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