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 	FILE *delta_cache = NULL;
154 
155 	*packfile = NULL;
156 	*pack_hash = NULL;
157 
158 	if (asprintf(&path, "%s/%s/packing.pack",
159 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
160 		err = got_error_from_errno("asprintf");
161 		goto done;
162 	}
163 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
164 	if (err)
165 		goto done;
166 
167 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
168 		err = got_error_from_errno2("fchmod", tmpfile_path);
169 		goto done;
170 	}
171 
172 	delta_cache = got_opentemp();
173 	if (delta_cache == NULL) {
174 		err = got_error_from_errno("got_opentemp");
175 		goto done;
176 	}
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, packfd, delta_cache,
205 	    theirs, ntheirs, ours, nours, repo, loose_obj_only, 0,
206 	    progress_cb, progress_arg, 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 (lseek(packfd, 0L, SEEK_SET) == -1) {
221 		err = got_error_from_errno("lseek");
222 		goto done;
223 	}
224 	if (rename(tmpfile_path, packfile_path) == -1) {
225 		err = got_error_from_errno3("rename", tmpfile_path,
226 		    packfile_path);
227 		goto done;
228 	}
229 	free(tmpfile_path);
230 	tmpfile_path = NULL;
231 
232 	*packfile = fdopen(packfd, "w");
233 	if (*packfile == NULL) {
234 		err = got_error_from_errno2("fdopen", tmpfile_path);
235 		goto done;
236 	}
237 	packfd = -1;
238 done:
239 	for (i = 0; i < nours; i++)
240 		free(ours[i]);
241 	free(ours);
242 	for (i = 0; i < ntheirs; i++)
243 		free(theirs[i]);
244 	free(theirs);
245 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
246 		err = got_error_from_errno2("close", packfile_path);
247 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
248 		err = got_error_from_errno("fclose");
249 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
250 		err = got_error_from_errno2("unlink", tmpfile_path);
251 	free(tmpfile_path);
252 	free(packfile_path);
253 	free(sha1_str);
254 	free(path);
255 	if (err) {
256 		free(*pack_hash);
257 		*pack_hash = NULL;
258 		if (*packfile)
259 			fclose(*packfile);
260 		*packfile = NULL;
261 	}
262 	return err;
263 }
264 
265 const struct got_error *
266 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
267     struct got_repository *repo,
268     got_pack_index_progress_cb progress_cb, void *progress_arg,
269     got_cancel_cb cancel_cb, void *cancel_arg)
270 {
271 	size_t i;
272 	char *path;
273 	int imsg_idxfds[2];
274 	int npackfd = -1, idxfd = -1, nidxfd = -1;
275 	int tmpfds[3];
276 	int idxstatus, done = 0;
277 	const struct got_error *err;
278 	struct imsgbuf idxibuf;
279 	pid_t idxpid;
280 	char *tmpidxpath = NULL;
281 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
282 	const char *repo_path = got_repo_get_path_git_dir(repo);
283 	struct stat sb;
284 
285 	for (i = 0; i < nitems(tmpfds); i++)
286 		tmpfds[i] = -1;
287 
288 	if (asprintf(&path, "%s/%s/indexing.idx",
289 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
290 		err = got_error_from_errno("asprintf");
291 		goto done;
292 	}
293 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
294 	free(path);
295 	if (err)
296 		goto done;
297 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
298 		err = got_error_from_errno2("fchmod", tmpidxpath);
299 		goto done;
300 	}
301 
302 	nidxfd = dup(idxfd);
303 	if (nidxfd == -1) {
304 		err = got_error_from_errno("dup");
305 		goto done;
306 	}
307 
308 	for (i = 0; i < nitems(tmpfds); i++) {
309 		tmpfds[i] = got_opentempfd();
310 		if (tmpfds[i] == -1) {
311 			err = got_error_from_errno("got_opentempfd");
312 			goto done;
313 		}
314 	}
315 
316 	err = got_object_id_str(&id_str, pack_hash);
317 	if (err)
318 		goto done;
319 
320 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
321 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
322 		err = got_error_from_errno("asprintf");
323 		goto done;
324 	}
325 
326 	if (fstat(fileno(packfile), &sb) == -1) {
327 		err = got_error_from_errno2("fstat", packfile_path);
328 		goto done;
329 	}
330 
331 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
332 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
333 		err = got_error_from_errno("asprintf");
334 		goto done;
335 	}
336 
337 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
338 		err = got_error_from_errno("socketpair");
339 		goto done;
340 	}
341 	idxpid = fork();
342 	if (idxpid == -1) {
343 		err= got_error_from_errno("fork");
344 		goto done;
345 	} else if (idxpid == 0)
346 		got_privsep_exec_child(imsg_idxfds,
347 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
348 	if (close(imsg_idxfds[1]) == -1) {
349 		err = got_error_from_errno("close");
350 		goto done;
351 	}
352 	imsg_init(&idxibuf, imsg_idxfds[0]);
353 
354 	npackfd = dup(fileno(packfile));
355 	if (npackfd == -1) {
356 		err = got_error_from_errno("dup");
357 		goto done;
358 	}
359 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
360 	    npackfd);
361 	if (err != NULL)
362 		goto done;
363 	npackfd = -1;
364 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
365 	if (err != NULL)
366 		goto done;
367 	nidxfd = -1;
368 	for (i = 0; i < nitems(tmpfds); i++) {
369 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
370 		if (err != NULL)
371 			goto done;
372 		tmpfds[i] = -1;
373 	}
374 	done = 0;
375 	while (!done) {
376 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
377 
378 		if (cancel_cb) {
379 			err = cancel_cb(cancel_arg);
380 			if (err)
381 				goto done;
382 		}
383 
384 		err = got_privsep_recv_index_progress(&done, &nobj_total,
385 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
386 		    &idxibuf);
387 		if (err != NULL)
388 			goto done;
389 		if (nobj_indexed != 0) {
390 			err = progress_cb(progress_arg, sb.st_size,
391 			    nobj_total, nobj_indexed, nobj_loose,
392 			    nobj_resolved);
393 			if (err)
394 				break;
395 		}
396 	}
397 	if (close(imsg_idxfds[0]) == -1) {
398 		err = got_error_from_errno("close");
399 		goto done;
400 	}
401 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
402 		err = got_error_from_errno("waitpid");
403 		goto done;
404 	}
405 
406 	if (rename(tmpidxpath, idxpath) == -1) {
407 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
408 		goto done;
409 	}
410 	free(tmpidxpath);
411 	tmpidxpath = NULL;
412 
413 done:
414 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
415 		err = got_error_from_errno2("unlink", tmpidxpath);
416 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
417 		err = got_error_from_errno("close");
418 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
419 		err = got_error_from_errno("close");
420 	for (i = 0; i < nitems(tmpfds); i++) {
421 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
422 			err = got_error_from_errno("close");
423 	}
424 	free(tmpidxpath);
425 	free(idxpath);
426 	free(packfile_path);
427 	return err;
428 }
429 
430 const struct got_error *
431 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
432     struct got_repository *repo, const char *packfile_path)
433 {
434 	const struct got_error *err = NULL;
435 	const char *packdir_path = NULL;
436 	char *packfile_name = NULL, *p, *dot;
437 	struct got_object_id id;
438 	int packfd = -1;
439 
440 	*packfile = NULL;
441 	*pack_hash = NULL;
442 
443 	packdir_path = got_repo_get_path_objects_pack(repo);
444 	if (packdir_path == NULL)
445 		return got_error_from_errno("got_repo_get_path_objects_pack");
446 
447 	if (!got_path_is_child(packfile_path, packdir_path,
448 	    strlen(packdir_path))) {
449 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
450 		goto done;
451 
452 	}
453 
454 	err = got_path_basename(&packfile_name, packfile_path);
455 	if (err)
456 		goto done;
457 	p = packfile_name;
458 
459 	if (strncmp(p, "pack-", 5) != 0) {
460 		err = got_error_fmt(GOT_ERR_BAD_PATH,
461 		    "'%s' is not a valid pack file name",
462 		    packfile_name);
463 		goto done;
464 	}
465 	p += 5;
466 	dot = strchr(p, '.');
467 	if (dot == NULL) {
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 	if (strcmp(dot + 1, "pack") != 0) {
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 	*dot = '\0';
480 	if (!got_parse_sha1_digest(id.sha1, p)) {
481 		err = got_error_fmt(GOT_ERR_BAD_PATH,
482 		    "'%s' is not a valid pack file name",
483 		    packfile_name);
484 		goto done;
485 	}
486 
487 	*pack_hash = got_object_id_dup(&id);
488 	if (*pack_hash == NULL) {
489 		err = got_error_from_errno("got_object_id_dup");
490 		goto done;
491 	}
492 
493 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
494 	if (packfd == -1) {
495 		err = got_error_from_errno2("open", packfile_path);
496 		goto done;
497 	}
498 
499 	*packfile = fdopen(packfd, "r");
500 	if (*packfile == NULL) {
501 		err = got_error_from_errno2("fdopen", packfile_path);
502 		goto done;
503 	}
504 	packfd = -1;
505 done:
506 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
507 		err = got_error_from_errno2("close", packfile_path);
508 	free(packfile_name);
509 	if (err) {
510 		free(*pack_hash);
511 		*pack_hash = NULL;
512 	}
513 	return err;
514 }
515 
516 const struct got_error *
517 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
518     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
519     got_cancel_cb cancel_cb, void *cancel_arg)
520 {
521 	const struct got_error *err = NULL;
522 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
523 	struct got_packidx *packidx = NULL;
524 	struct got_pack *pack = NULL;
525 	uint32_t nobj, i;
526 
527 	err = got_object_id_str(&id_str, pack_hash);
528 	if (err)
529 		goto done;
530 
531 	if (asprintf(&packpath, "%s/pack-%s.pack",
532 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
533 		err = got_error_from_errno("asprintf");
534 		goto done;
535 	}
536 	if (asprintf(&idxpath, "%s/pack-%s.idx",
537 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
538 		err = got_error_from_errno("asprintf");
539 		goto done;
540 	}
541 
542 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
543 	if (err)
544 		goto done;
545 
546 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
547 	if (err)
548 		goto done;
549 
550 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
551 	for (i = 0; i < nobj; i++) {
552 		struct got_packidx_object_id *oid;
553 		struct got_object_id id, base_id;
554 		off_t offset, base_offset = 0;
555 		uint8_t type;
556 		uint64_t size;
557 		size_t tslen, len;
558 
559 		if (cancel_cb) {
560 			err = cancel_cb(cancel_arg);
561 			if (err)
562 				break;
563 		}
564 		oid = &packidx->hdr.sorted_ids[i];
565 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
566 
567 		offset = got_packidx_get_object_offset(packidx, i);
568 		if (offset == -1) {
569 			err = got_error(GOT_ERR_BAD_PACKIDX);
570 			goto done;
571 		}
572 
573 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
574 		    pack, offset);
575 		if (err)
576 			goto done;
577 
578 		switch (type) {
579 		case GOT_OBJ_TYPE_OFFSET_DELTA:
580 			err = got_pack_parse_offset_delta(&base_offset, &len,
581 			    pack, offset, tslen);
582 			if (err)
583 				goto done;
584 			break;
585 		case GOT_OBJ_TYPE_REF_DELTA:
586 			err = got_pack_parse_ref_delta(&base_id,
587 			    pack, offset, tslen);
588 			if (err)
589 				goto done;
590 			break;
591 		}
592 		err = (*list_cb)(list_arg, &id, type, offset, size,
593 		    base_offset, &base_id);
594 		if (err)
595 			goto done;
596 	}
597 
598 done:
599 	free(id_str);
600 	free(idxpath);
601 	free(packpath);
602 	if (packidx)
603 		got_packidx_close(packidx);
604 	return err;
605 }
606 
607 static const struct got_error *
608 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
609     void *progress_arg, struct got_ratelimit *rl,
610     int nloose, int ncommits, int npurged)
611 {
612 	const struct got_error *err;
613 	int elapsed;
614 
615 	if (progress_cb == NULL)
616 		return NULL;
617 
618 	err = got_ratelimit_check(&elapsed, rl);
619 	if (err || !elapsed)
620 		return err;
621 
622 	return progress_cb(progress_arg, nloose, ncommits, npurged);
623 }
624 
625 static const struct got_error *
626 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
627     got_cleanup_progress_cb progress_cb, void *progress_arg,
628     struct got_ratelimit *rl, struct got_repository *repo)
629 {
630 	const struct got_error *err = NULL;
631 	char *path_objects = NULL, *path = NULL;
632 	DIR *dir = NULL;
633 	struct got_object *obj = NULL;
634 	struct got_object_id id;
635 	int i, fd = -1;
636 	struct stat sb;
637 
638 	*ondisk_size = 0;
639 	*loose_ids = got_object_idset_alloc();
640 	if (*loose_ids == NULL)
641 		return got_error_from_errno("got_object_idset_alloc");
642 
643 	path_objects = got_repo_get_path_objects(repo);
644 	if (path_objects == NULL) {
645 		err = got_error_from_errno("got_repo_get_path_objects");
646 		goto done;
647 	}
648 
649 	for (i = 0; i <= 0xff; i++) {
650 		struct dirent *dent;
651 
652 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
653 			err = got_error_from_errno("asprintf");
654 			break;
655 		}
656 
657 		dir = opendir(path);
658 		if (dir == NULL) {
659 			if (errno == ENOENT) {
660 				err = NULL;
661 				continue;
662 			}
663 			err = got_error_from_errno2("opendir", path);
664 			break;
665 		}
666 
667 		while ((dent = readdir(dir)) != NULL) {
668 			char *id_str;
669 
670 			if (strcmp(dent->d_name, ".") == 0 ||
671 			    strcmp(dent->d_name, "..") == 0)
672 				continue;
673 
674 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
675 				err = got_error_from_errno("asprintf");
676 				goto done;
677 			}
678 
679 			memset(&id, 0, sizeof(id));
680 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
681 				free(id_str);
682 				continue;
683 			}
684 			free(id_str);
685 
686 			err = got_object_open_loose_fd(&fd, &id, repo);
687 			if (err)
688 				goto done;
689 			if (fstat(fd, &sb) == -1) {
690 				err = got_error_from_errno("fstat");
691 				goto done;
692 			}
693 			err = got_object_read_header_privsep(&obj, &id, repo,
694 			    fd);
695 			if (err)
696 				goto done;
697 			fd = -1; /* already closed */
698 
699 			switch (obj->type) {
700 			case GOT_OBJ_TYPE_COMMIT:
701 			case GOT_OBJ_TYPE_TREE:
702 			case GOT_OBJ_TYPE_BLOB:
703 			case GOT_OBJ_TYPE_TAG:
704 				break;
705 			default:
706 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
707 				    "%d", obj->type);
708 				goto done;
709 			}
710 			got_object_close(obj);
711 			obj = NULL;
712 			(*ondisk_size) += sb.st_size;
713 			err = got_object_idset_add(*loose_ids, &id, NULL);
714 			if (err)
715 				goto done;
716 			err = report_cleanup_progress(progress_cb,
717 			    progress_arg, rl,
718 			    got_object_idset_num_elements(*loose_ids), -1, -1);
719 			if (err)
720 				goto done;
721 		}
722 
723 		if (closedir(dir) != 0) {
724 			err = got_error_from_errno("closedir");
725 			goto done;
726 		}
727 		dir = NULL;
728 
729 		free(path);
730 		path = NULL;
731 	}
732 done:
733 	if (dir && closedir(dir) != 0 && err == NULL)
734 		err = got_error_from_errno("closedir");
735 	if (fd != -1 && close(fd) == -1 && err == NULL)
736 		err = got_error_from_errno("close");
737 	if (err) {
738 		got_object_idset_free(*loose_ids);
739 		*loose_ids = NULL;
740 	}
741 	if (obj)
742 		got_object_close(obj);
743 	free(path_objects);
744 	free(path);
745 	return err;
746 }
747 
748 static const struct got_error *
749 preserve_loose_object(struct got_object_idset *loose_ids,
750     struct got_object_id *id, struct got_repository *repo, int *npacked)
751 {
752 	const struct got_error *err = NULL;
753 	struct got_object *obj;
754 
755 	if (!got_object_idset_contains(loose_ids, id))
756 		return NULL;
757 
758 	/*
759 	 * Try to open this object from a pack file. This ensures that
760 	 * we do in fact have a valid packed copy of the object. Otherwise
761 	 * we should not delete the loose representation of this object.
762 	 */
763 	err = got_object_open_packed(&obj, id, repo);
764 	if (err == NULL) {
765 		got_object_close(obj);
766 		/*
767 		 * The object is referenced and packed.
768 		 * We can purge the redundantly stored loose object.
769 		 */
770 		(*npacked)++;
771 		return NULL;
772 	} else if (err->code != GOT_ERR_NO_OBJ)
773 		return err;
774 
775 	/*
776 	 * This object is referenced and not packed.
777 	 * Remove it from our purge set.
778 	 */
779 	return got_object_idset_remove(NULL, loose_ids, id);
780 }
781 
782 static const struct got_error *
783 load_tree_entries(struct got_object_id_queue *ids,
784     struct got_object_idset *loose_ids,
785     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
786     const char *dpath, struct got_repository *repo, int *npacked,
787     got_cancel_cb cancel_cb, void *cancel_arg)
788 {
789 	const struct got_error *err;
790 	struct got_tree_object *tree;
791 	char *p = NULL;
792 	int i;
793 
794 	err = got_object_open_as_tree(&tree, repo, tree_id);
795 	if (err)
796 		return err;
797 
798 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
799 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
800 		struct got_object_id *id = got_tree_entry_get_id(e);
801 		mode_t mode = got_tree_entry_get_mode(e);
802 
803 		if (cancel_cb) {
804 			err = (*cancel_cb)(cancel_arg);
805 			if (err)
806 				break;
807 		}
808 
809 		if (got_object_tree_entry_is_symlink(e) ||
810 		    got_object_tree_entry_is_submodule(e) ||
811 		    got_object_idset_contains(traversed_ids, id))
812 			continue;
813 
814 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
815 		    got_tree_entry_get_name(e)) == -1) {
816 			err = got_error_from_errno("asprintf");
817 			break;
818 		}
819 
820 		if (S_ISDIR(mode)) {
821 			struct got_object_qid *qid;
822 			err = got_object_qid_alloc(&qid, id);
823 			if (err)
824 				break;
825 			STAILQ_INSERT_TAIL(ids, qid, entry);
826 		} else if (S_ISREG(mode)) {
827 			/* This blob is referenced. */
828 			err = preserve_loose_object(loose_ids, id, repo,
829 			    npacked);
830 			if (err)
831 				break;
832 			err = got_object_idset_add(traversed_ids, id, NULL);
833 			if (err)
834 				break;
835 		}
836 		free(p);
837 		p = NULL;
838 	}
839 
840 	got_object_tree_close(tree);
841 	free(p);
842 	return err;
843 }
844 
845 static const struct got_error *
846 load_tree(struct got_object_idset *loose_ids,
847     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
848     const char *dpath, struct got_repository *repo, int *npacked,
849     got_cancel_cb cancel_cb, void *cancel_arg)
850 {
851 	const struct got_error *err = NULL;
852 	struct got_object_id_queue tree_ids;
853 	struct got_object_qid *qid;
854 
855 	err = got_object_qid_alloc(&qid, tree_id);
856 	if (err)
857 		return err;
858 
859 	STAILQ_INIT(&tree_ids);
860 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
861 
862 	while (!STAILQ_EMPTY(&tree_ids)) {
863 		if (cancel_cb) {
864 			err = (*cancel_cb)(cancel_arg);
865 			if (err)
866 				break;
867 		}
868 
869 		qid = STAILQ_FIRST(&tree_ids);
870 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
871 
872 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
873 			got_object_qid_free(qid);
874 			continue;
875 		}
876 
877 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
878 		if (err) {
879 			got_object_qid_free(qid);
880 			break;
881 		}
882 
883 		/* This tree is referenced. */
884 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
885 		if (err)
886 			break;
887 
888 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
889 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
890 		got_object_qid_free(qid);
891 		if (err)
892 			break;
893 	}
894 
895 	got_object_id_queue_free(&tree_ids);
896 	return err;
897 }
898 
899 static const struct got_error *
900 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
901     int *npacked, struct got_object_idset *traversed_ids,
902     struct got_object_id *id, struct got_repository *repo,
903     got_cleanup_progress_cb progress_cb, void *progress_arg,
904     struct got_ratelimit *rl, int nloose,
905     got_cancel_cb cancel_cb, void *cancel_arg)
906 {
907 	const struct got_error *err;
908 	struct got_commit_object *commit = NULL;
909 	struct got_tag_object *tag = NULL;
910 	struct got_object_id *tree_id = NULL;
911 	struct got_object_id_queue ids;
912 	struct got_object_qid *qid;
913 	int obj_type;
914 
915 	err = got_object_qid_alloc(&qid, id);
916 	if (err)
917 		return err;
918 
919 	STAILQ_INIT(&ids);
920 	STAILQ_INSERT_TAIL(&ids, qid, entry);
921 
922 	while (!STAILQ_EMPTY(&ids)) {
923 		if (cancel_cb) {
924 			err = (*cancel_cb)(cancel_arg);
925 			if (err)
926 				break;
927 		}
928 
929 		qid = STAILQ_FIRST(&ids);
930 		STAILQ_REMOVE_HEAD(&ids, entry);
931 
932 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
933 			got_object_qid_free(qid);
934 			qid = NULL;
935 			continue;
936 		}
937 
938 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
939 		if (err)
940 			break;
941 
942 		/* This commit or tag is referenced. */
943 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
944 		if (err)
945 			break;
946 
947 		err = got_object_get_type(&obj_type, repo, &qid->id);
948 		if (err)
949 			break;
950 		switch (obj_type) {
951 		case GOT_OBJ_TYPE_COMMIT:
952 			err = got_object_open_as_commit(&commit, repo,
953 			    &qid->id);
954 			if (err)
955 				goto done;
956 			break;
957 		case GOT_OBJ_TYPE_TAG:
958 			err = got_object_open_as_tag(&tag, repo, &qid->id);
959 			if (err)
960 				goto done;
961 			break;
962 		default:
963 			/* should not happen */
964 			err = got_error(GOT_ERR_OBJ_TYPE);
965 			goto done;
966 		}
967 
968 		/* Find a tree object to scan. */
969 		if (commit) {
970 			tree_id = got_object_commit_get_tree_id(commit);
971 		} else if (tag) {
972 			obj_type = got_object_tag_get_object_type(tag);
973 			switch (obj_type) {
974 			case GOT_OBJ_TYPE_COMMIT:
975 				err = got_object_open_as_commit(&commit, repo,
976 				    got_object_tag_get_object_id(tag));
977 				if (err)
978 					goto done;
979 				tree_id = got_object_commit_get_tree_id(commit);
980 				break;
981 			case GOT_OBJ_TYPE_TREE:
982 				tree_id = got_object_tag_get_object_id(tag);
983 				break;
984 			default:
985 				/*
986 				 * Tag points at something other than a
987 				 * commit or tree. Leave this weird tag object
988 				 * and the object it points to on disk.
989 				 */
990 				err = got_object_idset_remove(NULL, loose_ids,
991 				    &qid->id);
992 				if (err && err->code != GOT_ERR_NO_OBJ)
993 					goto done;
994 				err = got_object_idset_remove(NULL, loose_ids,
995 				    got_object_tag_get_object_id(tag));
996 				if (err && err->code != GOT_ERR_NO_OBJ)
997 					goto done;
998 				err = NULL;
999 				break;
1000 			}
1001 		}
1002 
1003 		if (tree_id) {
1004 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1005 			    repo, npacked, cancel_cb, cancel_arg);
1006 			if (err)
1007 				break;
1008 		}
1009 
1010 		if (commit || tag)
1011 			(*ncommits)++; /* scanned tags are counted as commits */
1012 
1013 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1014 		    nloose, *ncommits, -1);
1015 		if (err)
1016 			break;
1017 
1018 		if (commit) {
1019 			/* Find parent commits to scan. */
1020 			const struct got_object_id_queue *parent_ids;
1021 			parent_ids = got_object_commit_get_parent_ids(commit);
1022 			err = got_object_id_queue_copy(parent_ids, &ids);
1023 			if (err)
1024 				break;
1025 			got_object_commit_close(commit);
1026 			commit = NULL;
1027 		}
1028 		if (tag) {
1029 			got_object_tag_close(tag);
1030 			tag = NULL;
1031 		}
1032 		got_object_qid_free(qid);
1033 		qid = NULL;
1034 	}
1035 done:
1036 	if (qid)
1037 		got_object_qid_free(qid);
1038 	if (commit)
1039 		got_object_commit_close(commit);
1040 	if (tag)
1041 		got_object_tag_close(tag);
1042 	got_object_id_queue_free(&ids);
1043 	return err;
1044 }
1045 
1046 struct purge_loose_object_arg {
1047 	struct got_repository *repo;
1048 	got_cleanup_progress_cb progress_cb;
1049 	void *progress_arg;
1050 	struct got_ratelimit *rl;
1051 	int nloose;
1052 	int ncommits;
1053 	int npurged;
1054 	off_t size_purged;
1055 	int dry_run;
1056 	time_t max_mtime;
1057 	int ignore_mtime;
1058 };
1059 
1060 static const struct got_error *
1061 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1062 {
1063 	struct purge_loose_object_arg *a = arg;
1064 	const struct got_error *err, *unlock_err = NULL;
1065 	char *path = NULL;
1066 	int fd = -1;
1067 	struct stat sb;
1068 	struct got_lockfile *lf = NULL;
1069 
1070 	err = got_object_get_path(&path, id, a->repo);
1071 	if (err)
1072 		return err;
1073 
1074 	err = got_object_open_loose_fd(&fd, id, a->repo);
1075 	if (err)
1076 		goto done;
1077 
1078 	if (fstat(fd, &sb) == -1) {
1079 		err = got_error_from_errno("fstat");
1080 		goto done;
1081 	}
1082 
1083 	/*
1084 	 * Do not delete objects which are younger than our maximum
1085 	 * modification time threshold. This prevents a race where
1086 	 * new objects which are being added to the repository
1087 	 * concurrently would be deleted.
1088 	 */
1089 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1090 		if (!a->dry_run) {
1091 			err = got_lockfile_lock(&lf, path, -1);
1092 			if (err)
1093 				goto done;
1094 			if (unlink(path) == -1) {
1095 				err = got_error_from_errno2("unlink", path);
1096 				goto done;
1097 			}
1098 		}
1099 
1100 		a->npurged++;
1101 		a->size_purged += sb.st_size;
1102 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1103 		    a->rl, a->nloose, a->ncommits, a->npurged);
1104 		if (err)
1105 			goto done;
1106 	}
1107 done:
1108 	if (fd != -1 && close(fd) == -1 && err == NULL)
1109 		err = got_error_from_errno("close");
1110 	free(path);
1111 	if (lf)
1112 		unlock_err = got_lockfile_unlock(lf, -1);
1113 	return err ? err : unlock_err;
1114 }
1115 
1116 const struct got_error *
1117 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1118     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1119     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1120     got_cancel_cb cancel_cb, void *cancel_arg)
1121 {
1122 	const struct got_error *err;
1123 	struct got_object_idset *loose_ids;
1124 	struct got_object_idset *traversed_ids;
1125 	struct got_object_id **referenced_ids;
1126 	int i, nreferenced, nloose, ncommits = 0;
1127 	struct got_reflist_head refs;
1128 	struct got_reflist_entry *re;
1129 	struct purge_loose_object_arg arg;
1130 	time_t max_mtime = 0;
1131 	struct got_ratelimit rl;
1132 
1133 	TAILQ_INIT(&refs);
1134 	got_ratelimit_init(&rl, 0, 500);
1135 
1136 	*size_before = 0;
1137 	*size_after = 0;
1138 	*npacked = 0;
1139 
1140 	err = get_loose_object_ids(&loose_ids, size_before,
1141 	    progress_cb, progress_arg, &rl, repo);
1142 	if (err)
1143 		return err;
1144 	nloose = got_object_idset_num_elements(loose_ids);
1145 	if (nloose == 0) {
1146 		got_object_idset_free(loose_ids);
1147 		if (progress_cb) {
1148 			err = progress_cb(progress_arg, 0, 0, 0);
1149 			if (err)
1150 				return err;
1151 		}
1152 		return NULL;
1153 	}
1154 
1155 	traversed_ids = got_object_idset_alloc();
1156 	if (traversed_ids == NULL) {
1157 		err = got_error_from_errno("got_object_idset_alloc");
1158 		goto done;
1159 	}
1160 
1161 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1162 	if (err)
1163 		goto done;
1164 	if (!ignore_mtime) {
1165 		TAILQ_FOREACH(re, &refs, entry) {
1166 			time_t mtime = got_ref_get_mtime(re->ref);
1167 			if (mtime > max_mtime)
1168 				max_mtime = mtime;
1169 		}
1170 		/*
1171 		 * For safety, keep objects created within 10 minutes
1172 		 * before the youngest reference was created.
1173 		 */
1174 		if (max_mtime >= 600)
1175 			max_mtime -= 600;
1176 	}
1177 
1178 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1179 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1180 	    &refs, repo, cancel_cb, cancel_arg);
1181 	if (err)
1182 		goto done;
1183 
1184 	for (i = 0; i < nreferenced; i++) {
1185 		struct got_object_id *id = referenced_ids[i];
1186 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1187 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1188 		    nloose, cancel_cb, cancel_arg);
1189 		if (err)
1190 			goto done;
1191 	}
1192 
1193 	/* Any remaining loose objects are unreferenced and can be purged. */
1194 	arg.repo = repo;
1195 	arg.progress_arg = progress_arg;
1196 	arg.progress_cb = progress_cb;
1197 	arg.rl = &rl;
1198 	arg.nloose = nloose;
1199 	arg.npurged = 0;
1200 	arg.size_purged = 0;
1201 	arg.ncommits = ncommits;
1202 	arg.dry_run = dry_run;
1203 	arg.max_mtime = max_mtime;
1204 	arg.ignore_mtime = ignore_mtime;
1205 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1206 	if (err)
1207 		goto done;
1208 	*size_after = *size_before - arg.size_purged;
1209 
1210 	/* Produce a final progress report. */
1211 	if (progress_cb) {
1212 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1213 		if (err)
1214 			goto done;
1215 	}
1216 done:
1217 	got_object_idset_free(loose_ids);
1218 	got_object_idset_free(traversed_ids);
1219 	return err;
1220 }
1221 
1222 static const struct got_error *
1223 remove_packidx(int dir_fd, const char *relpath)
1224 {
1225 	const struct got_error *err, *unlock_err;
1226 	struct got_lockfile *lf;
1227 
1228 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1229 	if (err)
1230 		return err;
1231 	if (unlinkat(dir_fd, relpath, 0) == -1)
1232 		err = got_error_from_errno("unlinkat");
1233 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1234 	return err ? err : unlock_err;
1235 }
1236 
1237 const struct got_error *
1238 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1239     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1240     got_cancel_cb cancel_cb, void *cancel_arg)
1241 {
1242 	const struct got_error *err = NULL;
1243 	DIR *packdir = NULL;
1244 	struct dirent *dent;
1245 	char *pack_relpath = NULL;
1246 	int packdir_fd;
1247 	struct stat sb;
1248 
1249 	packdir_fd = openat(got_repo_get_fd(repo),
1250 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1251 	if (packdir_fd == -1) {
1252 		if (errno == ENOENT)
1253 			return NULL;
1254 		return got_error_from_errno_fmt("openat: %s/%s",
1255 		    got_repo_get_path_git_dir(repo),
1256 		    GOT_OBJECTS_PACK_DIR);
1257 	}
1258 
1259 	packdir = fdopendir(packdir_fd);
1260 	if (packdir == NULL) {
1261 		err = got_error_from_errno("fdopendir");
1262 		goto done;
1263 	}
1264 
1265 	while ((dent = readdir(packdir)) != NULL) {
1266 		if (cancel_cb) {
1267 			err = cancel_cb(cancel_arg);
1268 			if (err)
1269 				goto done;
1270 		}
1271 
1272 		if (!got_repo_is_packidx_filename(dent->d_name,
1273 		    strlen(dent->d_name)))
1274 			continue;
1275 
1276 		err = got_packidx_get_packfile_path(&pack_relpath,
1277 		    dent->d_name);
1278 		if (err)
1279 			goto done;
1280 
1281 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1282 			free(pack_relpath);
1283 			pack_relpath = NULL;
1284 			continue;
1285 		}
1286 		if (errno != ENOENT) {
1287 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1288 			    got_repo_get_path_git_dir(repo),
1289 			    GOT_OBJECTS_PACK_DIR,
1290 			    pack_relpath);
1291 			goto done;
1292 		}
1293 
1294 		if (!dry_run) {
1295 			err = remove_packidx(packdir_fd, dent->d_name);
1296 			if (err)
1297 				goto done;
1298 		}
1299 		if (progress_cb) {
1300 			char *path;
1301 			if (asprintf(&path, "%s/%s/%s",
1302 			    got_repo_get_path_git_dir(repo),
1303 			    GOT_OBJECTS_PACK_DIR,
1304 			    dent->d_name) == -1) {
1305 				err = got_error_from_errno("asprintf");
1306 				goto done;
1307 			}
1308 			err = progress_cb(progress_arg, path);
1309 			free(path);
1310 			if (err)
1311 				goto done;
1312 		}
1313 		free(pack_relpath);
1314 		pack_relpath = NULL;
1315 	}
1316 done:
1317 	if (packdir && closedir(packdir) != 0 && err == NULL)
1318 		err = got_error_from_errno("closedir");
1319 	free(pack_relpath);
1320 	return err;
1321 }