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_ratelimit.h"
55 #include "got_lib_pack_create.h"
56 #include "got_lib_sha1.h"
57 #include "got_lib_lockfile.h"
58 
59 #ifndef nitems
60 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
61 #endif
62 
63 static const struct got_error *
64 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
65     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
66     struct got_repository *repo,
67     got_cancel_cb cancel_cb, void *cancel_arg)
68 {
69 	const struct got_error *err = NULL;
70 	const size_t alloc_chunksz = 256;
71 	size_t nalloc;
72 	struct got_reflist_entry *re;
73 	int i;
74 
75 	*ids = NULL;
76 	*nobjects = 0;
77 
78 	err = got_reflist_sort(refs,
79 	    got_ref_cmp_by_commit_timestamp_descending, repo);
80 	if (err)
81 		return err;
82 
83 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
84 	if (*ids == NULL)
85 		return got_error_from_errno("reallocarray");
86 	nalloc = alloc_chunksz;
87 
88 	TAILQ_FOREACH(re, refs, entry) {
89 		struct got_object_id *id;
90 
91 		if (cancel_cb) {
92 			err = cancel_cb(cancel_arg);
93 			if (err)
94 				goto done;
95 		}
96 
97 		err = got_ref_resolve(&id, repo, re->ref);
98 		if (err)
99 			goto done;
100 
101 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
102 			int obj_type;
103 			err = got_object_get_type(&obj_type, repo, id);
104 			if (err)
105 				goto done;
106 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
107 				free(id);
108 				id = NULL;
109 				continue;
110 			}
111 		}
112 
113 		if (nalloc <= *nobjects) {
114 			struct got_object_id **new;
115 			new = recallocarray(*ids, nalloc,
116 			    nalloc + alloc_chunksz,
117 			    sizeof(struct got_object_id *));
118 			if (new == NULL) {
119 				err = got_error_from_errno(
120 				    "recallocarray");
121 				goto done;
122 			}
123 			*ids = new;
124 			nalloc += alloc_chunksz;
125 		}
126 		(*ids)[*nobjects] = id;
127 		if ((*ids)[*nobjects] == NULL) {
128 			err = got_error_from_errno("got_object_id_dup");
129 			goto done;
130 		}
131 		(*nobjects)++;
132 	}
133 done:
134 	if (err) {
135 		for (i = 0; i < *nobjects; i++)
136 			free((*ids)[i]);
137 		free(*ids);
138 		*ids = NULL;
139 		*nobjects = 0;
140 	}
141 	return err;
142 }
143 
144 const struct got_error *
145 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
146     struct got_reflist_head *include_refs,
147     struct got_reflist_head *exclude_refs, struct got_repository *repo,
148     int loose_obj_only,
149     got_pack_progress_cb progress_cb, void *progress_arg,
150     got_cancel_cb cancel_cb, void *cancel_arg)
151 {
152 	const struct got_error *err = NULL;
153 	struct got_object_id **ours = NULL, **theirs = NULL;
154 	int nours = 0, ntheirs = 0, packfd = -1, i;
155 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
156 	char *sha1_str = NULL;
157 	FILE *delta_cache = NULL;
158 	struct got_ratelimit rl;
159 
160 	*packfile = NULL;
161 	*pack_hash = NULL;
162 
163 	got_ratelimit_init(&rl, 0, 500);
164 
165 	if (asprintf(&path, "%s/%s/packing.pack",
166 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
167 		err = got_error_from_errno("asprintf");
168 		goto done;
169 	}
170 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
171 	if (err)
172 		goto done;
173 
174 	if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
175 		err = got_error_from_errno2("fchmod", tmpfile_path);
176 		goto done;
177 	}
178 
179 	delta_cache = got_opentemp();
180 	if (delta_cache == NULL) {
181 		err = got_error_from_errno("got_opentemp");
182 		goto done;
183 	}
184 
185 	err = get_reflist_object_ids(&ours, &nours,
186 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
187 	    include_refs, repo, cancel_cb, cancel_arg);
188 	if (err)
189 		goto done;
190 
191 	if (nours == 0) {
192 		err = got_error(GOT_ERR_CANNOT_PACK);
193 		goto done;
194 	}
195 
196 	if (!TAILQ_EMPTY(exclude_refs)) {
197 		err = get_reflist_object_ids(&theirs, &ntheirs,
198 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
199 		    exclude_refs, repo,
200 		    cancel_cb, cancel_arg);
201 		if (err)
202 			goto done;
203 	}
204 
205 	*pack_hash = calloc(1, sizeof(**pack_hash));
206 	if (*pack_hash == NULL) {
207 		err = got_error_from_errno("calloc");
208 		goto done;
209 	}
210 
211 	err = got_pack_create((*pack_hash)->sha1, packfd, delta_cache,
212 	    theirs, ntheirs, ours, nours, repo, loose_obj_only, 0,
213 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
214 	if (err)
215 		goto done;
216 
217 	err = got_object_id_str(&sha1_str, *pack_hash);
218 	if (err)
219 		goto done;
220 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
221 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
222 	    sha1_str) == -1) {
223 		err = got_error_from_errno("asprintf");
224 		goto done;
225 	}
226 
227 	if (lseek(packfd, 0L, SEEK_SET) == -1) {
228 		err = got_error_from_errno("lseek");
229 		goto done;
230 	}
231 	if (rename(tmpfile_path, packfile_path) == -1) {
232 		err = got_error_from_errno3("rename", tmpfile_path,
233 		    packfile_path);
234 		goto done;
235 	}
236 	free(tmpfile_path);
237 	tmpfile_path = NULL;
238 
239 	*packfile = fdopen(packfd, "w");
240 	if (*packfile == NULL) {
241 		err = got_error_from_errno2("fdopen", tmpfile_path);
242 		goto done;
243 	}
244 	packfd = -1;
245 done:
246 	for (i = 0; i < nours; i++)
247 		free(ours[i]);
248 	free(ours);
249 	for (i = 0; i < ntheirs; i++)
250 		free(theirs[i]);
251 	free(theirs);
252 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
253 		err = got_error_from_errno2("close", packfile_path);
254 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
255 		err = got_error_from_errno("fclose");
256 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
257 		err = got_error_from_errno2("unlink", tmpfile_path);
258 	free(tmpfile_path);
259 	free(packfile_path);
260 	free(sha1_str);
261 	free(path);
262 	if (err) {
263 		free(*pack_hash);
264 		*pack_hash = NULL;
265 		if (*packfile)
266 			fclose(*packfile);
267 		*packfile = NULL;
268 	}
269 	return err;
270 }
271 
272 const struct got_error *
273 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
274     struct got_repository *repo,
275     got_pack_index_progress_cb progress_cb, void *progress_arg,
276     got_cancel_cb cancel_cb, void *cancel_arg)
277 {
278 	size_t i;
279 	char *path;
280 	int imsg_idxfds[2];
281 	int npackfd = -1, idxfd = -1, nidxfd = -1;
282 	int tmpfds[3];
283 	int idxstatus, done = 0;
284 	const struct got_error *err;
285 	struct imsgbuf idxibuf;
286 	pid_t idxpid;
287 	char *tmpidxpath = NULL;
288 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
289 	const char *repo_path = got_repo_get_path_git_dir(repo);
290 	struct stat sb;
291 
292 	for (i = 0; i < nitems(tmpfds); i++)
293 		tmpfds[i] = -1;
294 
295 	if (asprintf(&path, "%s/%s/indexing.idx",
296 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
297 		err = got_error_from_errno("asprintf");
298 		goto done;
299 	}
300 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
301 	free(path);
302 	if (err)
303 		goto done;
304 	if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
305 		err = got_error_from_errno2("fchmod", tmpidxpath);
306 		goto done;
307 	}
308 
309 	nidxfd = dup(idxfd);
310 	if (nidxfd == -1) {
311 		err = got_error_from_errno("dup");
312 		goto done;
313 	}
314 
315 	for (i = 0; i < nitems(tmpfds); i++) {
316 		tmpfds[i] = got_opentempfd();
317 		if (tmpfds[i] == -1) {
318 			err = got_error_from_errno("got_opentempfd");
319 			goto done;
320 		}
321 	}
322 
323 	err = got_object_id_str(&id_str, pack_hash);
324 	if (err)
325 		goto done;
326 
327 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
328 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
329 		err = got_error_from_errno("asprintf");
330 		goto done;
331 	}
332 
333 	if (fstat(fileno(packfile), &sb) == -1) {
334 		err = got_error_from_errno2("fstat", packfile_path);
335 		goto done;
336 	}
337 
338 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
339 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
340 		err = got_error_from_errno("asprintf");
341 		goto done;
342 	}
343 
344 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
345 		err = got_error_from_errno("socketpair");
346 		goto done;
347 	}
348 	idxpid = fork();
349 	if (idxpid == -1) {
350 		err= got_error_from_errno("fork");
351 		goto done;
352 	} else if (idxpid == 0)
353 		got_privsep_exec_child(imsg_idxfds,
354 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
355 	if (close(imsg_idxfds[1]) == -1) {
356 		err = got_error_from_errno("close");
357 		goto done;
358 	}
359 	imsg_init(&idxibuf, imsg_idxfds[0]);
360 
361 	npackfd = dup(fileno(packfile));
362 	if (npackfd == -1) {
363 		err = got_error_from_errno("dup");
364 		goto done;
365 	}
366 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
367 	    npackfd);
368 	if (err != NULL)
369 		goto done;
370 	npackfd = -1;
371 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
372 	if (err != NULL)
373 		goto done;
374 	nidxfd = -1;
375 	for (i = 0; i < nitems(tmpfds); i++) {
376 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
377 		if (err != NULL)
378 			goto done;
379 		tmpfds[i] = -1;
380 	}
381 	done = 0;
382 	while (!done) {
383 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
384 
385 		if (cancel_cb) {
386 			err = cancel_cb(cancel_arg);
387 			if (err)
388 				goto done;
389 		}
390 
391 		err = got_privsep_recv_index_progress(&done, &nobj_total,
392 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
393 		    &idxibuf);
394 		if (err != NULL)
395 			goto done;
396 		if (nobj_indexed != 0) {
397 			err = progress_cb(progress_arg, sb.st_size,
398 			    nobj_total, nobj_indexed, nobj_loose,
399 			    nobj_resolved);
400 			if (err)
401 				break;
402 		}
403 	}
404 	if (close(imsg_idxfds[0]) == -1) {
405 		err = got_error_from_errno("close");
406 		goto done;
407 	}
408 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
409 		err = got_error_from_errno("waitpid");
410 		goto done;
411 	}
412 
413 	if (rename(tmpidxpath, idxpath) == -1) {
414 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
415 		goto done;
416 	}
417 	free(tmpidxpath);
418 	tmpidxpath = NULL;
419 
420 done:
421 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
422 		err = got_error_from_errno2("unlink", tmpidxpath);
423 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
424 		err = got_error_from_errno("close");
425 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
426 		err = got_error_from_errno("close");
427 	for (i = 0; i < nitems(tmpfds); i++) {
428 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
429 			err = got_error_from_errno("close");
430 	}
431 	free(tmpidxpath);
432 	free(idxpath);
433 	free(packfile_path);
434 	return err;
435 }
436 
437 const struct got_error *
438 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
439     struct got_repository *repo, const char *packfile_path)
440 {
441 	const struct got_error *err = NULL;
442 	const char *packdir_path = NULL;
443 	char *packfile_name = NULL, *p, *dot;
444 	struct got_object_id id;
445 	int packfd = -1;
446 
447 	*packfile = NULL;
448 	*pack_hash = NULL;
449 
450 	packdir_path = got_repo_get_path_objects_pack(repo);
451 	if (packdir_path == NULL)
452 		return got_error_from_errno("got_repo_get_path_objects_pack");
453 
454 	if (!got_path_is_child(packfile_path, packdir_path,
455 	    strlen(packdir_path))) {
456 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
457 		goto done;
458 
459 	}
460 
461 	err = got_path_basename(&packfile_name, packfile_path);
462 	if (err)
463 		goto done;
464 	p = packfile_name;
465 
466 	if (strncmp(p, "pack-", 5) != 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 	p += 5;
473 	dot = strchr(p, '.');
474 	if (dot == NULL) {
475 		err = got_error_fmt(GOT_ERR_BAD_PATH,
476 		    "'%s' is not a valid pack file name",
477 		    packfile_name);
478 		goto done;
479 	}
480 	if (strcmp(dot + 1, "pack") != 0) {
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 	*dot = '\0';
487 	if (!got_parse_sha1_digest(id.sha1, p)) {
488 		err = got_error_fmt(GOT_ERR_BAD_PATH,
489 		    "'%s' is not a valid pack file name",
490 		    packfile_name);
491 		goto done;
492 	}
493 
494 	*pack_hash = got_object_id_dup(&id);
495 	if (*pack_hash == NULL) {
496 		err = got_error_from_errno("got_object_id_dup");
497 		goto done;
498 	}
499 
500 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
501 	if (packfd == -1) {
502 		err = got_error_from_errno2("open", packfile_path);
503 		goto done;
504 	}
505 
506 	*packfile = fdopen(packfd, "r");
507 	if (*packfile == NULL) {
508 		err = got_error_from_errno2("fdopen", packfile_path);
509 		goto done;
510 	}
511 	packfd = -1;
512 done:
513 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
514 		err = got_error_from_errno2("close", packfile_path);
515 	free(packfile_name);
516 	if (err) {
517 		free(*pack_hash);
518 		*pack_hash = NULL;
519 	}
520 	return err;
521 }
522 
523 const struct got_error *
524 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
525     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
526     got_cancel_cb cancel_cb, void *cancel_arg)
527 {
528 	const struct got_error *err = NULL;
529 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
530 	struct got_packidx *packidx = NULL;
531 	struct got_pack *pack = NULL;
532 	uint32_t nobj, i;
533 
534 	err = got_object_id_str(&id_str, pack_hash);
535 	if (err)
536 		goto done;
537 
538 	if (asprintf(&packpath, "%s/pack-%s.pack",
539 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
540 		err = got_error_from_errno("asprintf");
541 		goto done;
542 	}
543 	if (asprintf(&idxpath, "%s/pack-%s.idx",
544 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
545 		err = got_error_from_errno("asprintf");
546 		goto done;
547 	}
548 
549 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
550 	if (err)
551 		goto done;
552 
553 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
554 	if (err)
555 		goto done;
556 
557 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
558 	for (i = 0; i < nobj; i++) {
559 		struct got_packidx_object_id *oid;
560 		struct got_object_id id, base_id;
561 		off_t offset, base_offset = 0;
562 		uint8_t type;
563 		uint64_t size;
564 		size_t tslen, len;
565 
566 		if (cancel_cb) {
567 			err = cancel_cb(cancel_arg);
568 			if (err)
569 				break;
570 		}
571 		oid = &packidx->hdr.sorted_ids[i];
572 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
573 
574 		offset = got_packidx_get_object_offset(packidx, i);
575 		if (offset == -1) {
576 			err = got_error(GOT_ERR_BAD_PACKIDX);
577 			goto done;
578 		}
579 
580 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
581 		    pack, offset);
582 		if (err)
583 			goto done;
584 
585 		switch (type) {
586 		case GOT_OBJ_TYPE_OFFSET_DELTA:
587 			err = got_pack_parse_offset_delta(&base_offset, &len,
588 			    pack, offset, tslen);
589 			if (err)
590 				goto done;
591 			break;
592 		case GOT_OBJ_TYPE_REF_DELTA:
593 			err = got_pack_parse_ref_delta(&base_id,
594 			    pack, offset, tslen);
595 			if (err)
596 				goto done;
597 			break;
598 		}
599 		err = (*list_cb)(list_arg, &id, type, offset, size,
600 		    base_offset, &base_id);
601 		if (err)
602 			goto done;
603 	}
604 
605 done:
606 	free(id_str);
607 	free(idxpath);
608 	free(packpath);
609 	if (packidx)
610 		got_packidx_close(packidx);
611 	return err;
612 }
613 
614 static const struct got_error *
615 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
616     void *progress_arg, struct got_ratelimit *rl,
617     int nloose, int ncommits, int npurged)
618 {
619 	const struct got_error *err;
620 	int elapsed;
621 
622 	if (progress_cb == NULL)
623 		return NULL;
624 
625 	err = got_ratelimit_check(&elapsed, rl);
626 	if (err || !elapsed)
627 		return err;
628 
629 	return progress_cb(progress_arg, nloose, ncommits, npurged);
630 }
631 
632 static const struct got_error *
633 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
634     got_cleanup_progress_cb progress_cb, void *progress_arg,
635     struct got_ratelimit *rl, struct got_repository *repo)
636 {
637 	const struct got_error *err = NULL;
638 	char *path_objects = NULL, *path = NULL;
639 	DIR *dir = NULL;
640 	struct got_object *obj = NULL;
641 	struct got_object_id id;
642 	int i, fd = -1;
643 	struct stat sb;
644 
645 	*ondisk_size = 0;
646 	*loose_ids = got_object_idset_alloc();
647 	if (*loose_ids == NULL)
648 		return got_error_from_errno("got_object_idset_alloc");
649 
650 	path_objects = got_repo_get_path_objects(repo);
651 	if (path_objects == NULL) {
652 		err = got_error_from_errno("got_repo_get_path_objects");
653 		goto done;
654 	}
655 
656 	for (i = 0; i <= 0xff; i++) {
657 		struct dirent *dent;
658 
659 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
660 			err = got_error_from_errno("asprintf");
661 			break;
662 		}
663 
664 		dir = opendir(path);
665 		if (dir == NULL) {
666 			if (errno == ENOENT) {
667 				err = NULL;
668 				continue;
669 			}
670 			err = got_error_from_errno2("opendir", path);
671 			break;
672 		}
673 
674 		while ((dent = readdir(dir)) != NULL) {
675 			char *id_str;
676 
677 			if (strcmp(dent->d_name, ".") == 0 ||
678 			    strcmp(dent->d_name, "..") == 0)
679 				continue;
680 
681 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
682 				err = got_error_from_errno("asprintf");
683 				goto done;
684 			}
685 
686 			memset(&id, 0, sizeof(id));
687 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
688 				free(id_str);
689 				continue;
690 			}
691 			free(id_str);
692 
693 			err = got_object_open_loose_fd(&fd, &id, repo);
694 			if (err)
695 				goto done;
696 			if (fstat(fd, &sb) == -1) {
697 				err = got_error_from_errno("fstat");
698 				goto done;
699 			}
700 			err = got_object_read_header_privsep(&obj, &id, repo,
701 			    fd);
702 			if (err)
703 				goto done;
704 			fd = -1; /* already closed */
705 
706 			switch (obj->type) {
707 			case GOT_OBJ_TYPE_COMMIT:
708 			case GOT_OBJ_TYPE_TREE:
709 			case GOT_OBJ_TYPE_BLOB:
710 			case GOT_OBJ_TYPE_TAG:
711 				break;
712 			default:
713 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
714 				    "%d", obj->type);
715 				goto done;
716 			}
717 			got_object_close(obj);
718 			obj = NULL;
719 			(*ondisk_size) += sb.st_size;
720 			err = got_object_idset_add(*loose_ids, &id, NULL);
721 			if (err)
722 				goto done;
723 			err = report_cleanup_progress(progress_cb,
724 			    progress_arg, rl,
725 			    got_object_idset_num_elements(*loose_ids), -1, -1);
726 			if (err)
727 				goto done;
728 		}
729 
730 		if (closedir(dir) != 0) {
731 			err = got_error_from_errno("closedir");
732 			goto done;
733 		}
734 		dir = NULL;
735 
736 		free(path);
737 		path = NULL;
738 	}
739 done:
740 	if (dir && closedir(dir) != 0 && err == NULL)
741 		err = got_error_from_errno("closedir");
742 	if (fd != -1 && close(fd) == -1 && err == NULL)
743 		err = got_error_from_errno("close");
744 	if (err) {
745 		got_object_idset_free(*loose_ids);
746 		*loose_ids = NULL;
747 	}
748 	if (obj)
749 		got_object_close(obj);
750 	free(path_objects);
751 	free(path);
752 	return err;
753 }
754 
755 static const struct got_error *
756 preserve_loose_object(struct got_object_idset *loose_ids,
757     struct got_object_id *id, struct got_repository *repo, int *npacked)
758 {
759 	const struct got_error *err = NULL;
760 	struct got_object *obj;
761 
762 	if (!got_object_idset_contains(loose_ids, id))
763 		return NULL;
764 
765 	/*
766 	 * Try to open this object from a pack file. This ensures that
767 	 * we do in fact have a valid packed copy of the object. Otherwise
768 	 * we should not delete the loose representation of this object.
769 	 */
770 	err = got_object_open_packed(&obj, id, repo);
771 	if (err == NULL) {
772 		got_object_close(obj);
773 		/*
774 		 * The object is referenced and packed.
775 		 * We can purge the redundantly stored loose object.
776 		 */
777 		(*npacked)++;
778 		return NULL;
779 	} else if (err->code != GOT_ERR_NO_OBJ)
780 		return err;
781 
782 	/*
783 	 * This object is referenced and not packed.
784 	 * Remove it from our purge set.
785 	 */
786 	return got_object_idset_remove(NULL, loose_ids, id);
787 }
788 
789 static const struct got_error *
790 load_tree_entries(struct got_object_id_queue *ids,
791     struct got_object_idset *loose_ids,
792     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
793     const char *dpath, struct got_repository *repo, int *npacked,
794     got_cancel_cb cancel_cb, void *cancel_arg)
795 {
796 	const struct got_error *err;
797 	struct got_tree_object *tree;
798 	char *p = NULL;
799 	int i;
800 
801 	err = got_object_open_as_tree(&tree, repo, tree_id);
802 	if (err)
803 		return err;
804 
805 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
806 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
807 		struct got_object_id *id = got_tree_entry_get_id(e);
808 		mode_t mode = got_tree_entry_get_mode(e);
809 
810 		if (cancel_cb) {
811 			err = (*cancel_cb)(cancel_arg);
812 			if (err)
813 				break;
814 		}
815 
816 		if (got_object_tree_entry_is_symlink(e) ||
817 		    got_object_tree_entry_is_submodule(e) ||
818 		    got_object_idset_contains(traversed_ids, id))
819 			continue;
820 
821 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
822 		    got_tree_entry_get_name(e)) == -1) {
823 			err = got_error_from_errno("asprintf");
824 			break;
825 		}
826 
827 		if (S_ISDIR(mode)) {
828 			struct got_object_qid *qid;
829 			err = got_object_qid_alloc(&qid, id);
830 			if (err)
831 				break;
832 			STAILQ_INSERT_TAIL(ids, qid, entry);
833 		} else if (S_ISREG(mode)) {
834 			/* This blob is referenced. */
835 			err = preserve_loose_object(loose_ids, id, repo,
836 			    npacked);
837 			if (err)
838 				break;
839 			err = got_object_idset_add(traversed_ids, id, NULL);
840 			if (err)
841 				break;
842 		}
843 		free(p);
844 		p = NULL;
845 	}
846 
847 	got_object_tree_close(tree);
848 	free(p);
849 	return err;
850 }
851 
852 static const struct got_error *
853 load_tree(struct got_object_idset *loose_ids,
854     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
855     const char *dpath, struct got_repository *repo, int *npacked,
856     got_cancel_cb cancel_cb, void *cancel_arg)
857 {
858 	const struct got_error *err = NULL;
859 	struct got_object_id_queue tree_ids;
860 	struct got_object_qid *qid;
861 
862 	err = got_object_qid_alloc(&qid, tree_id);
863 	if (err)
864 		return err;
865 
866 	STAILQ_INIT(&tree_ids);
867 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
868 
869 	while (!STAILQ_EMPTY(&tree_ids)) {
870 		if (cancel_cb) {
871 			err = (*cancel_cb)(cancel_arg);
872 			if (err)
873 				break;
874 		}
875 
876 		qid = STAILQ_FIRST(&tree_ids);
877 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
878 
879 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
880 			got_object_qid_free(qid);
881 			continue;
882 		}
883 
884 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
885 		if (err) {
886 			got_object_qid_free(qid);
887 			break;
888 		}
889 
890 		/* This tree is referenced. */
891 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
892 		if (err)
893 			break;
894 
895 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
896 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
897 		got_object_qid_free(qid);
898 		if (err)
899 			break;
900 	}
901 
902 	got_object_id_queue_free(&tree_ids);
903 	return err;
904 }
905 
906 static const struct got_error *
907 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
908     int *npacked, struct got_object_idset *traversed_ids,
909     struct got_object_id *id, struct got_repository *repo,
910     got_cleanup_progress_cb progress_cb, void *progress_arg,
911     struct got_ratelimit *rl, int nloose,
912     got_cancel_cb cancel_cb, void *cancel_arg)
913 {
914 	const struct got_error *err;
915 	struct got_commit_object *commit = NULL;
916 	struct got_tag_object *tag = NULL;
917 	struct got_object_id *tree_id = NULL;
918 	struct got_object_id_queue ids;
919 	struct got_object_qid *qid;
920 	int obj_type;
921 
922 	err = got_object_qid_alloc(&qid, id);
923 	if (err)
924 		return err;
925 
926 	STAILQ_INIT(&ids);
927 	STAILQ_INSERT_TAIL(&ids, qid, entry);
928 
929 	while (!STAILQ_EMPTY(&ids)) {
930 		if (cancel_cb) {
931 			err = (*cancel_cb)(cancel_arg);
932 			if (err)
933 				break;
934 		}
935 
936 		qid = STAILQ_FIRST(&ids);
937 		STAILQ_REMOVE_HEAD(&ids, entry);
938 
939 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
940 			got_object_qid_free(qid);
941 			qid = NULL;
942 			continue;
943 		}
944 
945 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
946 		if (err)
947 			break;
948 
949 		/* This commit or tag is referenced. */
950 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
951 		if (err)
952 			break;
953 
954 		err = got_object_get_type(&obj_type, repo, &qid->id);
955 		if (err)
956 			break;
957 		switch (obj_type) {
958 		case GOT_OBJ_TYPE_COMMIT:
959 			err = got_object_open_as_commit(&commit, repo,
960 			    &qid->id);
961 			if (err)
962 				goto done;
963 			break;
964 		case GOT_OBJ_TYPE_TAG:
965 			err = got_object_open_as_tag(&tag, repo, &qid->id);
966 			if (err)
967 				goto done;
968 			break;
969 		default:
970 			/* should not happen */
971 			err = got_error(GOT_ERR_OBJ_TYPE);
972 			goto done;
973 		}
974 
975 		/* Find a tree object to scan. */
976 		if (commit) {
977 			tree_id = got_object_commit_get_tree_id(commit);
978 		} else if (tag) {
979 			obj_type = got_object_tag_get_object_type(tag);
980 			switch (obj_type) {
981 			case GOT_OBJ_TYPE_COMMIT:
982 				err = got_object_open_as_commit(&commit, repo,
983 				    got_object_tag_get_object_id(tag));
984 				if (err)
985 					goto done;
986 				tree_id = got_object_commit_get_tree_id(commit);
987 				break;
988 			case GOT_OBJ_TYPE_TREE:
989 				tree_id = got_object_tag_get_object_id(tag);
990 				break;
991 			default:
992 				/*
993 				 * Tag points at something other than a
994 				 * commit or tree. Leave this weird tag object
995 				 * and the object it points to on disk.
996 				 */
997 				err = got_object_idset_remove(NULL, loose_ids,
998 				    &qid->id);
999 				if (err && err->code != GOT_ERR_NO_OBJ)
1000 					goto done;
1001 				err = got_object_idset_remove(NULL, loose_ids,
1002 				    got_object_tag_get_object_id(tag));
1003 				if (err && err->code != GOT_ERR_NO_OBJ)
1004 					goto done;
1005 				err = NULL;
1006 				break;
1007 			}
1008 		}
1009 
1010 		if (tree_id) {
1011 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1012 			    repo, npacked, cancel_cb, cancel_arg);
1013 			if (err)
1014 				break;
1015 		}
1016 
1017 		if (commit || tag)
1018 			(*ncommits)++; /* scanned tags are counted as commits */
1019 
1020 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1021 		    nloose, *ncommits, -1);
1022 		if (err)
1023 			break;
1024 
1025 		if (commit) {
1026 			/* Find parent commits to scan. */
1027 			const struct got_object_id_queue *parent_ids;
1028 			parent_ids = got_object_commit_get_parent_ids(commit);
1029 			err = got_object_id_queue_copy(parent_ids, &ids);
1030 			if (err)
1031 				break;
1032 			got_object_commit_close(commit);
1033 			commit = NULL;
1034 		}
1035 		if (tag) {
1036 			got_object_tag_close(tag);
1037 			tag = NULL;
1038 		}
1039 		got_object_qid_free(qid);
1040 		qid = NULL;
1041 	}
1042 done:
1043 	if (qid)
1044 		got_object_qid_free(qid);
1045 	if (commit)
1046 		got_object_commit_close(commit);
1047 	if (tag)
1048 		got_object_tag_close(tag);
1049 	got_object_id_queue_free(&ids);
1050 	return err;
1051 }
1052 
1053 struct purge_loose_object_arg {
1054 	struct got_repository *repo;
1055 	got_cleanup_progress_cb progress_cb;
1056 	void *progress_arg;
1057 	struct got_ratelimit *rl;
1058 	int nloose;
1059 	int ncommits;
1060 	int npurged;
1061 	off_t size_purged;
1062 	int dry_run;
1063 	time_t max_mtime;
1064 	int ignore_mtime;
1065 };
1066 
1067 static const struct got_error *
1068 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1069 {
1070 	struct purge_loose_object_arg *a = arg;
1071 	const struct got_error *err, *unlock_err = NULL;
1072 	char *path = NULL;
1073 	int fd = -1;
1074 	struct stat sb;
1075 	struct got_lockfile *lf = NULL;
1076 
1077 	err = got_object_get_path(&path, id, a->repo);
1078 	if (err)
1079 		return err;
1080 
1081 	err = got_object_open_loose_fd(&fd, id, a->repo);
1082 	if (err)
1083 		goto done;
1084 
1085 	if (fstat(fd, &sb) == -1) {
1086 		err = got_error_from_errno("fstat");
1087 		goto done;
1088 	}
1089 
1090 	/*
1091 	 * Do not delete objects which are younger than our maximum
1092 	 * modification time threshold. This prevents a race where
1093 	 * new objects which are being added to the repository
1094 	 * concurrently would be deleted.
1095 	 */
1096 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1097 		if (!a->dry_run) {
1098 			err = got_lockfile_lock(&lf, path, -1);
1099 			if (err)
1100 				goto done;
1101 			if (unlink(path) == -1) {
1102 				err = got_error_from_errno2("unlink", path);
1103 				goto done;
1104 			}
1105 		}
1106 
1107 		a->npurged++;
1108 		a->size_purged += sb.st_size;
1109 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1110 		    a->rl, a->nloose, a->ncommits, a->npurged);
1111 		if (err)
1112 			goto done;
1113 	}
1114 done:
1115 	if (fd != -1 && close(fd) == -1 && err == NULL)
1116 		err = got_error_from_errno("close");
1117 	free(path);
1118 	if (lf)
1119 		unlock_err = got_lockfile_unlock(lf, -1);
1120 	return err ? err : unlock_err;
1121 }
1122 
1123 const struct got_error *
1124 got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1125     off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1126     int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1127     got_cancel_cb cancel_cb, void *cancel_arg)
1128 {
1129 	const struct got_error *err;
1130 	struct got_object_idset *loose_ids;
1131 	struct got_object_idset *traversed_ids;
1132 	struct got_object_id **referenced_ids;
1133 	int i, nreferenced, nloose, ncommits = 0;
1134 	struct got_reflist_head refs;
1135 	struct got_reflist_entry *re;
1136 	struct purge_loose_object_arg arg;
1137 	time_t max_mtime = 0;
1138 	struct got_ratelimit rl;
1139 
1140 	TAILQ_INIT(&refs);
1141 	got_ratelimit_init(&rl, 0, 500);
1142 
1143 	*size_before = 0;
1144 	*size_after = 0;
1145 	*npacked = 0;
1146 
1147 	err = get_loose_object_ids(&loose_ids, size_before,
1148 	    progress_cb, progress_arg, &rl, repo);
1149 	if (err)
1150 		return err;
1151 	nloose = got_object_idset_num_elements(loose_ids);
1152 	if (nloose == 0) {
1153 		got_object_idset_free(loose_ids);
1154 		if (progress_cb) {
1155 			err = progress_cb(progress_arg, 0, 0, 0);
1156 			if (err)
1157 				return err;
1158 		}
1159 		return NULL;
1160 	}
1161 
1162 	traversed_ids = got_object_idset_alloc();
1163 	if (traversed_ids == NULL) {
1164 		err = got_error_from_errno("got_object_idset_alloc");
1165 		goto done;
1166 	}
1167 
1168 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1169 	if (err)
1170 		goto done;
1171 	if (!ignore_mtime) {
1172 		TAILQ_FOREACH(re, &refs, entry) {
1173 			time_t mtime = got_ref_get_mtime(re->ref);
1174 			if (mtime > max_mtime)
1175 				max_mtime = mtime;
1176 		}
1177 		/*
1178 		 * For safety, keep objects created within 10 minutes
1179 		 * before the youngest reference was created.
1180 		 */
1181 		if (max_mtime >= 600)
1182 			max_mtime -= 600;
1183 	}
1184 
1185 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1186 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1187 	    &refs, repo, cancel_cb, cancel_arg);
1188 	if (err)
1189 		goto done;
1190 
1191 	for (i = 0; i < nreferenced; i++) {
1192 		struct got_object_id *id = referenced_ids[i];
1193 		err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1194 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1195 		    nloose, cancel_cb, cancel_arg);
1196 		if (err)
1197 			goto done;
1198 	}
1199 
1200 	/* Any remaining loose objects are unreferenced and can be purged. */
1201 	arg.repo = repo;
1202 	arg.progress_arg = progress_arg;
1203 	arg.progress_cb = progress_cb;
1204 	arg.rl = &rl;
1205 	arg.nloose = nloose;
1206 	arg.npurged = 0;
1207 	arg.size_purged = 0;
1208 	arg.ncommits = ncommits;
1209 	arg.dry_run = dry_run;
1210 	arg.max_mtime = max_mtime;
1211 	arg.ignore_mtime = ignore_mtime;
1212 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1213 	if (err)
1214 		goto done;
1215 	*size_after = *size_before - arg.size_purged;
1216 
1217 	/* Produce a final progress report. */
1218 	if (progress_cb) {
1219 		err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1220 		if (err)
1221 			goto done;
1222 	}
1223 done:
1224 	got_object_idset_free(loose_ids);
1225 	got_object_idset_free(traversed_ids);
1226 	return err;
1227 }
1228 
1229 static const struct got_error *
1230 remove_packidx(int dir_fd, const char *relpath)
1231 {
1232 	const struct got_error *err, *unlock_err;
1233 	struct got_lockfile *lf;
1234 
1235 	err = got_lockfile_lock(&lf, relpath, dir_fd);
1236 	if (err)
1237 		return err;
1238 	if (unlinkat(dir_fd, relpath, 0) == -1)
1239 		err = got_error_from_errno("unlinkat");
1240 	unlock_err = got_lockfile_unlock(lf, dir_fd);
1241 	return err ? err : unlock_err;
1242 }
1243 
1244 const struct got_error *
1245 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1246     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1247     got_cancel_cb cancel_cb, void *cancel_arg)
1248 {
1249 	const struct got_error *err = NULL;
1250 	DIR *packdir = NULL;
1251 	struct dirent *dent;
1252 	char *pack_relpath = NULL;
1253 	int packdir_fd;
1254 	struct stat sb;
1255 
1256 	packdir_fd = openat(got_repo_get_fd(repo),
1257 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1258 	if (packdir_fd == -1) {
1259 		if (errno == ENOENT)
1260 			return NULL;
1261 		return got_error_from_errno_fmt("openat: %s/%s",
1262 		    got_repo_get_path_git_dir(repo),
1263 		    GOT_OBJECTS_PACK_DIR);
1264 	}
1265 
1266 	packdir = fdopendir(packdir_fd);
1267 	if (packdir == NULL) {
1268 		err = got_error_from_errno("fdopendir");
1269 		goto done;
1270 	}
1271 
1272 	while ((dent = readdir(packdir)) != NULL) {
1273 		if (cancel_cb) {
1274 			err = cancel_cb(cancel_arg);
1275 			if (err)
1276 				goto done;
1277 		}
1278 
1279 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1280 			continue;
1281 
1282 		err = got_packidx_get_packfile_path(&pack_relpath,
1283 		    dent->d_name);
1284 		if (err)
1285 			goto done;
1286 
1287 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1288 			free(pack_relpath);
1289 			pack_relpath = NULL;
1290 			continue;
1291 		}
1292 		if (errno != ENOENT) {
1293 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1294 			    got_repo_get_path_git_dir(repo),
1295 			    GOT_OBJECTS_PACK_DIR,
1296 			    pack_relpath);
1297 			goto done;
1298 		}
1299 
1300 		if (!dry_run) {
1301 			err = remove_packidx(packdir_fd, dent->d_name);
1302 			if (err)
1303 				goto done;
1304 		}
1305 		if (progress_cb) {
1306 			char *path;
1307 			if (asprintf(&path, "%s/%s/%s",
1308 			    got_repo_get_path_git_dir(repo),
1309 			    GOT_OBJECTS_PACK_DIR,
1310 			    dent->d_name) == -1) {
1311 				err = got_error_from_errno("asprintf");
1312 				goto done;
1313 			}
1314 			err = progress_cb(progress_arg, path);
1315 			free(path);
1316 			if (err)
1317 				goto done;
1318 		}
1319 		free(pack_relpath);
1320 		pack_relpath = NULL;
1321 	}
1322 done:
1323 	if (packdir && closedir(packdir) != 0 && err == NULL)
1324 		err = got_error_from_errno("closedir");
1325 	free(pack_relpath);
1326 	return err;
1327 }