 		if ((*ids)[*nobjects] == NULL) {
124 			err = got_error_from_errno("got_object_id_dup");
125 			goto done;
126 		}
127 		(*nobjects)++;
128 	}
129 done:
130 	if (err) {
131 		for (i = 0; i < *nobjects; i++)
132 			free((*ids)[i]);
133 		free(*ids);
134 		*ids = NULL;
135 		*nobjects = 0;
136 	}
137 	return err;
138 }
139 
140 const struct got_error *
141 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
142     struct got_reflist_head *include_refs,
143     struct got_reflist_head *exclude_refs, struct got_repository *repo,
144     int loose_obj_only,
145     got_pack_progress_cb progress_cb, void *progress_arg,
146     got_cancel_cb cancel_cb, void *cancel_arg)
147 {
148 	const struct got_error *err = NULL;
149 	struct got_object_id **ours = NULL, **theirs = NULL;
150 	int nours = 0, ntheirs = 0, packfd = -1, i;
151 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
152 	char *sha1_str = NULL;
153 
154 	*packfile = NULL;
155 	*pack_hash = NULL;
156 
157 	if (asprintf(&path, "%s/%s/packing.pack",
158 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
159 		err = got_error_from_errno("asprintf");
160 		goto done;
161 	}
162 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
163 	if (err)
164 		goto done;
165 
166 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
167 		err = got_error_from_errno2("fchmod", tmpfile_path);
168 		goto done;
169 	}
170 
171 	*packfile = fdopen(packfd, "w");
172 	if (*packfile == NULL) {
173 		err = got_error_from_errno2("fdopen", tmpfile_path);
174 		goto done;
175 	}
176 	packfd = -1;
177 
178 	err = get_reflist_object_ids(&ours, &nours,
179 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
180 	    include_refs, repo, cancel_cb, cancel_arg);
181 	if (err)
182 		goto done;
183 
184 	if (nours == 0) {
185 		err = got_error(GOT_ERR_CANNOT_PACK);
186 		goto done;
187 	}
188 
189 	if (!TAILQ_EMPTY(exclude_refs)) {
190 		err = get_reflist_object_ids(&theirs, &ntheirs,
191 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
192 		    exclude_refs, repo,
193 		    cancel_cb, cancel_arg);
194 		if (err)
195 			goto done;
196 	}
197 
198 	*pack_hash = calloc(1, sizeof(**pack_hash));
199 	if (*pack_hash == NULL) {
200 		err = got_error_from_errno("calloc");
201 		goto done;
202 	}
203 
204 	err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
205 	    ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
206 	    cancel_cb, cancel_arg);
207 	if (err)
208 		goto done;
209 
210 	err = got_object_id_str(&sha1_str, *pack_hash);
211 	if (err)
212 		goto done;
213 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
214 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
215 	    sha1_str) == -1) {
216 		err = got_error_from_errno("asprintf");
217 		goto done;
218 	}
219 
220 	if (fflush(*packfile) == -1) {
221 		err = got_error_from_errno("fflush");
222 		goto done;
223 	}
224 	if (fseek(*packfile, 0L, SEEK_SET) == -1) {
225 		err = got_error_from_errno("fseek");
226 		goto done;
227 	}
228 	if (rename(tmpfile_path, packfile_path) == -1) {
229 		err = got_error_from_errno3("rename", tmpfile_path,
230 		    packfile_path);
231 		goto done;
232 	}
233 	free(tmpfile_path);
234 	tmpfile_path = NULL;
235 done:
236 	for (i = 0; i < nours; i++)
237 		free(ours[i]);
238 	free(ours);
239 	for (i = 0; i < ntheirs; i++)
240 		free(theirs[i]);
241 	free(theirs);
242 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
243 		err = got_error_from_errno2("close", packfile_path);
244 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
245 		err = got_error_from_errno2("unlink", tmpfile_path);
246 	free(tmpfile_path);
247 	free(packfile_path);
248 	free(sha1_str);
249 	free(path);
250 	if (err) {
251 		free(*pack_hash);
252 		*pack_hash = NULL;
253 		if (*packfile)
254 			fclose(*packfile);
255 		*packfile = NULL;
256 	}
257 	return err;
258 }
259 
260 const struct got_error *
261 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
262     struct got_repository *repo,
263     got_pack_index_progress_cb progress_cb, void *progress_arg,
264     got_cancel_cb cancel_cb, void *cancel_arg)
265 {
266 	size_t i;
267 	char *path;
268 	int imsg_idxfds[2];
269 	int npackfd = -1, idxfd = -1, nidxfd = -1;
270 	int tmpfds[3];
271 	int idxstatus, done = 0;
272 	const struct got_error *err;
273 	struct imsgbuf idxibuf;
274 	pid_t idxpid;
275 	char *tmpidxpath = NULL;
276 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
277 	const char *repo_path = got_repo_get_path_git_dir(repo);
278 	struct stat sb;
279 
280 	for (i = 0; i < nitems(tmpfds); i++)
281 		tmpfds[i] = -1;
282 
283 	if (asprintf(&path, "%s/%s/indexing.idx",
284 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
285 		err = got_error_from_errno("asprintf");
286 		goto done;
287 	}
288 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
289 	free(path);
290 	if (err)
291 		goto done;
292 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
293 		err = got_error_from_errno2("fchmod", tmpidxpath);
294 		goto done;
295 	}
296 
297 	nidxfd = dup(idxfd);
298 	if (nidxfd == -1) {
299 		err = got_error_from_errno("dup");
300 		goto done;
301 	}
302 
303 	for (i = 0; i < nitems(tmpfds); i++) {
304 		tmpfds[i] = got_opentempfd();
305 		if (tmpfds[i] == -1) {
306 			err = got_error_from_errno("got_opentempfd");
307 			goto done;
308 		}
309 	}
310 
311 	err = got_object_id_str(&id_str, pack_hash);
312 	if (err)
313 		goto done;
314 
315 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
316 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
317 		err = got_error_from_errno("asprintf");
318 		goto done;
319 	}
320 
321 	if (fstat(fileno(packfile), &sb) == -1) {
322 		err = got_error_from_errno2("fstat", packfile_path);
323 		goto done;
324 	}
325 
326 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
327 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
328 		err = got_error_from_errno("asprintf");
329 		goto done;
330 	}
331 
332 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
333 		err = got_error_from_errno("socketpair");
334 		goto done;
335 	}
336 	idxpid = fork();
337 	if (idxpid == -1) {
338 		err= got_error_from_errno("fork");
339 		goto done;
340 	} else if (idxpid == 0)
341 		got_privsep_exec_child(imsg_idxfds,
342 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
343 	if (close(imsg_idxfds[1]) == -1) {
344 		err = got_error_from_errno("close");
345 		goto done;
346 	}
347 	imsg_init(&idxibuf, imsg_idxfds[0]);
348 
349 	npackfd = dup(fileno(packfile));
350 	if (npackfd == -1) {
351 		err = got_error_from_errno("dup");
352 		goto done;
353 	}
354 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
355 	    npackfd);
356 	if (err != NULL)
357 		goto done;
358 	npackfd = -1;
359 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
360 	if (err != NULL)
361 		goto done;
362 	nidxfd = -1;
363 	for (i = 0; i < nitems(tmpfds); i++) {
364 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
365 		if (err != NULL)
366 			goto done;
367 		tmpfds[i] = -1;
368 	}
369 	done = 0;
370 	while (!done) {
371 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
372 
373 		if (cancel_cb) {
374 			err = cancel_cb(cancel_arg);
375 			if (err)
376 				goto done;
377 		}
378 
379 		err = got_privsep_recv_index_progress(&done, &nobj_total,
380 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
381 		    &idxibuf);
382 		if (err != NULL)
383 			goto done;
384 		if (nobj_indexed != 0) {
385 			err = progress_cb(progress_arg, sb.st_size,
386 			    nobj_total, nobj_indexed, nobj_loose,
387 			    nobj_resolved);
388 			if (err)
389 				break;
390 		}
391 	}
392 	if (close(imsg_idxfds[0]) == -1) {
393 		err = got_error_from_errno("close");
394 		goto done;
395 	}
396 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
397 		err = got_error_from_errno("waitpid");
398 		goto done;
399 	}
400 
401 	if (rename(tmpidxpath, idxpath) == -1) {
402 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
403 		goto done;
404 	}
405 	free(tmpidxpath);
406 	tmpidxpath = NULL;
407 
408 done:
409 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
410 		err = got_error_from_errno2("unlink", tmpidxpath);
411 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
412 		err = got_error_from_errno("close");
413 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
414 		err = got_error_from_errno("close");
415 	for (i = 0; i < nitems(tmpfds); i++) {
416 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
417 			err = got_error_from_errno("close");
418 	}
419 	free(tmpidxpath);
420 	free(idxpath);
421 	free(packfile_path);
422 	return err;
423 }
424 
425 const struct got_error *
426 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
427     struct got_repository *repo, const char *packfile_path)
428 {
429 	const struct got_error *err = NULL;
430 	const char *packdir_path = NULL;
431 	char *packfile_name = NULL, *p, *dot;
432 	struct got_object_id id;
433 	int packfd = -1;
434 
435 	*packfile = NULL;
436 	*pack_hash = NULL;
437 
438 	packdir_path = got_repo_get_path_objects_pack(repo);
439 	if (packdir_path == NULL)
440 		return got_error_from_errno("got_repo_get_path_objects_pack");
441 
442 	if (!got_path_is_child(packfile_path, packdir_path,
443 	    strlen(packdir_path))) {
444 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
445 		goto done;
446 
447 	}
448 
449 	err = got_path_basename(&packfile_name, packfile_path);
450 	if (err)
451 		goto done;
452 	p = packfile_name;
453 
454 	if (strncmp(p, "pack-", 5) != 0) {
455 		err = got_error_fmt(GOT_ERR_BAD_PATH,
456 		    "'%s' is not a valid pack file name",
457 		    packfile_name);
458 		goto done;
459 	}
460 	p += 5;
461 	dot = strchr(p, '.');
462 	if (dot == NULL) {
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 	if (strcmp(dot + 1, "pack") != 0) {
469 		err = got_error_fmt(GOT_ERR_BAD_PATH,
470 		    "'%s' is not a valid pack file name",
471 		    packfile_name);
472 		goto done;
473 	}
474 	*dot = '\0';
475 	if (!got_parse_sha1_digest(id.sha1, p)) {
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 
482 	*pack_hash = got_object_id_dup(&id);
483 	if (*pack_hash == NULL) {
484 		err = got_error_from_errno("got_object_id_dup");
485 		goto done;
486 	}
487 
488 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
489 	if (packfd == -1) {
490 		err = got_error_from_errno2("open", packfile_path);
491 		goto done;
492 	}
493 
494 	*packfile = fdopen(packfd, "r");
495 	if (*packfile == NULL) {
496 		err = got_error_from_errno2("fdopen", packfile_path);
497 		goto done;
498 	}
499 	packfd = -1;
500 done:
501 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
502 		err = got_error_from_errno2("close", packfile_path);
503 	free(packfile_name);
504 	if (err) {
505 		free(*pack_hash);
506 		*pack_hash = NULL;
507 	}
508 	return err;
509 }
510 
511 const struct got_error *
512 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
513     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
514     got_cancel_cb cancel_cb, void *cancel_arg)
515 {
516 	const struct got_error *err = NULL;
517 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
518 	struct got_packidx *packidx = NULL;
519 	struct got_pack *pack = NULL;
520 	uint32_t nobj, i;
521 
522 	err = got_object_id_str(&id_str, pack_hash);
523 	if (err)
524 		goto done;
525 
526 	if (asprintf(&packpath, "%s/pack-%s.pack",
527 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
528 		err = got_error_from_errno("asprintf");
529 		goto done;
530 	}
531 	if (asprintf(&idxpath, "%s/pack-%s.idx",
532 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
533 		err = got_error_from_errno("asprintf");
534 		goto done;
535 	}
536 
537 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
538 	if (err)
539 		goto done;
540 
541 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
542 	if (err)
543 		goto done;
544 
545 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
546 	for (i = 0; i < nobj; i++) {
547 		struct got_packidx_object_id *oid;
548 		struct got_object_id id, base_id;
549 		off_t offset, base_offset = 0;
550 		uint8_t type;
551 		uint64_t size;
552 		size_t tslen, len;
553 
554 		if (cancel_cb) {
555 			err = cancel_cb(cancel_arg);
556 			if (err)
557 				break;
558 		}
559 		oid = &packidx->hdr.sorted_ids[i];
560 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
561 
562 		offset = got_packidx_get_object_offset(packidx, i);
563 		if (offset == -1) {
564 			err = got_error(GOT_ERR_BAD_PACKIDX);
565 			goto done;
566 		}
567 
568 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
569 		    pack, offset);
570 		if (err)
571 			goto done;
572 
573 		switch (type) {
574 		case GOT_OBJ_TYPE_OFFSET_DELTA:
575 			err = got_pack_parse_offset_delta(&base_offset, &len,
576 			    pack, offset, tslen);
577 			if (err)
578 				goto done;
579 			break;
580 		case GOT_OBJ_TYPE_REF_DELTA:
581 			err = got_pack_parse_ref_delta(&base_id,
582 			    pack, offset, tslen);
583 			if (err)
584 				goto done;
585 			break;
586 		}
587 		err = (*list_cb)(list_arg, &id, type, offset, size,
588 		    base_offset, &base_id);
589 		if (err)
590 			goto done;
591 	}
592 
593 done:
594 	free(id_str);
595 	free(idxpath);
596 	free(packpath);
597 	if (packidx)
598 		got_packidx_close(packidx);
599 	return err;
600 }
601 
602 static const struct got_error *
603 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
604     void *progress_arg, struct got_ratelimit *rl,
605     int nloose, int ncommits, int npurged)
606 {
607 	const struct got_error *err;
608 	int elapsed;
609 
610 	if (progress_cb == NULL)
611 		return NULL;
612 
613 	err = got_ratelimit_check(&elapsed, rl);
614 	if (err || !elapsed)
615 		return err;
616 
617 	return progress_cb(progress_arg, nloose, ncommits, npurged);
618 }
619 
620 static const struct got_error *
621 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
622     got_cleanup_progress_cb progress_cb, void *progress_arg,
623     struct got_ratelimit *rl, struct got_repository *repo)
624 {
625 	const struct got_error *err = NULL;
626 	char *path_objects = NULL, *path = NULL;
627 	DIR *dir = NULL;
628 	struct got_object *obj = NULL;
629 	struct got_object_id id;
630 	int i, fd = -1;
631 	struct stat sb;
632 
633 	*ondisk_size = 0;
634 	*loose_ids = got_object_idset_alloc();
635 	if (*loose_ids == NULL)
636 		return got_error_from_errno("got_object_idset_alloc");
637 
638 	path_objects = got_repo_get_path_objects(repo);
639 	if (path_objects == NULL) {
640 		err = got_error_from_errno("got_repo_get_path_objects");
641 		goto done;
642 	}
643 
644 	for (i = 0; i <= 0xff; i++) {
645 		struct dirent *dent;
646 
647 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
648 			err = got_error_from_errno("asprintf");
649 			break;
650 		}
651 
652 		dir = opendir(path);
653 		if (dir == NULL) {
654 			if (errno == ENOENT) {
655 				err = NULL;
656 				continue;
657 			}
658 			err = got_error_from_errno2("opendir", path);
659 			break;
660 		}
661 
662 		while ((dent = readdir(dir)) != NULL) {
663 			char *id_str;
664 
665 			if (strcmp(dent->d_name, ".") == 0 ||
666 			    strcmp(dent->d_name, "..") == 0)
667 				continue;
668 
669 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
670 				err = got_error_from_errno("asprintf");
671 				goto done;
672 			}
673 
674 			memset(&id, 0, sizeof(id));
675 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
676 				free(id_str);
677 				continue;
678 			}
679 			free(id_str);
680 
681 			err = got_object_open_loose_fd(&fd, &id, repo);
682 			if (err)
683 				goto done;
684 			if (fstat(fd, &sb) == -1) {
685 				err = got_error_from_errno("fstat");
686 				goto done;
687 			}
688 			err = got_object_read_header_privsep(&obj, &id, repo,
689 			    fd);
690 			if (err)
691 				goto done;
692 			fd = -1; /* already closed */
693 
694 			switch (obj->type) {
695 			case GOT_OBJ_TYPE_COMMIT:
696 			case GOT_OBJ_TYPE_TREE:
697 			case GOT_OBJ_TYPE_BLOB:
698 			case GOT_OBJ_TYPE_TAG:
699 				break;
700 			default:
701 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
702 				    "%d", obj->type);
703 				goto done;
704 			}
705 			got_object_close(obj);
706 			obj = NULL;
707 			(*ondisk_size) += sb.st_size;
708 			err = got_object_idset_add(*loose_ids, &id, NULL);
709 			if (err)
710 				goto done;
711 			err = report_cleanup_progress(progress_cb,
712 			    progress_arg, rl,
713 			    got_object_idset_num_elements(*loose_ids), -1, -1);
714 			if (err)
715 				goto done;
716 		}
717 
718 		if (closedir(dir) != 0) {
719 			err = got_error_from_errno("closedir");
720 			goto done;
721 		}
722 		dir = NULL;
723 
724 		free(path);
725 		path = NULL;
726 	}
727 done:
728 	if (dir && closedir(dir) != 0 && err == NULL)
729 		err = got_error_from_errno("closedir");
730 	if (fd != -1 && close(fd) == -1 && err == NULL)
731 		err = got_error_from_errno("close");
732 	if (err) {
733 		got_object_idset_free(*loose_ids);
734 		*loose_ids = NULL;
735 	}
736 	if (obj)
737 		got_object_close(obj);
738 	free(path_objects);
739 	free(path);
740 	return err;
741 }
742 
743 static const struct got_error *
744 preserve_loose_object(struct got_object_idset *loose_ids,
745     struct got_object_id *id, struct got_repository *repo, int *npacked)
746 {
747 	const struct got_error *err = NULL;
748 	struct got_object *obj;
749 
750 	if (!got_object_idset_contains(loose_ids, id))
751 		return NULL;
752 
753 	/*
754 	 * Try to open this object from a pack file. This ensures that
755 	 * we do in fact have a valid packed copy of the object. Otherwise
756 	 * we should not delete the loose representation of this object.
757 	 */
758 	err = got_object_open_packed(&obj, id, repo);
759 	if (err == NULL) {
760 		got_object_close(obj);
761 		/*
762 		 * The object is referenced and packed.
763 		 * We can purge the redundantly stored loose object.
764 		 */
765 		(*npacked)++;
766 		return NULL;
767 	} else if (err->code != GOT_ERR_NO_OBJ)
768 		return err;
769 
770 	/*
771 	 * This object is referenced and not packed.
772 	 * Remove it from our purge set.
773 	 */
774 	return got_object_idset_remove(NULL, loose_ids, id);
775 }
776 
777 static const struct got_error *
778 load_tree_entries(struct got_object_id_queue *ids,
779     struct got_object_idset *loose_ids,
780     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
781     const char *dpath, struct got_repository *repo, int *npacked,
782     got_cancel_cb cancel_cb, void *cancel_arg)
783 {
784 	const struct got_error *err;
785 	struct got_tree_object *tree;
786 	char *p = NULL;
787 	int i;
788 
789 	err = got_object_open_as_tree(&tree, repo, tree_id);
790 	if (err)
791 		return err;
792 
793 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
794 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
795 		struct got_object_id *id = got_tree_entry_get_id(e);
796 		mode_t mode = got_tree_entry_get_mode(e);
797 
798 		if (cancel_cb) {
799 			err = (*cancel_cb)(cancel_arg);
800 			if (err)
801 				break;
802 		}
803 
804 		if (got_object_tree_entry_is_symlink(e) ||
805 		    got_object_tree_entry_is_submodule(e) ||
806 		    got_object_idset_contains(traversed_ids, id))
807 			continue;
808 
809 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
810 		    got_tree_entry_get_name(e)) == -1) {
811 			err = got_error_from_errno("asprintf");
812 			break;
813 		}
814 
815 		if (S_ISDIR(mode)) {
816 			struct got_object_qid *qid;
817 			err = got_object_qid_alloc(&qid, id);
818 			if (err)
819 				break;
820 			STAILQ_INSERT_TAIL(ids, qid, entry);
821 		} else if (S_ISREG(mode)) {
822 			/* This blob is referenced. */
823 			err = preserve_loose_object(loose_ids, id, repo,
824 			    npacked);
825 			if (err)
826 				break;
827 			err = got_object_idset_add(traversed_ids, id, NULL);
828 			if (err)
829 				break;
830 		}
831 		free(p);
832 		p = NULL;
833 	}
834 
835 	got_object_tree_close(tree);
836 	free(p);
837 	return err;
838 }
839 
840 static const struct got_error *
841 load_tree(struct got_object_idset *loose_ids,
842     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
843     const char *dpath, struct got_repository *repo, int *npacked,
844     got_cancel_cb cancel_cb, void *cancel_arg)
845 {
846 	const struct got_error *err = NULL;
847 	struct got_object_id_queue tree_ids;
848 	struct got_object_qid *qid;
849 
850 	err = got_object_qid_alloc(&qid, tree_id);
851 	if (err)
852 		return err;
853 
854 	STAILQ_INIT(&tree_ids);
855 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
856 
857 	while (!STAILQ_EMPTY(&tree_ids)) {
858 		if (cancel_cb) {
859 			err = (*cancel_cb)(cancel_arg);
860 			if (err)
861 				break;
862 		}
863 
864 		qid = STAILQ_FIRST(&tree_ids);
865 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
866 
867 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
868 			got_object_qid_free(qid);
869 			continue;
870 		}
871 
872 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
873 		if (err) {
874 			got_object_qid_free(qid);
875 			break;
876 		}
877 
878 		/* This tree is referenced. */
879 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
880 		if (err)
881 			break;
882 
883 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
884 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
885 		got_object_qid_free(qid);
886 		if (err)
887 			break;
888 	}
889 
890 	got_object_id_queue_free(&tree_ids);
891 	return err;
892 }
893 
894 static const struct got_error *
895 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
896     int *npacked, struct got_object_idset *traversed_ids,
897     struct got_object_id *id, struct got_repository *repo,
898     got_cleanup_progress_cb progress_cb, void *progress_arg,
899     struct got_ratelimit *rl, int nloose,
900     got_cancel_cb cancel_cb, void *cancel_arg)
901 {
902 	const struct got_error *err;
903 	struct got_commit_object *commit = NULL;
904 	struct got_tag_object *tag = NULL;
905 	struct got_object_id *tree_id = NULL;
906 	struct got_object_id_queue ids;
907 	struct got_object_qid *qid;
908 	int obj_type;
909 
910 	err = got_object_qid_alloc(&qid, id);
911 	if (err)
912 		return err;
913 
914 	STAILQ_INIT(&ids);
915 	STAILQ_INSERT_TAIL(&ids, qid, entry);
916 
917 	while (!STAILQ_EMPTY(&ids)) {
918 		if (cancel_cb) {
919 			err = (*cancel_cb)(cancel_arg);
920 			if (err)
921 				break;
922 		}
923 
924 		qid = STAILQ_FIRST(&ids);
925 		STAILQ_REMOVE_HEAD(&ids, entry);
926 
927 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
928 			got_object_qid_free(qid);
929 			qid = NULL;
930 			continue;
931 		}
932 
933 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
934 		if (err)
935 			break;
936 
937 		/* This commit or tag is referenced. */
938 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
939 		if (err)
940 			break;
941 
942 		err = got_object_get_type(&obj_type, repo, &qid->id);
943 		if (err)
944 			break;
945 		switch (obj_type) {
946 		case GOT_OBJ_TYPE_COMMIT:
947 			err = got_object_open_as_commit(&commit, repo,
948 			    &qid->id);
949 			if (err)
950 				goto done;
951 			break;
952 		case GOT_OBJ_TYPE_TAG:
953 			err = got_object_open_as_tag(&tag, repo, &qid->id);
954 			if (err)
955 				goto done;
956 			break;
957 		default:
958 			/* should not happen */
959 			err = got_error(GOT_ERR_OBJ_TYPE);
960 			goto done;
961 		}
962 
963 		/* Find a tree object to scan. */
964 		if (commit) {
965 			tree_id = got_object_commit_get_tree_id(commit);
966 		} else if (tag) {
967 			obj_type = got_object_tag_get_object_type(tag);
968 			switch (obj_type) {
969 			case GOT_OBJ_TYPE_COMMIT:
970 				err = got_object_open_as_commit(&commit, repo,
971 				    got_object_tag_get_object_id(tag));
972 				if (err)
973 					goto done;
974 				tree_id = got_object_commit_get_tree_id(commit);
975 				break;
976 			case GOT_OBJ_TYPE_TREE:
977 				tree_id = got_object_tag_get_object_id(tag);
978 				break;
979 			default:
980 				/*
981 				 * Tag points at something other than a
982 				 * commit or tree. Leave this weird tag object
983 				 * and the object it points to on disk.
984 				 */
985 				err = got_object_idset_remove(NULL, loose_ids,
986 				    &qid->id);
987 				if (err && err->code != GOT_ERR_NO_OBJ)
988 					goto done;
989 				err = got_object_idset_remove(NULL, loose_ids,
990 				    got_object_tag_get_object_id(tag));
991 				if (err && err->code != GOT_ERR_NO_OBJ)
992 					goto done;
993 				err = NULL;
994 				break;
995 			}
996 		}
997 
998 		if (tree_id) {
999 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1000 			    repo, npacked, cancel_cb, cancel_arg);
1001 			if (err)
1002 				break;
1003 		}
1004 
1005 		if (commit || tag)
1006 			(*ncommits)++; /* scanned tags are counted as commits */
1007 
1008 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1009 		    nloose, *ncommits, -1);
1010 		if (err)
1011 			break;
1012 
1013 		if (commit) {
1014 			/* Find parent commits to scan. */
1015 			const struct got_object_id_queue *parent_ids;
1016 			parent_ids = got_object_commit_get_parent_ids(commit);
1017 			err = got_object_id_queue_copy(parent_ids, &ids);
1018 			if (err)
1019 				break;
1020 			got_object_commit_close(commit);
1021 			commit = NULL;
1022 		}
1023 		if (tag) {
1024 			got_object_tag_close(tag);
1025 			tag = NULL;
1026 		}
1027 		got_object_qid_free(qid);
1028 		qid = NULL;
1029 	}
1030 done:
1031 	if (qid)
1032 		got_object_qid_free(qid);
1033 	if (commit)
1034 		got_object_commit_close(commit);
1035 	if (tag)
1036 		got_object_tag_close(tag);
1037 	got_object_id_queue_free(&ids);
1038 	return err;
1039 }
1040 
1041 struct purge_loose_object_arg {
1042 	struct got_repository *repo;
1043 	got_cleanup_progress_cb progress_cb;
1044 	void *progress_arg;
1045 	struct got_ratelimit *rl;
1046 	int nloose;
1047 	int ncommits;
1048 	int npurged;
1049 	off_t size_purged;
1050 	int dry_run;
1051 	time_t max_mtime;
1052 	int ignore_mtime;
1053 };
1054 
1055 static const struct got_error *
1056 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1057 {
1058 	struct purge_loose_object_arg *a = arg;
1059 	const struct got_error *err, *unlock_err = NULL;
1060 	char *path = NULL;
1061 	int fd = -1;
1062 	struct stat sb;
1063 	struct got_lockfile *lf = NULL;
1064 
1065 	err = got_object_get_path(&path, id, a->repo);
1066 	if (err)
1067 		return err;
1068 
1069 	err = got_object_open_loose_fd(&fd, id, a->repo);
1070 	if (err)
1071 		goto done;
1072 
1073 	if (fstat(fd, &sb) == -1) {
1074 		err = got_error_from_errno("fstat");
1075 		goto done;
1076 	}
1077 
1078 	/*
1079 	 * Do not delete objects which are younger than our maximum
1080 	 * modification time threshold. This prevents a race where
1081 	 * new objects which are being added to the repository
1082 	 * concurrently would be deleted.
1083 	 */
1084 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1085 		if (!a->dry_run) {
1086 			err = got_lockfile_lock(&lf, path, -1);
1087 			if (err)
1088 				goto done;
1089 			if (unlink(path) == -1) {
1090 				err = got_error_from_errno2("unlink", path);
1091 				goto done;
1092 			}
1093 		}
1094 
1095 		a->npurged++;
1096 		a->size_purged += sb.st_size;
1097 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1098 		    a->rl, a->nloose, a->ncommits, a->npurged);
1099 		if (err)
1100 			goto done;
1101 	}
1102 done:
1103 	if (fd != -1 && close(fd) == -1 && err == NULL)
1104 		err = got_error_from_errno("close");
1105 	free(path);
1106 	if (lf)
1107 		unlock_err = got_lockfile_unlock(lf, -1);
1108 	return err ? err : unlock_err;
1109 }
1110 
1111 const struct got_error *
1112 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1113     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1114     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1115     got_cancel_cb cancel_cb, void *cancel_arg)
1116 {
1117 	const struct got_error *err;
1118 	struct got_object_idset *loose_ids;
1119 	struct got_object_idset *traversed_ids;
1120 	struct got_object_id **referenced_ids;
1121 	int i, nreferenced, nloose, ncommits = 0;
1122 	struct got_reflist_head refs;
1123 	struct got_reflist_entry *re;
1124 	struct purge_loose_object_arg arg;
1125 	time_t max_mtime = 0;
1126 	struct got_ratelimit rl;
1127 
1128 	TAILQ_INIT(&refs);
1129 	got_ratelimit_init(&rl, 0, 500);
1130 
1131 	*size_before = 0;
1132 	*size_after = 0;
1133 	*npacked = 0;
1134 
1135 	err = get_loose_object_ids(&loose_ids, size_before,
1136 	    progress_cb, progress_arg, &rl, repo);
1137 	if (err)
1138 		return err;
1139 	nloose = got_object_idset_num_elements(loose_ids);
1140 	if (nloose == 0) {
1141 		got_object_idset_free(loose_ids);
1142 		if (progress_cb) {
1143 			err = progress_cb(progress_arg, 0, 0, 0);
1144 			if (err)
1145 				return err;
1146 		}
1147 		return NULL;
1148 	}
1149 
1150 	traversed_ids = got_object_idset_alloc();
1151 	if (traversed_ids == NULL) {
1152 		err = got_error_from_errno("got_object_idset_alloc");
1153 		goto done;
1154 	}
1155 
1156 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1157 	if (err)
1158 		goto done;
1159 	if (!ignore_mtime) {
1160 		TAILQ_FOREACH(re, &refs, entry) {
1161 			time_t mtime = got_ref_get_mtime(re->ref);
1162 			if (mtime > max_mtime)
1163 				max_mtime = mtime;
1164 		}
1165 		/*
1166 		 * For safety, keep objects created within 10 minutes
1167 		 * before the youngest reference was created.
1168 		 */
1169 		if (max_mtime >= 600)
1170 			max_mtime -= 600;
1171 	}
1172 
1173 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1174 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1175 	    &refs, repo, cancel_cb, cancel_arg);
1176 	if (err)
1177 		goto done;
1178 
1179 	for (i = 0; i < nreferenced; i++) {
1180 		struct got_object_id *id = referenced_ids[i];
1181 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1182 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1183 		    nloose, cancel_cb, cancel_arg);
1184 		if (err)
1185 			goto done;
1186 	}
1187 
1188 	/* Any remaining loose objects are unreferenced and can be purged. */
1189 	arg.repo = repo;
1190 	arg.progress_arg = progress_arg;
1191 	arg.progress_cb = progress_cb;
1192 	arg.rl = &rl;
1193 	arg.nloose = nloose;
1194 	arg.npurged = 0;
1195 	arg.size_purged = 0;
1196 	arg.ncommits = ncommits;
1197 	arg.dry_run = dry_run;
1198 	arg.max_mtime = max_mtime;
1199 	arg.ignore_mtime = ignore_mtime;
1200 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1201 	if (err)
1202 		goto done;
1203 	*size_after = *size_before - arg.size_purged;
1204 
1205 	/* Produce a final progress report. */
1206 	if (progress_cb) {
1207 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1208 		if (err)
1209 			goto done;
1210 	}
1211 done:
1212 	got_object_idset_free(loose_ids);
1213 	got_object_idset_free(traversed_ids);
1214 	return err;
1215 }
1216 
1217 static const struct got_error *
1218 remove_packidx(int dir_fd, const char *relpath)
1219 {
1220 	const struct got_error *err, *unlock_err;
1221 	struct got_lockfile *lf;
1222 
1223 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1224 	if (err)
1225 		return err;
1226 	if (unlinkat(dir_fd, relpath, 0) == -1)
1227 		err = got_error_from_errno("unlinkat");
1228 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1229 	return err ? err : unlock_err;
1230 }
1231 
1232 const struct got_error *
1233 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1234     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1235     got_cancel_cb cancel_cb, void *cancel_arg)
1236 {
1237 	const struct got_error *err = NULL;
1238 	DIR *packdir = NULL;
1239 	struct dirent *dent;
1240 	char *pack_relpath = NULL;
1241 	int packdir_fd;
1242 	struct stat sb;
1243 
1244 	packdir_fd = openat(got_repo_get_fd(repo),
1245 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1246 	if (packdir_fd == -1) {
1247 		if (errno == ENOENT)
1248 			return NULL;
1249 		return got_error_from_errno_fmt("openat: %s/%s",
1250 		    got_repo_get_path_git_dir(repo),
1251 		    GOT_OBJECTS_PACK_DIR);
1252 	}
1253 
1254 	packdir = fdopendir(packdir_fd);
1255 	if (packdir == NULL) {
1256 		err = got_error_from_errno("fdopendir");
1257 		goto done;
1258 	}
1259 
1260 	while ((dent = readdir(packdir)) != NULL) {
1261 		if (cancel_cb) {
1262 			err = cancel_cb(cancel_arg);
1263 			if (err)
1264 				goto done;
1265 		}
1266 
1267 		if (!got_repo_is_packidx_filename(dent->d_name,
1268 		    strlen(dent->d_name)))
1269 			continue;
1270 
1271 		err = got_packidx_get_packfile_path(&pack_relpath,
1272 		    dent->d_name);
1273 		if (err)
1274 			goto done;
1275 
1276 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1277 			free(pack_relpath);
1278 			pack_relpath = NULL;
1279 			continue;
1280 		}
1281 		if (errno != ENOENT) {
1282 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1283 			    got_repo_get_path_git_dir(repo),
1284 			    GOT_OBJECTS_PACK_DIR,
1285 			    pack_relpath);
1286 			goto done;
1287 		}
1288 
1289 		if (!dry_run) {
1290 			err = remove_packidx(packdir_fd, dent->d_name);
1291 			if (err)
1292 				goto done;
1293 		}
1294 		if (progress_cb) {
1295 			char *path;
1296 			if (asprintf(&path, "%s/%s/%s",
1297 			    got_repo_get_path_git_dir(repo),
1298 			    GOT_OBJECTS_PACK_DIR,
1299 			    dent->d_name) == -1) {
1300 				err = got_error_from_errno("asprintf");
1301 				goto done;
1302 			}
1303 			err = progress_cb(progress_arg, path);
1304 			free(path);
1305 			if (err)
1306 				goto done;
1307 		}
1308 		free(pack_relpath);
1309 		pack_relpath = NULL;
1310 	}
1311 done:
1312 	if (packdir && closedir(packdir) != 0 && err == NULL)
1313 		err = got_error_from_errno("closedir");
1314 	free(pack_relpath);
1315 	return err;
1316 }