Got Mirror

Blob

Date:: Fri Sep 24 22:01:09 2021 UTC
Message:: fix 'got send' with tree objects which contain symlinks; reported by Omar
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 #include <sys/uio.h>
20 #include <sys/stat.h>
21 
22 #include <stdint.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <sha1.h>
27 #include <limits.h>
28 #include <zlib.h>
29 
30 #include "got_error.h"
31 #include "got_cancel.h"
32 #include "got_object.h"
33 #include "got_reference.h"
34 #include "got_repository_admin.h"
35 
36 #include "got_lib_deltify.h"
37 #include "got_lib_delta.h"
38 #include "got_lib_object.h"
39 #include "got_lib_object_idset.h"
40 #include "got_lib_object_cache.h"
41 #include "got_lib_deflate.h"
42 #include "got_lib_pack.h"
43 #include "got_lib_privsep.h"
44 #include "got_lib_repository.h"
45 
46 #ifndef MAX
47 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
48 #endif
49 
50 struct got_pack_meta {
51 	struct got_object_id id;
52 	char	*path;
53 	int	obj_type;
54 	time_t	mtime;
55 
56 	/* The best delta we picked */
57 	struct got_pack_meta *head;
58 	struct got_pack_meta *prev;
59 	struct got_delta_instruction *deltas;
60 	int	ndeltas;
61 	int	nchain;
62 
63 	/* Only used for delta window */
64 	struct got_delta_table *dtab;
65 
66 	/* Only used for writing offset deltas */
67 	off_t	off;
68 };
69 
70 struct got_pack_metavec {
71 	struct got_pack_meta **meta;
72 	int nmeta;
73 	int metasz;
74 };
75 
76 static const struct got_error *
77 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
78     const char *path, int obj_type, time_t mtime)
79 {
80 	const struct got_error *err = NULL;
81 	struct got_pack_meta *m;
82 
83 	*new = NULL;
84 
85 	m = calloc(1, sizeof(*m));
86 	if (m == NULL)
87 		return got_error_from_errno("malloc");
88 
89 	memcpy(&m->id, id, sizeof(m->id));
90 
91 	m->path = strdup(path);
92 	if (m->path == NULL) {
93 		err = got_error_from_errno("strdup");
94 		free(m);
95 		return err;
96 	}
97 
98 	m->obj_type = obj_type;
99 	m->mtime = mtime;
100 	*new = m;
101 	return NULL;
102 }
103 
104 static void
105 clear_meta(struct got_pack_meta *meta)
106 {
107 	if (meta == NULL)
108 		return;
109 	free(meta->deltas);
110 	meta->deltas = NULL;
111 	free(meta->path);
112 	meta->path = NULL;
113 }
114 
115 static void
116 free_nmeta(struct got_pack_meta **meta, int nmeta)
117 {
118 	int i;
119 
120 	for (i = 0; i < nmeta; i++)
121 		clear_meta(meta[i]);
122 	free(meta);
123 }
124 
125 static int
126 delta_order_cmp(const void *pa, const void *pb)
127 {
128 	struct got_pack_meta *a, *b;
129 	int cmp;
130 
131 	a = *(struct got_pack_meta **)pa;
132 	b = *(struct got_pack_meta **)pb;
133 
134 	if (a->obj_type != b->obj_type)
135 		return a->obj_type - b->obj_type;
136 	cmp = strcmp(a->path, b->path);
137 	if (cmp != 0)
138 		return cmp;
139 	if (a->mtime != b->mtime)
140 		return a->mtime - b->mtime;
141 	return got_object_id_cmp(&a->id, &b->id);
142 }
143 
144 static int
145 delta_size(struct got_delta_instruction *deltas, int ndeltas)
146 {
147 	int i, size = 32;
148 	for (i = 0; i < ndeltas; i++) {
149 		if (deltas[i].copy)
150 			size += GOT_DELTA_SIZE_SHIFT;
151 		else
152 			size += deltas[i].len + 1;
153 	}
154 	return size;
155 }
156 
157 
158 static const struct got_error *
159 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
160     struct got_repository *repo,
161     got_pack_progress_cb progress_cb, void *progress_arg,
162     got_cancel_cb cancel_cb, void *cancel_arg)
163 {
164 	const struct got_error *err = NULL;
165 	struct got_pack_meta *m = NULL, *base = NULL;
166 	struct got_raw_object *raw = NULL, *base_raw = NULL;
167 	struct got_delta_instruction *deltas;
168 	int i, j, size, ndeltas, best;
169 	const int max_base_candidates = 10;
170 
171 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
172 	for (i = 0; i < nmeta; i++) {
173 		if (cancel_cb) {
174 			err = (*cancel_cb)(cancel_arg);
175 			if (err)
176 				break;
177 		}
178 		if (progress_cb) {
179 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
180 			if (err)
181 				goto done;
182 		}
183 		m = meta[i];
184 		m->deltas = NULL;
185 		m->ndeltas = 0;
186 
187 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
188 		    m->obj_type == GOT_OBJ_TYPE_TAG)
189 			continue;
190 
191 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
192 		if (err)
193 			goto done;
194 
195 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
196 		    raw->size + raw->hdrlen);
197 		if (err)
198 			goto done;
199 
200 		if (i > max_base_candidates) {
201 			struct got_pack_meta *n = NULL;
202 			n = meta[i - (max_base_candidates + 1)];
203 			got_deltify_free(n->dtab);
204 			n->dtab = NULL;
205 		}
206 
207 		best = raw->size;
208 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
209 			if (cancel_cb) {
210 				err = (*cancel_cb)(cancel_arg);
211 				if (err)
212 					goto done;
213 			}
214 			base = meta[j];
215 			/* long chains make unpacking slow, avoid such bases */
216 			if (base->nchain >= 128 ||
217 			    base->obj_type != m->obj_type)
218 				continue;
219 
220 			err = got_object_raw_open(&base_raw, repo, &base->id,
221 			    8192);
222 			if (err)
223 				goto done;
224 			err = got_deltify(&deltas, &ndeltas,
225 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
226 			    base->dtab, base_raw->f, base_raw->hdrlen,
227 			    base_raw->size + base_raw->hdrlen);
228 			got_object_raw_close(base_raw);
229 			base_raw = NULL;
230 			if (err)
231 				goto done;
232 
233 			size = delta_size(deltas, ndeltas);
234 			if (size + 32 < best){
235 				/*
236 				 * if we already picked a best delta,
237 				 * replace it.
238 				 */
239 				free(m->deltas);
240 				best = size;
241 				m->deltas = deltas;
242 				m->ndeltas = ndeltas;
243 				m->nchain = base->nchain + 1;
244 				m->prev = base;
245 				m->head = base->head;
246 				if (m->head == NULL)
247 					m->head = base;
248 			} else {
249 				free(deltas);
250 				deltas = NULL;
251 				ndeltas = 0;
252 			}
253 		}
254 
255 		got_object_raw_close(raw);
256 		raw = NULL;
257 	}
258 done:
259 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
260 		got_deltify_free(meta[i]->dtab);
261 		meta[i]->dtab = NULL;
262 	}
263 	if (raw)
264 		got_object_raw_close(raw);
265 	if (base_raw)
266 		got_object_raw_close(base_raw);
267 	return err;
268 }
269 
270 static const struct got_error *
271 search_packidx(int *found, struct got_object_id *id,
272     struct got_repository *repo)
273 {
274 	const struct got_error *err = NULL;
275 	struct got_packidx *packidx = NULL;
276 	int idx;
277 
278 	*found = 0;
279 
280 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
281 	if (err == NULL)
282 		*found = 1; /* object is already packed */
283 	else if (err->code == GOT_ERR_NO_OBJ)
284 		err = NULL;
285 	return err;
286 }
287 
288 static const int obj_types[] = {
289 	GOT_OBJ_TYPE_ANY,
290 	GOT_OBJ_TYPE_COMMIT,
291 	GOT_OBJ_TYPE_TREE,
292 	GOT_OBJ_TYPE_BLOB,
293 	GOT_OBJ_TYPE_TAG,
294 	GOT_OBJ_TYPE_OFFSET_DELTA,
295 	GOT_OBJ_TYPE_REF_DELTA
296 };
297 
298 static const struct got_error *
299 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
300     struct got_object_id *id, const char *path, int obj_type,
301     time_t mtime, int loose_obj_only, struct got_repository *repo)
302 {
303 	const struct got_error *err;
304 	struct got_pack_meta *m;
305 
306 	if (loose_obj_only) {
307 		int is_packed;
308 		err = search_packidx(&is_packed, id, repo);
309 		if (err)
310 			return err;
311 		if (is_packed)
312 			return NULL;
313 	}
314 
315 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
316 	if (err)
317 		return err;
318 
319 	if (v == NULL)
320 		return NULL;
321 
322 	err = alloc_meta(&m, id, path, obj_type, mtime);
323 	if (err)
324 		goto done;
325 
326 	if (v->nmeta == v->metasz){
327 		size_t newsize = 2 * v->metasz;
328 		struct got_pack_meta **new;
329 		new = reallocarray(v->meta, newsize, sizeof(*new));
330 		if (new == NULL) {
331 			err = got_error_from_errno("reallocarray");
332 			goto done;
333 		}
334 		v->meta = new;
335 		v->metasz = newsize; 
336 	}
337 done:
338 	if (err) {
339 		clear_meta(m);
340 		free(m);
341 	} else
342 		v->meta[v->nmeta++] = m;
343 
344 	return err;
345 }
346 
347 static const struct got_error *
348 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
349     struct got_object_idset *idset, struct got_object_id *tree_id,
350     const char *dpath, time_t mtime, struct got_repository *repo,
351     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
352 {
353 	const struct got_error *err;
354 	struct got_tree_object *tree;
355 	char *p = NULL;
356 	int i;
357 
358 	err = got_object_open_as_tree(&tree, repo, tree_id);
359 	if (err)
360 		return err;
361 
362 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
363 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
364 		struct got_object_id *id = got_tree_entry_get_id(e);
365 		mode_t mode = got_tree_entry_get_mode(e);
366 
367 		if (cancel_cb) {
368 			err = (*cancel_cb)(cancel_arg);
369 			if (err)
370 				break;
371 		}
372 
373 		if (got_object_tree_entry_is_submodule(e) ||
374 		    got_object_idset_contains(idset, id))
375 			continue;
376 		
377 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
378 		    got_tree_entry_get_name(e)) == -1) {
379 			err = got_error_from_errno("asprintf");
380 			break;
381 		}
382 
383 		if (S_ISDIR(mode)) {
384 			struct got_object_qid *qid;
385 			err = got_object_qid_alloc(&qid, id);
386 			if (err)
387 				break;
388 			STAILQ_INSERT_TAIL(ids, qid, entry);
389 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
390 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
391 			    mtime, loose_obj_only, repo);
392 			if (err)
393 				break;
394 		}
395 		free(p);
396 		p = NULL;
397 	}
398 
399 	got_object_tree_close(tree);
400 	free(p);
401 	return err;
402 }
403 
404 static const struct got_error *
405 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
406     struct got_object_id *tree_id, const char *dpath, time_t mtime,
407     int loose_obj_only, struct got_repository *repo,
408     got_cancel_cb cancel_cb, void *cancel_arg)
409 {
410 	const struct got_error *err = NULL;
411 	struct got_object_id_queue tree_ids;
412 	struct got_object_qid *qid;
413 
414 	if (got_object_idset_contains(idset, tree_id))
415 		return NULL;
416 
417 	err = got_object_qid_alloc(&qid, tree_id);
418 	if (err)
419 		return err;
420 
421 	STAILQ_INIT(&tree_ids);
422 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
423 
424 	while (!STAILQ_EMPTY(&tree_ids)) {
425 		if (cancel_cb) {
426 			err = (*cancel_cb)(cancel_arg);
427 			if (err)
428 				break;
429 		}
430 
431 		qid = STAILQ_FIRST(&tree_ids);
432 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
433 
434 		if (got_object_idset_contains(idset, qid->id)) {
435 			got_object_qid_free(qid);
436 			continue;
437 		}
438 
439 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
440 		    mtime, loose_obj_only, repo);
441 		if (err) {
442 			got_object_qid_free(qid);
443 			break;
444 		}
445 
446 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
447 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
448 		got_object_qid_free(qid);
449 		if (err)
450 			break;
451 	}
452 
453 	got_object_id_queue_free(&tree_ids);
454 	return err;
455 }
456 
457 static const struct got_error *
458 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
459     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
460     got_cancel_cb cancel_cb, void *cancel_arg)
461 {
462 	const struct got_error *err;
463 	struct got_commit_object *commit;
464 
465 	if (got_object_idset_contains(idset, id))
466 		return NULL;
467 
468 	if (loose_obj_only) {
469 		int is_packed;
470 		err = search_packidx(&is_packed, id, repo);
471 		if (err)
472 			return err;
473 		if (is_packed)
474 			return NULL;
475 	}
476 
477 	err = got_object_open_as_commit(&commit, repo, id);
478 	if (err)
479 		return err;
480 
481 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
482 	    got_object_commit_get_committer_time(commit),
483 	    loose_obj_only, repo);
484 	if (err)
485 		goto done;
486 
487 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
488 	    "", got_object_commit_get_committer_time(commit),
489 	    loose_obj_only, repo, cancel_cb, cancel_arg);
490 done:
491 	got_object_commit_close(commit);
492 	return err;
493 }
494 
495 static const struct got_error *
496 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
497     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
498     got_cancel_cb cancel_cb, void *cancel_arg)
499 {
500 	const struct got_error *err;
501 	struct got_tag_object *tag = NULL;
502 
503 	if (got_object_idset_contains(idset, id))
504 		return NULL;
505 
506 	if (loose_obj_only) {
507 		int is_packed;
508 		err = search_packidx(&is_packed, id, repo);
509 		if (err)
510 			return err;
511 		if (is_packed)
512 			return NULL;
513 	}
514 
515 	err = got_object_open_as_tag(&tag, repo, id);
516 	if (err)
517 		return err;
518 
519 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
520 	    got_object_tag_get_tagger_time(tag),
521 	    loose_obj_only, repo);
522 	if (err)
523 		goto done;
524 
525 	switch (got_object_tag_get_object_type(tag)) {
526 	case GOT_OBJ_TYPE_COMMIT:
527 		err = load_commit(v, idset,
528 		    got_object_tag_get_object_id(tag), repo,
529 		    loose_obj_only, cancel_cb, cancel_arg);
530 		break;
531 	case GOT_OBJ_TYPE_TREE:
532 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
533 		    "", got_object_tag_get_tagger_time(tag),
534 		    loose_obj_only, repo, cancel_cb, cancel_arg);
535 		break;
536 	default:
537 		break;
538 	}
539 
540 done:
541 	got_object_tag_close(tag);
542 	return err;
543 }
544 
545 enum findtwixt_color {
546 	COLOR_KEEP = 0,
547 	COLOR_DROP,
548 	COLOR_BLANK,
549 };
550 static const int findtwixt_colors[] = {
551 	COLOR_KEEP,
552 	COLOR_DROP,
553 	COLOR_BLANK
554 };
555 
556 static const struct got_error *
557 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
558     int color, struct got_repository *repo)
559 {
560 	const struct got_error *err;
561 	struct got_object_qid *qid;
562 
563 	err = got_object_qid_alloc(&qid, id);
564 	if (err)
565 		return err;
566 
567 	STAILQ_INSERT_TAIL(ids, qid, entry);
568 	qid->data = (void *)&findtwixt_colors[color];
569 	return NULL;
570 }
571 
572 static const struct got_error *
573 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
574     struct got_object_id *id, struct got_repository *repo,
575     got_cancel_cb cancel_cb, void *cancel_arg)
576 {
577 	const struct got_error *err = NULL;
578 	struct got_commit_object *commit;
579 	const struct got_object_id_queue *parents;
580 	struct got_object_id_queue ids;
581 	struct got_object_qid *qid;
582 
583 	STAILQ_INIT(&ids);
584 
585 	err = got_object_qid_alloc(&qid, id);
586 	if (err)
587 		return err;
588 	STAILQ_INSERT_HEAD(&ids, qid, entry);
589 
590 	while (!STAILQ_EMPTY(&ids)) {
591 		if (cancel_cb) {
592 			err = (*cancel_cb)(cancel_arg);
593 			if (err)
594 				break;
595 		}
596 
597 		qid = STAILQ_FIRST(&ids);
598 		STAILQ_REMOVE_HEAD(&ids, entry);
599 
600 		if (got_object_idset_contains(drop, qid->id)) {
601 			got_object_qid_free(qid);
602 			continue;
603 		}
604 
605 		err = got_object_idset_add(drop, qid->id,
606 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
607 		if (err) {
608 			got_object_qid_free(qid);
609 			break;
610 		}
611 
612 		if (!got_object_idset_contains(keep, qid->id)) {
613 			got_object_qid_free(qid);
614 			continue;
615 		}
616 
617 		err = got_object_open_as_commit(&commit, repo, qid->id);
618 		got_object_qid_free(qid);
619 		if (err)
620 			break;
621 
622 		parents = got_object_commit_get_parent_ids(commit);
623 		if (parents) {
624 			err = got_object_id_queue_copy(parents, &ids);
625 			if (err) {
626 				got_object_commit_close(commit);
627 				break;
628 			}
629 		}
630 		got_object_commit_close(commit);
631 	}
632 
633 	got_object_id_queue_free(&ids);
634 	return err;
635 }
636 
637 struct append_id_arg {
638 	struct got_object_id **array;
639 	int idx;
640 };
641 
642 static const struct got_error *
643 append_id(struct got_object_id *id, void *data, void *arg)
644 {
645 	struct append_id_arg *a = arg;
646 
647 	a->array[a->idx] = got_object_id_dup(id);
648 	if (a->array[a->idx] == NULL)
649 		return got_error_from_errno("got_object_id_dup");
650 
651 	a->idx++;
652 	return NULL;
653 }
654 
655 static const struct got_error *
656 findtwixt(struct got_object_id ***res, int *nres,
657     struct got_object_id **head, int nhead,
658     struct got_object_id **tail, int ntail,
659     struct got_repository *repo,
660     got_cancel_cb cancel_cb, void *cancel_arg)
661 {
662 	const struct got_error *err = NULL;
663 	struct got_object_id_queue ids;
664 	struct got_object_idset *keep, *drop;
665 	struct got_object_qid *qid;
666 	int i, ncolor, nkeep, obj_type;
667 
668 	STAILQ_INIT(&ids);
669 	*res = NULL;
670 	*nres = 0;
671 
672 	keep = got_object_idset_alloc();
673 	if (keep == NULL)
674 		return got_error_from_errno("got_object_idset_alloc");
675 
676 	drop = got_object_idset_alloc();
677 	if (drop == NULL) {
678 		err = got_error_from_errno("got_object_idset_alloc");
679 		goto done;
680 	}
681 
682 	for (i = 0; i < nhead; i++) {
683 		struct got_object_id *id = head[i];
684 		if (id == NULL)
685 			continue;
686 		err = got_object_get_type(&obj_type, repo, id);
687 		if (err)
688 			return err;
689 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
690 			continue;
691 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
692 		if (err)
693 			goto done;
694 	}		
695 	for (i = 0; i < ntail; i++) {
696 		struct got_object_id *id = tail[i];
697 		if (id == NULL)
698 			continue;
699 		err = got_object_get_type(&obj_type, repo, id);
700 		if (err)
701 			return err;
702 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
703 			continue;
704 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
705 		if (err)
706 			goto done;
707 	}
708 
709 	while (!STAILQ_EMPTY(&ids)) {
710 		int qcolor;
711 		qid = STAILQ_FIRST(&ids);
712 		qcolor = *((int *)qid->data);
713 
714 		if (got_object_idset_contains(drop, qid->id))
715 			ncolor = COLOR_DROP;
716 		else if (got_object_idset_contains(keep, qid->id))
717 			ncolor = COLOR_KEEP;
718 		else
719 			ncolor = COLOR_BLANK;
720 
721 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
722 		    qcolor == COLOR_KEEP)) {
723 			STAILQ_REMOVE_HEAD(&ids, entry);
724 			got_object_qid_free(qid);
725 			continue;
726 		}
727 
728 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
729 			err = drop_commit(keep, drop, qid->id, repo,
730 			    cancel_cb, cancel_arg);
731 			if (err)
732 				goto done;
733 		} else if (ncolor == COLOR_BLANK) {
734 			struct got_commit_object *commit;
735 			struct got_object_id *id;
736 			const struct got_object_id_queue *parents;
737 			struct got_object_qid *pid;
738 
739 			id = got_object_id_dup(qid->id);
740 			if (id == NULL) {
741 				err = got_error_from_errno("got_object_id_dup");
742 				goto done;
743 			}
744 			if (qcolor == COLOR_KEEP)
745 				err = got_object_idset_add(keep, id,
746 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
747 			else
748 				err = got_object_idset_add(drop, id,
749 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
750 			if (err) {
751 				free(id);
752 				goto done;
753 			}
754 
755 			err = got_object_open_as_commit(&commit, repo, id);
756 			if (err) {
757 				free(id);
758 				goto done;
759 			}
760 			parents = got_object_commit_get_parent_ids(commit);
761 			if (parents) {
762 				STAILQ_FOREACH(pid, parents, entry) {
763 					err = queue_commit_id(&ids, pid->id,
764 					    qcolor, repo);
765 					if (err) {
766 						free(id);
767 						goto done;
768 					}
769 				}
770 			}
771 			got_object_commit_close(commit);
772 			commit = NULL;
773 		} else {
774 			/* should not happen */
775 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
776 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
777 			goto done;
778 		}
779 
780 		STAILQ_REMOVE_HEAD(&ids, entry);
781 		got_object_qid_free(qid);
782 	}
783 
784 	nkeep = got_object_idset_num_elements(keep);
785 	if (nkeep > 0) {
786 		struct append_id_arg arg;
787 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
788 		if (arg.array == NULL) {
789 			err = got_error_from_errno("calloc");
790 			goto done;
791 		}
792 		arg.idx = 0;
793 		err = got_object_idset_for_each(keep, append_id, &arg);
794 		if (err) {
795 			free(arg.array);
796 			goto done;
797 		}
798 		*res = arg.array;
799 		*nres = nkeep;
800 	}
801 done:
802 	got_object_idset_free(keep);
803 	got_object_idset_free(drop);
804 	got_object_id_queue_free(&ids);
805 	return err;
806 }
807 
808 static const struct got_error *
809 read_meta(struct got_pack_meta ***meta, int *nmeta,
810     struct got_object_id **theirs, int ntheirs,
811     struct got_object_id **ours, int nours, struct got_repository *repo,
812     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
813     got_cancel_cb cancel_cb, void *cancel_arg)
814 {
815 	const struct got_error *err = NULL;
816 	struct got_object_id **ids = NULL;
817 	struct got_object_idset *idset;
818 	int i, nobj = 0, obj_type;
819 	struct got_pack_metavec v;
820 
821 	*meta = NULL;
822 	*nmeta = 0;
823 
824 	idset = got_object_idset_alloc();
825 	if (idset == NULL)
826 		return got_error_from_errno("got_object_idset_alloc");
827 
828 	v.nmeta = 0;
829 	v.metasz = 64;
830 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
831 	if (v.meta == NULL) {
832 		err = got_error_from_errno("reallocarray");
833 		goto done;
834 	}
835 
836 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
837 	    cancel_cb, cancel_arg);
838 	if (err || nobj == 0)
839 		goto done;
840 
841 	for (i = 0; i < ntheirs; i++) {
842 		struct got_object_id *id = theirs[i];
843 		if (id == NULL)
844 			continue;
845 		err = got_object_get_type(&obj_type, repo, id);
846 		if (err)
847 			return err;
848 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
849 			continue;
850 		err = load_commit(NULL, idset, id, repo,
851 		    loose_obj_only, cancel_cb, cancel_arg);
852 		if (err)
853 			goto done;
854 		if (progress_cb) {
855 			err = progress_cb(progress_arg, 0L, nours,
856 			    v.nmeta, 0, 0);
857 			if (err)
858 				goto done;
859 		}
860 	}
861 
862 	for (i = 0; i < ntheirs; i++) {
863 		struct got_object_id *id = theirs[i];
864 		int *cached_type;
865 		if (id == NULL)
866 			continue;
867 		cached_type = got_object_idset_get(idset, id);
868 		if (cached_type == NULL) {
869 			err = got_object_get_type(&obj_type, repo, id);
870 			if (err)
871 				goto done;
872 		} else
873 			obj_type = *cached_type;
874 		if (obj_type != GOT_OBJ_TYPE_TAG)
875 			continue;
876 		err = load_tag(NULL, idset, id, repo,
877 		    loose_obj_only, cancel_cb, cancel_arg);
878 		if (err)
879 			goto done;
880 		if (progress_cb) {
881 			err = progress_cb(progress_arg, 0L, nours,
882 			    v.nmeta, 0, 0);
883 			if (err)
884 				goto done;
885 		}
886 	}
887 
888 	for (i = 0; i < nobj; i++) {
889 		err = load_commit(&v, idset, ids[i], repo,
890 		    loose_obj_only, cancel_cb, cancel_arg);
891 		if (err)
892 			goto done;
893 		if (progress_cb) {
894 			err = progress_cb(progress_arg, 0L, nours,
895 			    v.nmeta, 0, 0);
896 			if (err)
897 				goto done;
898 		}
899 	}
900 
901 	for (i = 0; i < nours; i++) {
902 		struct got_object_id *id = ours[i];
903 		int *cached_type;
904 		if (id == NULL)
905 			continue;
906 		cached_type = got_object_idset_get(idset, id);
907 		if (cached_type == NULL) {
908 			err = got_object_get_type(&obj_type, repo, id);
909 			if (err)
910 				goto done;
911 		} else
912 			obj_type = *cached_type;
913 		if (obj_type != GOT_OBJ_TYPE_TAG)
914 			continue;
915 		err = load_tag(&v, idset, id, repo,
916 		    loose_obj_only, cancel_cb, cancel_arg);
917 		if (err)
918 			goto done;
919 		if (progress_cb) {
920 			err = progress_cb(progress_arg, 0L, nours,
921 			    v.nmeta, 0, 0);
922 			if (err)
923 				goto done;
924 		}
925 	}
926 
927 done:
928 	for (i = 0; i < nobj; i++) {
929 		free(ids[i]);
930 	}
931 	free(ids);
932 	got_object_idset_free(idset);
933 	if (err == NULL) {
934 		*meta = v.meta;
935 		*nmeta = v.nmeta;
936 	} else
937 		free(v.meta);
938 
939 	return err;
940 }
941 
942 const struct got_error *
943 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
944 {
945 	size_t n;
946 
947 	SHA1Update(ctx, buf, len);
948 	n = fwrite(buf, 1, len, f);
949 	if (n != len)
950 		return got_ferror(f, GOT_ERR_IO);
951 	return NULL;
952 }
953 
954 static void
955 putbe32(char *b, uint32_t n)
956 {
957 	b[0] = n >> 24;
958 	b[1] = n >> 16;
959 	b[2] = n >> 8;
960 	b[3] = n >> 0;
961 }
962 
963 static int
964 write_order_cmp(const void *pa, const void *pb)
965 {
966 	struct got_pack_meta *a, *b, *ahd, *bhd;
967 
968 	a = *(struct got_pack_meta **)pa;
969 	b = *(struct got_pack_meta **)pb;
970 	ahd = (a->head == NULL) ? a : a->head;
971 	bhd = (b->head == NULL) ? b : b->head;
972 	if (ahd->mtime != bhd->mtime)
973 		return bhd->mtime - ahd->mtime;
974 	if (ahd != bhd)
975 		return (uintptr_t)bhd - (uintptr_t)ahd;
976 	if (a->nchain != b->nchain)
977 		return a->nchain - b->nchain;
978 	return a->mtime - b->mtime;
979 }
980 
981 static const struct got_error *
982 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
983 {
984 	size_t i;
985 
986 	*hdrlen = 0;
987 
988 	hdr[0] = obj_type << 4;
989 	hdr[0] |= len & 0xf;
990 	len >>= 4;
991 	for (i = 1; len != 0; i++){
992 		if (i >= bufsize)
993 			return got_error(GOT_ERR_NO_SPACE);
994 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
995 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
996 		len >>= GOT_DELTA_SIZE_SHIFT;
997 	}
998 
999 	*hdrlen = i;
1000 	return NULL;
1001 }
1002 
1003 static const struct got_error *
1004 append(char **p, int *len, int *sz, void *seg, int nseg)
1005 {
1006 	char *n;
1007 
1008 	if (*len + nseg >= *sz) {
1009 		while (*len + nseg >= *sz)
1010 			*sz += *sz / 2;
1011 		n = realloc(*p, *sz);
1012 		if (n == NULL)
1013 			return got_error_from_errno("realloc");
1014 		*p = n;
1015 	}
1016 	memcpy(*p + *len, seg, nseg);
1017 	*len += nseg;
1018 	return NULL;
1019 }
1020 
1021 
1022 static const struct got_error *
1023 encodedelta(int *nd, struct got_pack_meta *m, struct got_raw_object *o,
1024     off_t base_size, char **pp)
1025 {
1026 	const struct got_error *err = NULL;
1027 	char *p;
1028 	unsigned char buf[16], *bp;
1029 	int len, sz, i, j;
1030 	off_t n;
1031 	struct got_delta_instruction *d;
1032 
1033 	*pp = NULL;
1034 	*nd = 0;
1035 
1036 	sz = 128;
1037 	len = 0;
1038 	p = malloc(sz);
1039 	if (p == NULL)
1040 		return got_error_from_errno("malloc");
1041 
1042 	/* base object size */
1043 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1044 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1045 	for (i = 1; n > 0; i++) {
1046 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1047 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1048 		n >>= GOT_DELTA_SIZE_SHIFT;
1049 	}
1050 	err = append(&p, &len, &sz, buf, i);
1051 	if (err)
1052 		return err;
1053 
1054 	/* target object size */
1055 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1056 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1057 	for (i = 1; n > 0; i++) {
1058 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1059 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1060 		n >>= GOT_DELTA_SIZE_SHIFT;
1061 	}
1062 	err = append(&p, &len, &sz, buf, i);
1063 	if (err)
1064 		return err;
1065 	for (j = 0; j < m->ndeltas; j++) {
1066 		d = &m->deltas[j];
1067 		if (d->copy) {
1068 			n = d->offset;
1069 			bp = &buf[1];
1070 			buf[0] = GOT_DELTA_BASE_COPY;
1071 			for (i = 0; i < 4; i++) {
1072 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1073 				buf[0] |= 1 << i;
1074 				*bp++ = n & 0xff;
1075 				n >>= 8;
1076 				if (n == 0)
1077 					break;
1078 			}
1079 
1080 			n = d->len;
1081 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1082 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1083 				for (i = 0; i < 3 && n > 0; i++) {
1084 					buf[0] |= 1 << (i + 4);
1085 					*bp++ = n & 0xff;
1086 					n >>= 8;
1087 				}
1088 			}
1089 			err = append(&p, &len, &sz, buf, bp - buf);
1090 			if (err)
1091 				return err;
1092 		} else {
1093 			char content[128];
1094 			size_t r;
1095 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1096 				return got_error_from_errno("fseeko");
1097 			n = 0;
1098 			while (n != d->len) {
1099 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1100 				err = append(&p, &len, &sz, buf, 1);
1101 				if (err)
1102 					return err;
1103 				r = fread(content, 1, buf[0], o->f);
1104 				if (r != buf[0])
1105 					return got_ferror(o->f, GOT_ERR_IO);
1106 				err = append(&p, &len, &sz, content, buf[0]);
1107 				if (err)
1108 					return err;
1109 				n += buf[0];
1110 			}
1111 		}
1112 	}
1113 	*pp = p;
1114 	*nd = len;
1115 	return NULL;
1116 }
1117 
1118 static int
1119 packoff(char *hdr, off_t off)
1120 {
1121 	int i, j;
1122 	char rbuf[8];
1123 
1124 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1125 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1126 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1127 		    GOT_DELTA_SIZE_MORE;
1128 	}
1129 
1130 	j = 0;
1131 	while (i > 0)
1132 		hdr[j++] = rbuf[--i];
1133 	return j;
1134 }
1135 
1136 static const struct got_error *
1137 genpack(uint8_t *pack_sha1, FILE *packfile,
1138     struct got_pack_meta **meta, int nmeta, int nours,
1139     int use_offset_deltas, struct got_repository *repo,
1140     got_pack_progress_cb progress_cb, void *progress_arg,
1141     got_cancel_cb cancel_cb, void *cancel_arg)
1142 {
1143 	const struct got_error *err = NULL;
1144 	int i, nh, nd;
1145 	SHA1_CTX ctx;
1146 	struct got_pack_meta *m;
1147 	struct got_raw_object *raw;
1148 	char *p = NULL, buf[32];
1149 	size_t outlen, n;
1150 	struct got_deflate_checksum csum;
1151 	off_t packfile_size = 0;
1152 
1153 	SHA1Init(&ctx);
1154 	csum.output_sha1 = &ctx;
1155 	csum.output_crc = NULL;
1156 
1157 	err = hwrite(packfile, "PACK", 4, &ctx);
1158 	if (err)
1159 		return err;
1160 	putbe32(buf, GOT_PACKFILE_VERSION);
1161 	err = hwrite(packfile, buf, 4, &ctx);
1162 	if (err)
1163 		goto done;
1164 	putbe32(buf, nmeta);
1165 	err = hwrite(packfile, buf, 4, &ctx);
1166 	if (err)
1167 		goto done;
1168 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1169 	for (i = 0; i < nmeta; i++) {
1170 		if (progress_cb) {
1171 			err = progress_cb(progress_arg, packfile_size, nours,
1172 			    nmeta, nmeta, i);
1173 			if (err)
1174 				goto done;
1175 		}
1176 		m = meta[i];
1177 		m->off = ftello(packfile);
1178 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
1179 		if (err)
1180 			goto done;
1181 		if (m->deltas == NULL) {
1182 			err = packhdr(&nh, buf, sizeof(buf),
1183 			    m->obj_type, raw->size);
1184 			if (err)
1185 				goto done;
1186 			err = hwrite(packfile, buf, nh, &ctx);
1187 			if (err)
1188 				goto done;
1189 			packfile_size += nh;
1190 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1191 				err = got_error_from_errno("fseeko");
1192 				goto done;
1193 			}
1194 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1195 			    &csum);
1196 			if (err)
1197 				goto done;
1198 			packfile_size += outlen;
1199 		} else {
1200 			FILE *delta_file;
1201 			struct got_raw_object *base_raw;
1202 			err = got_object_raw_open(&base_raw, repo,
1203 			    &m->prev->id, 8192);
1204 			if (err)
1205 				goto done;
1206 			err = encodedelta(&nd, m, raw, base_raw->size, &p);
1207 			if (err)
1208 				goto done;
1209 			got_object_raw_close(base_raw);
1210 			if (use_offset_deltas && m->prev->off != 0) {
1211 				err = packhdr(&nh, buf, sizeof(buf),
1212 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1213 				if (err)
1214 					goto done;
1215 				nh += packoff(buf + nh,
1216 				    m->off - m->prev->off);
1217 				err = hwrite(packfile, buf, nh, &ctx);
1218 				if (err)
1219 					goto done;
1220 				packfile_size += nh;
1221 			} else {
1222 				err = packhdr(&nh, buf, sizeof(buf),
1223 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1224 				err = hwrite(packfile, buf, nh, &ctx);
1225 				if (err)
1226 					goto done;
1227 				packfile_size += nh;
1228 				err = hwrite(packfile, m->prev->id.sha1,
1229 				    sizeof(m->prev->id.sha1), &ctx);
1230 				packfile_size += sizeof(m->prev->id.sha1);
1231 				if (err)
1232 					goto done;
1233 			}
1234 			/* XXX need got_deflate_from_mem() */
1235 			delta_file = fmemopen(p, nd, "r");
1236 			if (delta_file == NULL) {
1237 				err = got_error_from_errno("fmemopen");
1238 				goto done;
1239 			}
1240 			err = got_deflate_to_file(&outlen, delta_file,
1241 			    packfile, &csum);
1242 			fclose(delta_file);
1243 			if (err)
1244 				goto done;
1245 			packfile_size += outlen;
1246 			free(p);
1247 			p = NULL;
1248 		}
1249 		got_object_raw_close(raw);
1250 		raw = NULL;
1251 	}
1252 	SHA1Final(pack_sha1, &ctx);
1253 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1254 	if (n != SHA1_DIGEST_LENGTH)
1255 		err = got_ferror(packfile, GOT_ERR_IO);
1256 	packfile_size += SHA1_DIGEST_LENGTH;
1257 	packfile_size += sizeof(struct got_packfile_hdr);
1258 	err = progress_cb(progress_arg, packfile_size, nours,
1259 	    nmeta, nmeta, nmeta);
1260 	if (err)
1261 		goto done;
1262 done:
1263 	free(p);
1264 	return err;
1265 }
1266 
1267 const struct got_error *
1268 got_pack_create(uint8_t *packsha1, FILE *packfile,
1269     struct got_object_id **theirs, int ntheirs,
1270     struct got_object_id **ours, int nours,
1271     struct got_repository *repo, int loose_obj_only, int allow_empty,
1272     got_pack_progress_cb progress_cb, void *progress_arg,
1273     got_cancel_cb cancel_cb, void *cancel_arg)
1274 {
1275 	const struct got_error *err;
1276 	struct got_pack_meta **meta;
1277 	int nmeta;
1278 
1279 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1280 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1281 	if (err)
1282 		return err;
1283 
1284 	if (nmeta == 0 && !allow_empty) {
1285 		err = got_error(GOT_ERR_CANNOT_PACK);
1286 		goto done;
1287 	}
1288 	if (nmeta > 0) {
1289 		err = pick_deltas(meta, nmeta, nours, repo,
1290 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1291 		if (err)
1292 			goto done;
1293 	}
1294 
1295 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1296 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1297 	if (err)
1298 		goto done;
1299 done:
1300 	free_nmeta(meta, nmeta);
1301 	return err;
1302 }