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