Gotweb

Blob

Date:: Fri Sep 24 22:02:58 2021 UTC
Message:: portable: add FreeBSD support This adds the capability to compile got-portable on FreeBSD.
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 <limits.h>
27 #include <zlib.h>
28 
29 #include "got_error.h"
30 #include "got_cancel.h"
31 #include "got_object.h"
32 #include "got_reference.h"
33 #include "got_repository_admin.h"
34 
35 #include "got_lib_deltify.h"
36 #include "got_lib_delta.h"
37 #include "got_lib_object.h"
38 #include "got_lib_object_idset.h"
39 #include "got_lib_object_cache.h"
40 #include "got_lib_deflate.h"
41 #include "got_lib_pack.h"
42 #include "got_lib_privsep.h"
43 #include "got_lib_repository.h"
44 
45 #ifndef MAX
46 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
47 #endif
48 
49 struct got_pack_meta {
50 	struct got_object_id id;
51 	char	*path;
52 	int	obj_type;
53 	time_t	mtime;
54 
55 	/* The best delta we picked */
56 	struct got_pack_meta *head;
57 	struct got_pack_meta *prev;
58 	struct got_delta_instruction *deltas;
59 	int	ndeltas;
60 	int	nchain;
61 
62 	/* Only used for delta window */
63 	struct got_delta_table *dtab;
64 
65 	/* Only used for writing offset deltas */
66 	off_t	off;
67 };
68 
69 struct got_pack_metavec {
70 	struct got_pack_meta **meta;
71 	int nmeta;
72 	int metasz;
73 };
74 
75 static const struct got_error *
76 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
77     const char *path, int obj_type, time_t mtime)
78 {
79 	const struct got_error *err = NULL;
80 	struct got_pack_meta *m;
81 
82 	*new = NULL;
83 
84 	m = calloc(1, sizeof(*m));
85 	if (m == NULL)
86 		return got_error_from_errno("malloc");
87 
88 	memcpy(&m->id, id, sizeof(m->id));
89 
90 	m->path = strdup(path);
91 	if (m->path == NULL) {
92 		err = got_error_from_errno("strdup");
93 		free(m);
94 		return err;
95 	}
96 
97 	m->obj_type = obj_type;
98 	m->mtime = mtime;
99 	*new = m;
100 	return NULL;
101 }
102 
103 static void
104 clear_meta(struct got_pack_meta *meta)
105 {
106 	if (meta == NULL)
107 		return;
108 	free(meta->deltas);
109 	meta->deltas = NULL;
110 	free(meta->path);
111 	meta->path = NULL;
112 }
113 
114 static void
115 free_nmeta(struct got_pack_meta **meta, int nmeta)
116 {
117 	int i;
118 
119 	for (i = 0; i < nmeta; i++)
120 		clear_meta(meta[i]);
121 	free(meta);
122 }
123 
124 static int
125 delta_order_cmp(const void *pa, const void *pb)
126 {
127 	struct got_pack_meta *a, *b;
128 	int cmp;
129 
130 	a = *(struct got_pack_meta **)pa;
131 	b = *(struct got_pack_meta **)pb;
132 
133 	if (a->obj_type != b->obj_type)
134 		return a->obj_type - b->obj_type;
135 	cmp = strcmp(a->path, b->path);
136 	if (cmp != 0)
137 		return cmp;
138 	if (a->mtime != b->mtime)
139 		return a->mtime - b->mtime;
140 	return got_object_id_cmp(&a->id, &b->id);
141 }
142 
143 static int
144 delta_size(struct got_delta_instruction *deltas, int ndeltas)
145 {
146 	int i, size = 32;
147 	for (i = 0; i < ndeltas; i++) {
148 		if (deltas[i].copy)
149 			size += GOT_DELTA_SIZE_SHIFT;
150 		else
151 			size += deltas[i].len + 1;
152 	}
153 	return size;
154 }
155 
156 
157 static const struct got_error *
158 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
159     struct got_repository *repo,
160     got_pack_progress_cb progress_cb, void *progress_arg,
161     got_cancel_cb cancel_cb, void *cancel_arg)
162 {
163 	const struct got_error *err = NULL;
164 	struct got_pack_meta *m = NULL, *base = NULL;
165 	struct got_raw_object *raw = NULL, *base_raw = NULL;
166 	struct got_delta_instruction *deltas;
167 	int i, j, size, ndeltas, best;
168 	const int max_base_candidates = 10;
169 
170 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
171 	for (i = 0; i < nmeta; i++) {
172 		if (cancel_cb) {
173 			err = (*cancel_cb)(cancel_arg);
174 			if (err)
175 				break;
176 		}
177 		if (progress_cb) {
178 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
179 			if (err)
180 				goto done;
181 		}
182 		m = meta[i];
183 		m->deltas = NULL;
184 		m->ndeltas = 0;
185 
186 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
187 		    m->obj_type == GOT_OBJ_TYPE_TAG)
188 			continue;
189 
190 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
191 		if (err)
192 			goto done;
193 
194 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
195 		    raw->size + raw->hdrlen);
196 		if (err)
197 			goto done;
198 
199 		if (i > max_base_candidates) {
200 			struct got_pack_meta *n = NULL;
201 			n = meta[i - (max_base_candidates + 1)];
202 			got_deltify_free(n->dtab);
203 			n->dtab = NULL;
204 		}
205 
206 		best = raw->size;
207 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
208 			if (cancel_cb) {
209 				err = (*cancel_cb)(cancel_arg);
210 				if (err)
211 					goto done;
212 			}
213 			base = meta[j];
214 			/* long chains make unpacking slow, avoid such bases */
215 			if (base->nchain >= 128 ||
216 			    base->obj_type != m->obj_type)
217 				continue;
218 
219 			err = got_object_raw_open(&base_raw, repo, &base->id,
220 			    8192);
221 			if (err)
222 				goto done;
223 			err = got_deltify(&deltas, &ndeltas,
224 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
225 			    base->dtab, base_raw->f, base_raw->hdrlen,
226 			    base_raw->size + base_raw->hdrlen);
227 			got_object_raw_close(base_raw);
228 			base_raw = NULL;
229 			if (err)
230 				goto done;
231 
232 			size = delta_size(deltas, ndeltas);
233 			if (size + 32 < best){
234 				/*
235 				 * if we already picked a best delta,
236 				 * replace it.
237 				 */
238 				free(m->deltas);
239 				best = size;
240 				m->deltas = deltas;
241 				m->ndeltas = ndeltas;
242 				m->nchain = base->nchain + 1;
243 				m->prev = base;
244 				m->head = base->head;
245 				if (m->head == NULL)
246 					m->head = base;
247 			} else {
248 				free(deltas);
249 				deltas = NULL;
250 				ndeltas = 0;
251 			}
252 		}
253 
254 		got_object_raw_close(raw);
255 		raw = NULL;
256 	}
257 done:
258 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
259 		got_deltify_free(meta[i]->dtab);
260 		meta[i]->dtab = NULL;
261 	}
262 	if (raw)
263 		got_object_raw_close(raw);
264 	if (base_raw)
265 		got_object_raw_close(base_raw);
266 	return err;
267 }
268 
269 static const struct got_error *
270 search_packidx(int *found, struct got_object_id *id,
271     struct got_repository *repo)
272 {
273 	const struct got_error *err = NULL;
274 	struct got_packidx *packidx = NULL;
275 	int idx;
276 
277 	*found = 0;
278 
279 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
280 	if (err == NULL)
281 		*found = 1; /* object is already packed */
282 	else if (err->code == GOT_ERR_NO_OBJ)
283 		err = NULL;
284 	return err;
285 }
286 
287 static const int obj_types[] = {
288 	GOT_OBJ_TYPE_ANY,
289 	GOT_OBJ_TYPE_COMMIT,
290 	GOT_OBJ_TYPE_TREE,
291 	GOT_OBJ_TYPE_BLOB,
292 	GOT_OBJ_TYPE_TAG,
293 	GOT_OBJ_TYPE_OFFSET_DELTA,
294 	GOT_OBJ_TYPE_REF_DELTA
295 };
296 
297 static const struct got_error *
298 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
299     struct got_object_id *id, const char *path, int obj_type,
300     time_t mtime, int loose_obj_only, struct got_repository *repo)
301 {
302 	const struct got_error *err;
303 	struct got_pack_meta *m;
304 
305 	if (loose_obj_only) {
306 		int is_packed;
307 		err = search_packidx(&is_packed, id, repo);
308 		if (err)
309 			return err;
310 		if (is_packed)
311 			return NULL;
312 	}
313 
314 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
315 	if (err)
316 		return err;
317 
318 	if (v == NULL)
319 		return NULL;
320 
321 	err = alloc_meta(&m, id, path, obj_type, mtime);
322 	if (err)
323 		goto done;
324 
325 	if (v->nmeta == v->metasz){
326 		size_t newsize = 2 * v->metasz;
327 		struct got_pack_meta **new;
328 		new = reallocarray(v->meta, newsize, sizeof(*new));
329 		if (new == NULL) {
330 			err = got_error_from_errno("reallocarray");
331 			goto done;
332 		}
333 		v->meta = new;
334 		v->metasz = newsize; 
335 	}
336 done:
337 	if (err) {
338 		clear_meta(m);
339 		free(m);
340 	} else
341 		v->meta[v->nmeta++] = m;
342 
343 	return err;
344 }
345 
346 static const struct got_error *
347 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
348     struct got_object_idset *idset, struct got_object_id *tree_id,
349     const char *dpath, time_t mtime, struct got_repository *repo,
350     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
351 {
352 	const struct got_error *err;
353 	struct got_tree_object *tree;
354 	char *p = NULL;
355 	int i;
356 
357 	err = got_object_open_as_tree(&tree, repo, tree_id);
358 	if (err)
359 		return err;
360 
361 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
362 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
363 		struct got_object_id *id = got_tree_entry_get_id(e);
364 		mode_t mode = got_tree_entry_get_mode(e);
365 
366 		if (cancel_cb) {
367 			err = (*cancel_cb)(cancel_arg);
368 			if (err)
369 				break;
370 		}
371 
372 		if (got_object_tree_entry_is_submodule(e) ||
373 		    got_object_idset_contains(idset, id))
374 			continue;
375 		
376 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
377 		    got_tree_entry_get_name(e)) == -1) {
378 			err = got_error_from_errno("asprintf");
379 			break;
380 		}
381 
382 		if (S_ISDIR(mode)) {
383 			struct got_object_qid *qid;
384 			err = got_object_qid_alloc(&qid, id);
385 			if (err)
386 				break;
387 			STAILQ_INSERT_TAIL(ids, qid, entry);
388 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
389 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
390 			    mtime, loose_obj_only, repo);
391 			if (err)
392 				break;
393 		}
394 		free(p);
395 		p = NULL;
396 	}
397 
398 	got_object_tree_close(tree);
399 	free(p);
400 	return err;
401 }
402 
403 static const struct got_error *
404 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
405     struct got_object_id *tree_id, const char *dpath, time_t mtime,
406     int loose_obj_only, struct got_repository *repo,
407     got_cancel_cb cancel_cb, void *cancel_arg)
408 {
409 	const struct got_error *err = NULL;
410 	struct got_object_id_queue tree_ids;
411 	struct got_object_qid *qid;
412 
413 	if (got_object_idset_contains(idset, tree_id))
414 		return NULL;
415 
416 	err = got_object_qid_alloc(&qid, tree_id);
417 	if (err)
418 		return err;
419 
420 	STAILQ_INIT(&tree_ids);
421 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
422 
423 	while (!STAILQ_EMPTY(&tree_ids)) {
424 		if (cancel_cb) {
425 			err = (*cancel_cb)(cancel_arg);
426 			if (err)
427 				break;
428 		}
429 
430 		qid = STAILQ_FIRST(&tree_ids);
431 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
432 
433 		if (got_object_idset_contains(idset, qid->id)) {
434 			got_object_qid_free(qid);
435 			continue;
436 		}
437 
438 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
439 		    mtime, loose_obj_only, repo);
440 		if (err) {
441 			got_object_qid_free(qid);
442 			break;
443 		}
444 
445 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
446 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
447 		got_object_qid_free(qid);
448 		if (err)
449 			break;
450 	}
451 
452 	got_object_id_queue_free(&tree_ids);
453 	return err;
454 }
455 
456 static const struct got_error *
457 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
458     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
459     got_cancel_cb cancel_cb, void *cancel_arg)
460 {
461 	const struct got_error *err;
462 	struct got_commit_object *commit;
463 
464 	if (got_object_idset_contains(idset, id))
465 		return NULL;
466 
467 	if (loose_obj_only) {
468 		int is_packed;
469 		err = search_packidx(&is_packed, id, repo);
470 		if (err)
471 			return err;
472 		if (is_packed)
473 			return NULL;
474 	}
475 
476 	err = got_object_open_as_commit(&commit, repo, id);
477 	if (err)
478 		return err;
479 
480 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
481 	    got_object_commit_get_committer_time(commit),
482 	    loose_obj_only, repo);
483 	if (err)
484 		goto done;
485 
486 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
487 	    "", got_object_commit_get_committer_time(commit),
488 	    loose_obj_only, repo, cancel_cb, cancel_arg);
489 done:
490 	got_object_commit_close(commit);
491 	return err;
492 }
493 
494 static const struct got_error *
495 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
496     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
497     got_cancel_cb cancel_cb, void *cancel_arg)
498 {
499 	const struct got_error *err;
500 	struct got_tag_object *tag = NULL;
501 
502 	if (got_object_idset_contains(idset, id))
503 		return NULL;
504 
505 	if (loose_obj_only) {
506 		int is_packed;
507 		err = search_packidx(&is_packed, id, repo);
508 		if (err)
509 			return err;
510 		if (is_packed)
511 			return NULL;
512 	}
513 
514 	err = got_object_open_as_tag(&tag, repo, id);
515 	if (err)
516 		return err;
517 
518 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
519 	    got_object_tag_get_tagger_time(tag),
520 	    loose_obj_only, repo);
521 	if (err)
522 		goto done;
523 
524 	switch (got_object_tag_get_object_type(tag)) {
525 	case GOT_OBJ_TYPE_COMMIT:
526 		err = load_commit(v, idset,
527 		    got_object_tag_get_object_id(tag), repo,
528 		    loose_obj_only, cancel_cb, cancel_arg);
529 		break;
530 	case GOT_OBJ_TYPE_TREE:
531 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
532 		    "", got_object_tag_get_tagger_time(tag),
533 		    loose_obj_only, repo, cancel_cb, cancel_arg);
534 		break;
535 	default:
536 		break;
537 	}
538 
539 done:
540 	got_object_tag_close(tag);
541 	return err;
542 }
543 
544 enum findtwixt_color {
545 	COLOR_KEEP = 0,
546 	COLOR_DROP,
547 	COLOR_BLANK,
548 };
549 static const int findtwixt_colors[] = {
550 	COLOR_KEEP,
551 	COLOR_DROP,
552 	COLOR_BLANK
553 };
554 
555 static const struct got_error *
556 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
557     int color, struct got_repository *repo)
558 {
559 	const struct got_error *err;
560 	struct got_object_qid *qid;
561 
562 	err = got_object_qid_alloc(&qid, id);
563 	if (err)
564 		return err;
565 
566 	STAILQ_INSERT_TAIL(ids, qid, entry);
567 	qid->data = (void *)&findtwixt_colors[color];
568 	return NULL;
569 }
570 
571 static const struct got_error *
572 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
573     struct got_object_id *id, struct got_repository *repo,
574     got_cancel_cb cancel_cb, void *cancel_arg)
575 {
576 	const struct got_error *err = NULL;
577 	struct got_commit_object *commit;
578 	const struct got_object_id_queue *parents;
579 	struct got_object_id_queue ids;
580 	struct got_object_qid *qid;
581 
582 	STAILQ_INIT(&ids);
583 
584 	err = got_object_qid_alloc(&qid, id);
585 	if (err)
586 		return err;
587 	STAILQ_INSERT_HEAD(&ids, qid, entry);
588 
589 	while (!STAILQ_EMPTY(&ids)) {
590 		if (cancel_cb) {
591 			err = (*cancel_cb)(cancel_arg);
592 			if (err)
593 				break;
594 		}
595 
596 		qid = STAILQ_FIRST(&ids);
597 		STAILQ_REMOVE_HEAD(&ids, entry);
598 
599 		if (got_object_idset_contains(drop, qid->id)) {
600 			got_object_qid_free(qid);
601 			continue;
602 		}
603 
604 		err = got_object_idset_add(drop, qid->id,
605 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
606 		if (err) {
607 			got_object_qid_free(qid);
608 			break;
609 		}
610 
611 		if (!got_object_idset_contains(keep, qid->id)) {
612 			got_object_qid_free(qid);
613 			continue;
614 		}
615 
616 		err = got_object_open_as_commit(&commit, repo, qid->id);
617 		got_object_qid_free(qid);
618 		if (err)
619 			break;
620 
621 		parents = got_object_commit_get_parent_ids(commit);
622 		if (parents) {
623 			err = got_object_id_queue_copy(parents, &ids);
624 			if (err) {
625 				got_object_commit_close(commit);
626 				break;
627 			}
628 		}
629 		got_object_commit_close(commit);
630 	}
631 
632 	got_object_id_queue_free(&ids);
633 	return err;
634 }
635 
636 struct append_id_arg {
637 	struct got_object_id **array;
638 	int idx;
639 };
640 
641 static const struct got_error *
642 append_id(struct got_object_id *id, void *data, void *arg)
643 {
644 	struct append_id_arg *a = arg;
645 
646 	a->array[a->idx] = got_object_id_dup(id);
647 	if (a->array[a->idx] == NULL)
648 		return got_error_from_errno("got_object_id_dup");
649 
650 	a->idx++;
651 	return NULL;
652 }
653 
654 static const struct got_error *
655 findtwixt(struct got_object_id ***res, int *nres,
656     struct got_object_id **head, int nhead,
657     struct got_object_id **tail, int ntail,
658     struct got_repository *repo,
659     got_cancel_cb cancel_cb, void *cancel_arg)
660 {
661 	const struct got_error *err = NULL;
662 	struct got_object_id_queue ids;
663 	struct got_object_idset *keep, *drop;
664 	struct got_object_qid *qid;
665 	int i, ncolor, nkeep, obj_type;
666 
667 	STAILQ_INIT(&ids);
668 	*res = NULL;
669 	*nres = 0;
670 
671 	keep = got_object_idset_alloc();
672 	if (keep == NULL)
673 		return got_error_from_errno("got_object_idset_alloc");
674 
675 	drop = got_object_idset_alloc();
676 	if (drop == NULL) {
677 		err = got_error_from_errno("got_object_idset_alloc");
678 		goto done;
679 	}
680 
681 	for (i = 0; i < nhead; i++) {
682 		struct got_object_id *id = head[i];
683 		if (id == NULL)
684 			continue;
685 		err = got_object_get_type(&obj_type, repo, id);
686 		if (err)
687 			return err;
688 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
689 			continue;
690 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
691 		if (err)
692 			goto done;
693 	}		
694 	for (i = 0; i < ntail; i++) {
695 		struct got_object_id *id = tail[i];
696 		if (id == NULL)
697 			continue;
698 		err = got_object_get_type(&obj_type, repo, id);
699 		if (err)
700 			return err;
701 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
702 			continue;
703 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
704 		if (err)
705 			goto done;
706 	}
707 
708 	while (!STAILQ_EMPTY(&ids)) {
709 		int qcolor;
710 		qid = STAILQ_FIRST(&ids);
711 		qcolor = *((int *)qid->data);
712 
713 		if (got_object_idset_contains(drop, qid->id))
714 			ncolor = COLOR_DROP;
715 		else if (got_object_idset_contains(keep, qid->id))
716 			ncolor = COLOR_KEEP;
717 		else
718 			ncolor = COLOR_BLANK;
719 
720 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
721 		    qcolor == COLOR_KEEP)) {
722 			STAILQ_REMOVE_HEAD(&ids, entry);
723 			got_object_qid_free(qid);
724 			continue;
725 		}
726 
727 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
728 			err = drop_commit(keep, drop, qid->id, repo,
729 			    cancel_cb, cancel_arg);
730 			if (err)
731 				goto done;
732 		} else if (ncolor == COLOR_BLANK) {
733 			struct got_commit_object *commit;
734 			struct got_object_id *id;
735 			const struct got_object_id_queue *parents;
736 			struct got_object_qid *pid;
737 
738 			id = got_object_id_dup(qid->id);
739 			if (id == NULL) {
740 				err = got_error_from_errno("got_object_id_dup");
741 				goto done;
742 			}
743 			if (qcolor == COLOR_KEEP)
744 				err = got_object_idset_add(keep, id,
745 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
746 			else
747 				err = got_object_idset_add(drop, id,
748 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
749 			if (err) {
750 				free(id);
751 				goto done;
752 			}
753 
754 			err = got_object_open_as_commit(&commit, repo, id);
755 			if (err) {
756 				free(id);
757 				goto done;
758 			}
759 			parents = got_object_commit_get_parent_ids(commit);
760 			if (parents) {
761 				STAILQ_FOREACH(pid, parents, entry) {
762 					err = queue_commit_id(&ids, pid->id,
763 					    qcolor, repo);
764 					if (err) {
765 						free(id);
766 						goto done;
767 					}
768 				}
769 			}
770 			got_object_commit_close(commit);
771 			commit = NULL;
772 		} else {
773 			/* should not happen */
774 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
775 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
776 			goto done;
777 		}
778 
779 		STAILQ_REMOVE_HEAD(&ids, entry);
780 		got_object_qid_free(qid);
781 	}
782 
783 	nkeep = got_object_idset_num_elements(keep);
784 	if (nkeep > 0) {
785 		struct append_id_arg arg;
786 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
787 		if (arg.array == NULL) {
788 			err = got_error_from_errno("calloc");
789 			goto done;
790 		}
791 		arg.idx = 0;
792 		err = got_object_idset_for_each(keep, append_id, &arg);
793 		if (err) {
794 			free(arg.array);
795 			goto done;
796 		}
797 		*res = arg.array;
798 		*nres = nkeep;
799 	}
800 done:
801 	got_object_idset_free(keep);
802 	got_object_idset_free(drop);
803 	got_object_id_queue_free(&ids);
804 	return err;
805 }
806 
807 static const struct got_error *
808 read_meta(struct got_pack_meta ***meta, int *nmeta,
809     struct got_object_id **theirs, int ntheirs,
810     struct got_object_id **ours, int nours, struct got_repository *repo,
811     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
812     got_cancel_cb cancel_cb, void *cancel_arg)
813 {
814 	const struct got_error *err = NULL;
815 	struct got_object_id **ids = NULL;
816 	struct got_object_idset *idset;
817 	int i, nobj = 0, obj_type;
818 	struct got_pack_metavec v;
819 
820 	*meta = NULL;
821 	*nmeta = 0;
822 
823 	idset = got_object_idset_alloc();
824 	if (idset == NULL)
825 		return got_error_from_errno("got_object_idset_alloc");
826 
827 	v.nmeta = 0;
828 	v.metasz = 64;
829 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
830 	if (v.meta == NULL) {
831 		err = got_error_from_errno("reallocarray");
832 		goto done;
833 	}
834 
835 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
836 	    cancel_cb, cancel_arg);
837 	if (err || nobj == 0)
838 		goto done;
839 
840 	for (i = 0; i < ntheirs; i++) {
841 		struct got_object_id *id = theirs[i];
842 		if (id == NULL)
843 			continue;
844 		err = got_object_get_type(&obj_type, repo, id);
845 		if (err)
846 			return err;
847 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
848 			continue;
849 		err = load_commit(NULL, idset, id, repo,
850 		    loose_obj_only, cancel_cb, cancel_arg);
851 		if (err)
852 			goto done;
853 		if (progress_cb) {
854 			err = progress_cb(progress_arg, 0L, nours,
855 			    v.nmeta, 0, 0);
856 			if (err)
857 				goto done;
858 		}
859 	}
860 
861 	for (i = 0; i < ntheirs; i++) {
862 		struct got_object_id *id = theirs[i];
863 		int *cached_type;
864 		if (id == NULL)
865 			continue;
866 		cached_type = got_object_idset_get(idset, id);
867 		if (cached_type == NULL) {
868 			err = got_object_get_type(&obj_type, repo, id);
869 			if (err)
870 				goto done;
871 		} else
872 			obj_type = *cached_type;
873 		if (obj_type != GOT_OBJ_TYPE_TAG)
874 			continue;
875 		err = load_tag(NULL, idset, id, repo,
876 		    loose_obj_only, cancel_cb, cancel_arg);
877 		if (err)
878 			goto done;
879 		if (progress_cb) {
880 			err = progress_cb(progress_arg, 0L, nours,
881 			    v.nmeta, 0, 0);
882 			if (err)
883 				goto done;
884 		}
885 	}
886 
887 	for (i = 0; i < nobj; i++) {
888 		err = load_commit(&v, idset, ids[i], repo,
889 		    loose_obj_only, cancel_cb, cancel_arg);
890 		if (err)
891 			goto done;
892 		if (progress_cb) {
893 			err = progress_cb(progress_arg, 0L, nours,
894 			    v.nmeta, 0, 0);
895 			if (err)
896 				goto done;
897 		}
898 	}
899 
900 	for (i = 0; i < nours; i++) {
901 		struct got_object_id *id = ours[i];
902 		int *cached_type;
903 		if (id == NULL)
904 			continue;
905 		cached_type = got_object_idset_get(idset, id);
906 		if (cached_type == NULL) {
907 			err = got_object_get_type(&obj_type, repo, id);
908 			if (err)
909 				goto done;
910 		} else
911 			obj_type = *cached_type;
912 		if (obj_type != GOT_OBJ_TYPE_TAG)
913 			continue;
914 		err = load_tag(&v, idset, id, repo,
915 		    loose_obj_only, cancel_cb, cancel_arg);
916 		if (err)
917 			goto done;
918 		if (progress_cb) {
919 			err = progress_cb(progress_arg, 0L, nours,
920 			    v.nmeta, 0, 0);
921 			if (err)
922 				goto done;
923 		}
924 	}
925 
926 done:
927 	for (i = 0; i < nobj; i++) {
928 		free(ids[i]);
929 	}
930 	free(ids);
931 	got_object_idset_free(idset);
932 	if (err == NULL) {
933 		*meta = v.meta;
934 		*nmeta = v.nmeta;
935 	} else
936 		free(v.meta);
937 
938 	return err;
939 }
940 
941 const struct got_error *
942 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
943 {
944 	size_t n;
945 
946 	SHA1Update(ctx, buf, len);
947 	n = fwrite(buf, 1, len, f);
948 	if (n != len)
949 		return got_ferror(f, GOT_ERR_IO);
950 	return NULL;
951 }
952 
953 static void
954 putbe32(char *b, uint32_t n)
955 {
956 	b[0] = n >> 24;
957 	b[1] = n >> 16;
958 	b[2] = n >> 8;
959 	b[3] = n >> 0;
960 }
961 
962 static int
963 write_order_cmp(const void *pa, const void *pb)
964 {
965 	struct got_pack_meta *a, *b, *ahd, *bhd;
966 
967 	a = *(struct got_pack_meta **)pa;
968 	b = *(struct got_pack_meta **)pb;
969 	ahd = (a->head == NULL) ? a : a->head;
970 	bhd = (b->head == NULL) ? b : b->head;
971 	if (ahd->mtime != bhd->mtime)
972 		return bhd->mtime - ahd->mtime;
973 	if (ahd != bhd)
974 		return (uintptr_t)bhd - (uintptr_t)ahd;
975 	if (a->nchain != b->nchain)
976 		return a->nchain - b->nchain;
977 	return a->mtime - b->mtime;
978 }
979 
980 static const struct got_error *
981 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
982 {
983 	size_t i;
984 
985 	*hdrlen = 0;
986 
987 	hdr[0] = obj_type << 4;
988 	hdr[0] |= len & 0xf;
989 	len >>= 4;
990 	for (i = 1; len != 0; i++){
991 		if (i >= bufsize)
992 			return got_error(GOT_ERR_NO_SPACE);
993 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
994 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
995 		len >>= GOT_DELTA_SIZE_SHIFT;
996 	}
997 
998 	*hdrlen = i;
999 	return NULL;
1000 }
1001 
1002 static const struct got_error *
1003 append(char **p, int *len, int *sz, void *seg, int nseg)
1004 {
1005 	char *n;
1006 
1007 	if (*len + nseg >= *sz) {
1008 		while (*len + nseg >= *sz)
1009 			*sz += *sz / 2;
1010 		n = realloc(*p, *sz);
1011 		if (n == NULL)
1012 			return got_error_from_errno("realloc");
1013 		*p = n;
1014 	}
1015 	memcpy(*p + *len, seg, nseg);
1016 	*len += nseg;
1017 	return NULL;
1018 }
1019 
1020 
1021 static const struct got_error *
1022 encodedelta(int *nd, struct got_pack_meta *m, struct got_raw_object *o,
1023     off_t base_size, char **pp)
1024 {
1025 	const struct got_error *err = NULL;
1026 	char *p;
1027 	unsigned char buf[16], *bp;
1028 	int len, sz, i, j;
1029 	off_t n;
1030 	struct got_delta_instruction *d;
1031 
1032 	*pp = NULL;
1033 	*nd = 0;
1034 
1035 	sz = 128;
1036 	len = 0;
1037 	p = malloc(sz);
1038 	if (p == NULL)
1039 		return got_error_from_errno("malloc");
1040 
1041 	/* base object size */
1042 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1043 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1044 	for (i = 1; n > 0; i++) {
1045 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1046 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1047 		n >>= GOT_DELTA_SIZE_SHIFT;
1048 	}
1049 	err = append(&p, &len, &sz, buf, i);
1050 	if (err)
1051 		return err;
1052 
1053 	/* target object size */
1054 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1055 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1056 	for (i = 1; n > 0; i++) {
1057 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1058 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1059 		n >>= GOT_DELTA_SIZE_SHIFT;
1060 	}
1061 	err = append(&p, &len, &sz, buf, i);
1062 	if (err)
1063 		return err;
1064 	for (j = 0; j < m->ndeltas; j++) {
1065 		d = &m->deltas[j];
1066 		if (d->copy) {
1067 			n = d->offset;
1068 			bp = &buf[1];
1069 			buf[0] = GOT_DELTA_BASE_COPY;
1070 			for (i = 0; i < 4; i++) {
1071 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1072 				buf[0] |= 1 << i;
1073 				*bp++ = n & 0xff;
1074 				n >>= 8;
1075 				if (n == 0)
1076 					break;
1077 			}
1078 
1079 			n = d->len;
1080 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1081 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1082 				for (i = 0; i < 3 && n > 0; i++) {
1083 					buf[0] |= 1 << (i + 4);
1084 					*bp++ = n & 0xff;
1085 					n >>= 8;
1086 				}
1087 			}
1088 			err = append(&p, &len, &sz, buf, bp - buf);
1089 			if (err)
1090 				return err;
1091 		} else {
1092 			char content[128];
1093 			size_t r;
1094 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1095 				return got_error_from_errno("fseeko");
1096 			n = 0;
1097 			while (n != d->len) {
1098 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1099 				err = append(&p, &len, &sz, buf, 1);
1100 				if (err)
1101 					return err;
1102 				r = fread(content, 1, buf[0], o->f);
1103 				if (r != buf[0])
1104 					return got_ferror(o->f, GOT_ERR_IO);
1105 				err = append(&p, &len, &sz, content, buf[0]);
1106 				if (err)
1107 					return err;
1108 				n += buf[0];
1109 			}
1110 		}
1111 	}
1112 	*pp = p;
1113 	*nd = len;
1114 	return NULL;
1115 }
1116 
1117 static int
1118 packoff(char *hdr, off_t off)
1119 {
1120 	int i, j;
1121 	char rbuf[8];
1122 
1123 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1124 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1125 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1126 		    GOT_DELTA_SIZE_MORE;
1127 	}
1128 
1129 	j = 0;
1130 	while (i > 0)
1131 		hdr[j++] = rbuf[--i];
1132 	return j;
1133 }
1134 
1135 static const struct got_error *
1136 genpack(uint8_t *pack_sha1, FILE *packfile,
1137     struct got_pack_meta **meta, int nmeta, int nours,
1138     int use_offset_deltas, struct got_repository *repo,
1139     got_pack_progress_cb progress_cb, void *progress_arg,
1140     got_cancel_cb cancel_cb, void *cancel_arg)
1141 {
1142 	const struct got_error *err = NULL;
1143 	int i, nh, nd;
1144 	SHA1_CTX ctx;
1145 	struct got_pack_meta *m;
1146 	struct got_raw_object *raw;
1147 	char *p = NULL, buf[32];
1148 	size_t outlen, n;
1149 	struct got_deflate_checksum csum;
1150 	off_t packfile_size = 0;
1151 
1152 	SHA1Init(&ctx);
1153 	csum.output_sha1 = &ctx;
1154 	csum.output_crc = NULL;
1155 
1156 	err = hwrite(packfile, "PACK", 4, &ctx);
1157 	if (err)
1158 		return err;
1159 	putbe32(buf, GOT_PACKFILE_VERSION);
1160 	err = hwrite(packfile, buf, 4, &ctx);
1161 	if (err)
1162 		goto done;
1163 	putbe32(buf, nmeta);
1164 	err = hwrite(packfile, buf, 4, &ctx);
1165 	if (err)
1166 		goto done;
1167 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1168 	for (i = 0; i < nmeta; i++) {
1169 		if (progress_cb) {
1170 			err = progress_cb(progress_arg, packfile_size, nours,
1171 			    nmeta, nmeta, i);
1172 			if (err)
1173 				goto done;
1174 		}
1175 		m = meta[i];
1176 		m->off = ftello(packfile);
1177 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
1178 		if (err)
1179 			goto done;
1180 		if (m->deltas == NULL) {
1181 			err = packhdr(&nh, buf, sizeof(buf),
1182 			    m->obj_type, raw->size);
1183 			if (err)
1184 				goto done;
1185 			err = hwrite(packfile, buf, nh, &ctx);
1186 			if (err)
1187 				goto done;
1188 			packfile_size += nh;
1189 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1190 				err = got_error_from_errno("fseeko");
1191 				goto done;
1192 			}
1193 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1194 			    &csum);
1195 			if (err)
1196 				goto done;
1197 			packfile_size += outlen;
1198 		} else {
1199 			FILE *delta_file;
1200 			struct got_raw_object *base_raw;
1201 			err = got_object_raw_open(&base_raw, repo,
1202 			    &m->prev->id, 8192);
1203 			if (err)
1204 				goto done;
1205 			err = encodedelta(&nd, m, raw, base_raw->size, &p);
1206 			if (err)
1207 				goto done;
1208 			got_object_raw_close(base_raw);
1209 			if (use_offset_deltas && m->prev->off != 0) {
1210 				err = packhdr(&nh, buf, sizeof(buf),
1211 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1212 				if (err)
1213 					goto done;
1214 				nh += packoff(buf + nh,
1215 				    m->off - m->prev->off);
1216 				err = hwrite(packfile, buf, nh, &ctx);
1217 				if (err)
1218 					goto done;
1219 				packfile_size += nh;
1220 			} else {
1221 				err = packhdr(&nh, buf, sizeof(buf),
1222 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1223 				err = hwrite(packfile, buf, nh, &ctx);
1224 				if (err)
1225 					goto done;
1226 				packfile_size += nh;
1227 				err = hwrite(packfile, m->prev->id.sha1,
1228 				    sizeof(m->prev->id.sha1), &ctx);
1229 				packfile_size += sizeof(m->prev->id.sha1);
1230 				if (err)
1231 					goto done;
1232 			}
1233 			/* XXX need got_deflate_from_mem() */
1234 			delta_file = fmemopen(p, nd, "r");
1235 			if (delta_file == NULL) {
1236 				err = got_error_from_errno("fmemopen");
1237 				goto done;
1238 			}
1239 			err = got_deflate_to_file(&outlen, delta_file,
1240 			    packfile, &csum);
1241 			fclose(delta_file);
1242 			if (err)
1243 				goto done;
1244 			packfile_size += outlen;
1245 			free(p);
1246 			p = NULL;
1247 		}
1248 		got_object_raw_close(raw);
1249 		raw = NULL;
1250 	}
1251 	SHA1Final(pack_sha1, &ctx);
1252 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1253 	if (n != SHA1_DIGEST_LENGTH)
1254 		err = got_ferror(packfile, GOT_ERR_IO);
1255 	packfile_size += SHA1_DIGEST_LENGTH;
1256 	packfile_size += sizeof(struct got_packfile_hdr);
1257 	err = progress_cb(progress_arg, packfile_size, nours,
1258 	    nmeta, nmeta, nmeta);
1259 	if (err)
1260 		goto done;
1261 done:
1262 	free(p);
1263 	return err;
1264 }
1265 
1266 const struct got_error *
1267 got_pack_create(uint8_t *packsha1, FILE *packfile,
1268     struct got_object_id **theirs, int ntheirs,
1269     struct got_object_id **ours, int nours,
1270     struct got_repository *repo, int loose_obj_only, int allow_empty,
1271     got_pack_progress_cb progress_cb, void *progress_arg,
1272     got_cancel_cb cancel_cb, void *cancel_arg)
1273 {
1274 	const struct got_error *err;
1275 	struct got_pack_meta **meta;
1276 	int nmeta;
1277 
1278 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1279 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1280 	if (err)
1281 		return err;
1282 
1283 	if (nmeta == 0 && !allow_empty) {
1284 		err = got_error(GOT_ERR_CANNOT_PACK);
1285 		goto done;
1286 	}
1287 	if (nmeta > 0) {
1288 		err = pick_deltas(meta, nmeta, nours, repo,
1289 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1290 		if (err)
1291 			goto done;
1292 	}
1293 
1294 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1295 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1296 	if (err)
1297 		goto done;
1298 done:
1299 	free_nmeta(meta, nmeta);
1300 	return err;
1301 }