op public repos

Blob

Date:: Fri Sep 3 09:51:31 2021 UTC
Message:: fix 'got send' adding too many objects to the pack file in some cases Load server-side tags before loading local commits. Otherwise objects which are reachable via server-side tags will not be filtered out.
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/queue.h>
20 #include <sys/uio.h>
21 #include <sys/stat.h>
22 
23 #include <stdint.h>
24 #include <imsg.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <sha1.h>
29 #include <limits.h>
30 #include <zlib.h>
31 
32 #include "got_error.h"
33 #include "got_cancel.h"
34 #include "got_object.h"
35 #include "got_reference.h"
36 #include "got_repository_admin.h"
37 
38 #include "got_lib_deltify.h"
39 #include "got_lib_delta.h"
40 #include "got_lib_object.h"
41 #include "got_lib_object_idset.h"
42 #include "got_lib_object_cache.h"
43 #include "got_lib_deflate.h"
44 #include "got_lib_pack.h"
45 #include "got_lib_privsep.h"
46 #include "got_lib_repository.h"
47 
48 #ifndef MAX
49 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
50 #endif
51 
52 struct got_pack_meta {
53 	struct got_object_id id;
54 	char	*path;
55 	int	obj_type;
56 	time_t	mtime;
57 
58 	/* The best delta we picked */
59 	struct got_pack_meta *head;
60 	struct got_pack_meta *prev;
61 	struct got_delta_instruction *deltas;
62 	int	ndeltas;
63 	int	nchain;
64 
65 	/* Only used for delta window */
66 	struct got_delta_table *dtab;
67 
68 	/* Only used for writing offset deltas */
69 	off_t	off;
70 };
71 
72 struct got_pack_metavec {
73 	struct got_pack_meta **meta;
74 	int nmeta;
75 	int metasz;
76 };
77 
78 static const struct got_error *
79 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
80     const char *path, int obj_type, time_t mtime)
81 {
82 	const struct got_error *err = NULL;
83 	struct got_pack_meta *m;
84 
85 	*new = NULL;
86 
87 	m = calloc(1, sizeof(*m));
88 	if (m == NULL)
89 		return got_error_from_errno("malloc");
90 
91 	memcpy(&m->id, id, sizeof(m->id));
92 
93 	m->path = strdup(path);
94 	if (m->path == NULL) {
95 		err = got_error_from_errno("strdup");
96 		free(m);
97 		return err;
98 	}
99 
100 	m->obj_type = obj_type;
101 	m->mtime = mtime;
102 	*new = m;
103 	return NULL;
104 }
105 
106 static void
107 clear_meta(struct got_pack_meta *meta)
108 {
109 	if (meta == NULL)
110 		return;
111 	free(meta->deltas);
112 	meta->deltas = NULL;
113 	free(meta->path);
114 	meta->path = NULL;
115 }
116 
117 static void
118 free_nmeta(struct got_pack_meta **meta, int nmeta)
119 {
120 	int i;
121 
122 	for (i = 0; i < nmeta; i++)
123 		clear_meta(meta[i]);
124 	free(meta);
125 }
126 
127 static int
128 delta_order_cmp(const void *pa, const void *pb)
129 {
130 	struct got_pack_meta *a, *b;
131 	int cmp;
132 
133 	a = *(struct got_pack_meta **)pa;
134 	b = *(struct got_pack_meta **)pb;
135 
136 	if (a->obj_type != b->obj_type)
137 		return a->obj_type - b->obj_type;
138 	cmp = strcmp(a->path, b->path);
139 	if (cmp != 0)
140 		return cmp;
141 	if (a->mtime != b->mtime)
142 		return a->mtime - b->mtime;
143 	return got_object_id_cmp(&a->id, &b->id);
144 }
145 
146 static int
147 delta_size(struct got_delta_instruction *deltas, int ndeltas)
148 {
149 	int i, size = 32;
150 	for (i = 0; i < ndeltas; i++) {
151 		if (deltas[i].copy)
152 			size += GOT_DELTA_SIZE_SHIFT;
153 		else
154 			size += deltas[i].len + 1;
155 	}
156 	return size;
157 }
158 
159 
160 static const struct got_error *
161 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
162     struct got_repository *repo,
163     got_pack_progress_cb progress_cb, void *progress_arg,
164     got_cancel_cb cancel_cb, void *cancel_arg)
165 {
166 	const struct got_error *err = NULL;
167 	struct got_pack_meta *m = NULL, *base = NULL;
168 	struct got_raw_object *raw = NULL, *base_raw = NULL;
169 	struct got_delta_instruction *deltas;
170 	int i, j, size, ndeltas, best;
171 	const int max_base_candidates = 10;
172 
173 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
174 	for (i = 0; i < nmeta; i++) {
175 		if (cancel_cb) {
176 			err = (*cancel_cb)(cancel_arg);
177 			if (err)
178 				break;
179 		}
180 		if (progress_cb) {
181 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
182 			if (err)
183 				goto done;
184 		}
185 		m = meta[i];
186 		m->deltas = NULL;
187 		m->ndeltas = 0;
188 
189 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
190 		    m->obj_type == GOT_OBJ_TYPE_TAG)
191 			continue;
192 
193 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
194 		if (err)
195 			goto done;
196 
197 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
198 		    raw->size + raw->hdrlen);
199 		if (err)
200 			goto done;
201 
202 		if (i > max_base_candidates) {
203 			struct got_pack_meta *n = NULL;
204 			n = meta[i - (max_base_candidates + 1)];
205 			got_deltify_free(n->dtab);
206 			n->dtab = NULL;
207 		}
208 
209 		best = raw->size;
210 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
211 			if (cancel_cb) {
212 				err = (*cancel_cb)(cancel_arg);
213 				if (err)
214 					goto done;
215 			}
216 			base = meta[j];
217 			/* long chains make unpacking slow, avoid such bases */
218 			if (base->nchain >= 128 ||
219 			    base->obj_type != m->obj_type)
220 				continue;
221 
222 			err = got_object_raw_open(&base_raw, repo, &base->id,
223 			    8192);
224 			if (err)
225 				goto done;
226 			err = got_deltify(&deltas, &ndeltas,
227 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
228 			    base->dtab, base_raw->f, base_raw->hdrlen,
229 			    base_raw->size + base_raw->hdrlen);
230 			got_object_raw_close(base_raw);
231 			base_raw = NULL;
232 			if (err)
233 				goto done;
234 
235 			size = delta_size(deltas, ndeltas);
236 			if (size + 32 < best){
237 				/*
238 				 * if we already picked a best delta,
239 				 * replace it.
240 				 */
241 				free(m->deltas);
242 				best = size;
243 				m->deltas = deltas;
244 				m->ndeltas = ndeltas;
245 				m->nchain = base->nchain + 1;
246 				m->prev = base;
247 				m->head = base->head;
248 				if (m->head == NULL)
249 					m->head = base;
250 			} else {
251 				free(deltas);
252 				deltas = NULL;
253 				ndeltas = 0;
254 			}
255 		}
256 
257 		got_object_raw_close(raw);
258 		raw = NULL;
259 	}
260 done:
261 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
262 		got_deltify_free(meta[i]->dtab);
263 		meta[i]->dtab = NULL;
264 	}
265 	if (raw)
266 		got_object_raw_close(raw);
267 	if (base_raw)
268 		got_object_raw_close(base_raw);
269 	return err;
270 }
271 
272 static const struct got_error *
273 search_packidx(int *found, struct got_object_id *id,
274     struct got_repository *repo)
275 {
276 	const struct got_error *err = NULL;
277 	struct got_packidx *packidx = NULL;
278 	int idx;
279 
280 	*found = 0;
281 
282 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
283 	if (err == NULL)
284 		*found = 1; /* object is already packed */
285 	else if (err->code == GOT_ERR_NO_OBJ)
286 		err = NULL;
287 	return err;
288 }
289 
290 static const int obj_types[] = {
291 	GOT_OBJ_TYPE_ANY,
292 	GOT_OBJ_TYPE_COMMIT,
293 	GOT_OBJ_TYPE_TREE,
294 	GOT_OBJ_TYPE_BLOB,
295 	GOT_OBJ_TYPE_TAG,
296 	GOT_OBJ_TYPE_OFFSET_DELTA,
297 	GOT_OBJ_TYPE_REF_DELTA
298 };
299 
300 static const struct got_error *
301 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
302     struct got_object_id *id, const char *path, int obj_type,
303     time_t mtime, int loose_obj_only, struct got_repository *repo)
304 {
305 	const struct got_error *err;
306 	struct got_pack_meta *m;
307 
308 	if (loose_obj_only) {
309 		int is_packed;
310 		err = search_packidx(&is_packed, id, repo);
311 		if (err)
312 			return err;
313 		if (is_packed)
314 			return NULL;
315 	}
316 
317 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
318 	if (err)
319 		return err;
320 
321 	if (v == NULL)
322 		return NULL;
323 
324 	err = alloc_meta(&m, id, path, obj_type, mtime);
325 	if (err)
326 		goto done;
327 
328 	if (v->nmeta == v->metasz){
329 		size_t newsize = 2 * v->metasz;
330 		struct got_pack_meta **new;
331 		new = reallocarray(v->meta, newsize, sizeof(*new));
332 		if (new == NULL) {
333 			err = got_error_from_errno("reallocarray");
334 			goto done;
335 		}
336 		v->meta = new;
337 		v->metasz = newsize; 
338 	}
339 done:
340 	if (err) {
341 		clear_meta(m);
342 		free(m);
343 	} else
344 		v->meta[v->nmeta++] = m;
345 
346 	return err;
347 }
348 
349 static const struct got_error *
350 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
351     struct got_object_idset *idset, struct got_object_id *tree_id,
352     const char *dpath, time_t mtime, struct got_repository *repo,
353     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
354 {
355 	const struct got_error *err;
356 	struct got_tree_object *tree;
357 	char *p = NULL;
358 	int i;
359 
360 	err = got_object_open_as_tree(&tree, repo, tree_id);
361 	if (err)
362 		return err;
363 
364 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
365 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
366 		struct got_object_id *id = got_tree_entry_get_id(e);
367 		mode_t mode = got_tree_entry_get_mode(e);
368 
369 		if (cancel_cb) {
370 			err = (*cancel_cb)(cancel_arg);
371 			if (err)
372 				break;
373 		}
374 
375 		if (got_object_tree_entry_is_symlink(e) ||
376 		    got_object_tree_entry_is_submodule(e) ||
377 		    got_object_idset_contains(idset, id))
378 			continue;
379 		
380 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
381 		    got_tree_entry_get_name(e)) == -1) {
382 			err = got_error_from_errno("asprintf");
383 			break;
384 		}
385 
386 		if (S_ISDIR(mode)) {
387 			struct got_object_qid *qid;
388 			err = got_object_qid_alloc(&qid, id);
389 			if (err)
390 				break;
391 			STAILQ_INSERT_TAIL(ids, qid, entry);
392 		} else if (S_ISREG(mode)) {
393 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
394 			    mtime, loose_obj_only, repo);
395 			if (err)
396 				break;
397 		}
398 		free(p);
399 		p = NULL;
400 	}
401 
402 	got_object_tree_close(tree);
403 	free(p);
404 	return err;
405 }
406 
407 static const struct got_error *
408 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
409     struct got_object_id *tree_id, const char *dpath, time_t mtime,
410     int loose_obj_only, struct got_repository *repo,
411     got_cancel_cb cancel_cb, void *cancel_arg)
412 {
413 	const struct got_error *err = NULL;
414 	struct got_object_id_queue tree_ids;
415 	struct got_object_qid *qid;
416 
417 	if (got_object_idset_contains(idset, tree_id))
418 		return NULL;
419 
420 	err = got_object_qid_alloc(&qid, tree_id);
421 	if (err)
422 		return err;
423 
424 	STAILQ_INIT(&tree_ids);
425 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
426 
427 	while (!STAILQ_EMPTY(&tree_ids)) {
428 		if (cancel_cb) {
429 			err = (*cancel_cb)(cancel_arg);
430 			if (err)
431 				break;
432 		}
433 
434 		qid = STAILQ_FIRST(&tree_ids);
435 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
436 
437 		if (got_object_idset_contains(idset, qid->id)) {
438 			got_object_qid_free(qid);
439 			continue;
440 		}
441 
442 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
443 		    mtime, loose_obj_only, repo);
444 		if (err) {
445 			got_object_qid_free(qid);
446 			break;
447 		}
448 
449 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
450 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
451 		got_object_qid_free(qid);
452 		if (err)
453 			break;
454 	}
455 
456 	got_object_id_queue_free(&tree_ids);
457 	return err;
458 }
459 
460 static const struct got_error *
461 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
462     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
463     got_cancel_cb cancel_cb, void *cancel_arg)
464 {
465 	const struct got_error *err;
466 	struct got_commit_object *commit;
467 
468 	if (got_object_idset_contains(idset, id))
469 		return NULL;
470 
471 	if (loose_obj_only) {
472 		int is_packed;
473 		err = search_packidx(&is_packed, id, repo);
474 		if (err)
475 			return err;
476 		if (is_packed)
477 			return NULL;
478 	}
479 
480 	err = got_object_open_as_commit(&commit, repo, id);
481 	if (err)
482 		return err;
483 
484 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
485 	    got_object_commit_get_committer_time(commit),
486 	    loose_obj_only, repo);
487 	if (err)
488 		goto done;
489 
490 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
491 	    "", got_object_commit_get_committer_time(commit),
492 	    loose_obj_only, repo, cancel_cb, cancel_arg);
493 done:
494 	got_object_commit_close(commit);
495 	return err;
496 }
497 
498 static const struct got_error *
499 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
500     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
501     got_cancel_cb cancel_cb, void *cancel_arg)
502 {
503 	const struct got_error *err;
504 	struct got_tag_object *tag = NULL;
505 
506 	if (got_object_idset_contains(idset, id))
507 		return NULL;
508 
509 	if (loose_obj_only) {
510 		int is_packed;
511 		err = search_packidx(&is_packed, id, repo);
512 		if (err)
513 			return err;
514 		if (is_packed)
515 			return NULL;
516 	}
517 
518 	err = got_object_open_as_tag(&tag, repo, id);
519 	if (err)
520 		return err;
521 
522 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
523 	    got_object_tag_get_tagger_time(tag),
524 	    loose_obj_only, repo);
525 	if (err)
526 		goto done;
527 
528 	switch (got_object_tag_get_object_type(tag)) {
529 	case GOT_OBJ_TYPE_COMMIT:
530 		err = load_commit(NULL, idset,
531 		    got_object_tag_get_object_id(tag), repo,
532 		    loose_obj_only, cancel_cb, cancel_arg);
533 		break;
534 	case GOT_OBJ_TYPE_TREE:
535 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
536 		    "", got_object_tag_get_tagger_time(tag),
537 		    loose_obj_only, repo, cancel_cb, cancel_arg);
538 		break;
539 	default:
540 		break;
541 	}
542 
543 done:
544 	got_object_tag_close(tag);
545 	return err;
546 }
547 
548 enum findtwixt_color {
549 	COLOR_KEEP = 0,
550 	COLOR_DROP,
551 	COLOR_BLANK,
552 };
553 static const int findtwixt_colors[] = {
554 	COLOR_KEEP,
555 	COLOR_DROP,
556 	COLOR_BLANK
557 };
558 
559 static const struct got_error *
560 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
561     int color, struct got_repository *repo)
562 {
563 	const struct got_error *err;
564 	struct got_object_qid *qid;
565 
566 	err = got_object_qid_alloc(&qid, id);
567 	if (err)
568 		return err;
569 
570 	STAILQ_INSERT_TAIL(ids, qid, entry);
571 	qid->data = (void *)&findtwixt_colors[color];
572 	return NULL;
573 }
574 
575 static const struct got_error *
576 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
577     struct got_object_id *id, struct got_repository *repo,
578     got_cancel_cb cancel_cb, void *cancel_arg)
579 {
580 	const struct got_error *err = NULL;
581 	struct got_commit_object *commit;
582 	const struct got_object_id_queue *parents;
583 	struct got_object_id_queue ids;
584 	struct got_object_qid *qid;
585 
586 	STAILQ_INIT(&ids);
587 
588 	err = got_object_qid_alloc(&qid, id);
589 	if (err)
590 		return err;
591 	STAILQ_INSERT_HEAD(&ids, qid, entry);
592 
593 	while (!STAILQ_EMPTY(&ids)) {
594 		if (cancel_cb) {
595 			err = (*cancel_cb)(cancel_arg);
596 			if (err)
597 				break;
598 		}
599 
600 		qid = STAILQ_FIRST(&ids);
601 		STAILQ_REMOVE_HEAD(&ids, entry);
602 
603 		if (got_object_idset_contains(drop, qid->id)) {
604 			got_object_qid_free(qid);
605 			continue;
606 		}
607 
608 		err = got_object_idset_add(drop, qid->id,
609 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
610 		if (err) {
611 			got_object_qid_free(qid);
612 			break;
613 		}
614 
615 		if (!got_object_idset_contains(keep, qid->id)) {
616 			got_object_qid_free(qid);
617 			continue;
618 		}
619 
620 		err = got_object_open_as_commit(&commit, repo, qid->id);
621 		got_object_qid_free(qid);
622 		if (err)
623 			break;
624 
625 		parents = got_object_commit_get_parent_ids(commit);
626 		if (parents) {
627 			err = got_object_id_queue_copy(parents, &ids);
628 			if (err) {
629 				got_object_commit_close(commit);
630 				break;
631 			}
632 		}
633 		got_object_commit_close(commit);
634 	}
635 
636 	got_object_id_queue_free(&ids);
637 	return err;
638 }
639 
640 struct append_id_arg {
641 	struct got_object_id **array;
642 	int idx;
643 };
644 
645 static const struct got_error *
646 append_id(struct got_object_id *id, void *data, void *arg)
647 {
648 	struct append_id_arg *a = arg;
649 
650 	a->array[a->idx] = got_object_id_dup(id);
651 	if (a->array[a->idx] == NULL)
652 		return got_error_from_errno("got_object_id_dup");
653 
654 	a->idx++;
655 	return NULL;
656 }
657 
658 static const struct got_error *
659 findtwixt(struct got_object_id ***res, int *nres,
660     struct got_object_id **head, int nhead,
661     struct got_object_id **tail, int ntail,
662     struct got_repository *repo,
663     got_cancel_cb cancel_cb, void *cancel_arg)
664 {
665 	const struct got_error *err = NULL;
666 	struct got_object_id_queue ids;
667 	struct got_object_idset *keep, *drop;
668 	struct got_object_qid *qid;
669 	int i, ncolor, nkeep, obj_type;
670 
671 	STAILQ_INIT(&ids);
672 	*res = NULL;
673 	*nres = 0;
674 
675 	keep = got_object_idset_alloc();
676 	if (keep == NULL)
677 		return got_error_from_errno("got_object_idset_alloc");
678 
679 	drop = got_object_idset_alloc();
680 	if (drop == NULL) {
681 		err = got_error_from_errno("got_object_idset_alloc");
682 		goto done;
683 	}
684 
685 	for (i = 0; i < nhead; i++) {
686 		struct got_object_id *id = head[i];
687 		if (id == NULL)
688 			continue;
689 		err = got_object_get_type(&obj_type, repo, id);
690 		if (err)
691 			return err;
692 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
693 			continue;
694 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
695 		if (err)
696 			goto done;
697 	}		
698 	for (i = 0; i < ntail; i++) {
699 		struct got_object_id *id = tail[i];
700 		if (id == NULL)
701 			continue;
702 		err = got_object_get_type(&obj_type, repo, id);
703 		if (err)
704 			return err;
705 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
706 			continue;
707 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
708 		if (err)
709 			goto done;
710 	}
711 
712 	while (!STAILQ_EMPTY(&ids)) {
713 		int qcolor;
714 		qid = STAILQ_FIRST(&ids);
715 		qcolor = *((int *)qid->data);
716 
717 		if (got_object_idset_contains(drop, qid->id))
718 			ncolor = COLOR_DROP;
719 		else if (got_object_idset_contains(keep, qid->id))
720 			ncolor = COLOR_KEEP;
721 		else
722 			ncolor = COLOR_BLANK;
723 
724 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
725 		    qcolor == COLOR_KEEP)) {
726 			STAILQ_REMOVE_HEAD(&ids, entry);
727 			got_object_qid_free(qid);
728 			continue;
729 		}
730 
731 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
732 			err = drop_commit(keep, drop, qid->id, repo,
733 			    cancel_cb, cancel_arg);
734 			if (err)
735 				goto done;
736 		} else if (ncolor == COLOR_BLANK) {
737 			struct got_commit_object *commit;
738 			struct got_object_id *id;
739 			const struct got_object_id_queue *parents;
740 			struct got_object_qid *pid;
741 
742 			id = got_object_id_dup(qid->id);
743 			if (id == NULL) {
744 				err = got_error_from_errno("got_object_id_dup");
745 				goto done;
746 			}
747 			if (qcolor == COLOR_KEEP)
748 				err = got_object_idset_add(keep, id,
749 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
750 			else
751 				err = got_object_idset_add(drop, id,
752 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
753 			if (err) {
754 				free(id);
755 				goto done;
756 			}
757 
758 			err = got_object_open_as_commit(&commit, repo, id);
759 			if (err) {
760 				free(id);
761 				goto done;
762 			}
763 			parents = got_object_commit_get_parent_ids(commit);
764 			if (parents) {
765 				STAILQ_FOREACH(pid, parents, entry) {
766 					err = queue_commit_id(&ids, pid->id,
767 					    qcolor, repo);
768 					if (err) {
769 						free(id);
770 						goto done;
771 					}
772 				}
773 			}
774 			got_object_commit_close(commit);
775 			commit = NULL;
776 		} else {
777 			/* should not happen */
778 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
779 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
780 			goto done;
781 		}
782 
783 		STAILQ_REMOVE_HEAD(&ids, entry);
784 		got_object_qid_free(qid);
785 	}
786 
787 	nkeep = got_object_idset_num_elements(keep);
788 	if (nkeep > 0) {
789 		struct append_id_arg arg;
790 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
791 		if (arg.array == NULL) {
792 			err = got_error_from_errno("calloc");
793 			goto done;
794 		}
795 		arg.idx = 0;
796 		err = got_object_idset_for_each(keep, append_id, &arg);
797 		if (err) {
798 			free(arg.array);
799 			goto done;
800 		}
801 		*res = arg.array;
802 		*nres = nkeep;
803 	}
804 done:
805 	got_object_idset_free(keep);
806 	got_object_idset_free(drop);
807 	got_object_id_queue_free(&ids);
808 	return err;
809 }
810 
811 static const struct got_error *
812 read_meta(struct got_pack_meta ***meta, int *nmeta,
813     struct got_object_id **theirs, int ntheirs,
814     struct got_object_id **ours, int nours, struct got_repository *repo,
815     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
816     got_cancel_cb cancel_cb, void *cancel_arg)
817 {
818 	const struct got_error *err = NULL;
819 	struct got_object_id **ids = NULL;
820 	struct got_object_idset *idset;
821 	int i, nobj = 0, obj_type;
822 	struct got_pack_metavec v;
823 
824 	*meta = NULL;
825 	*nmeta = 0;
826 
827 	idset = got_object_idset_alloc();
828 	if (idset == NULL)
829 		return got_error_from_errno("got_object_idset_alloc");
830 
831 	v.nmeta = 0;
832 	v.metasz = 64;
833 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
834 	if (v.meta == NULL) {
835 		err = got_error_from_errno("reallocarray");
836 		goto done;
837 	}
838 
839 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
840 	    cancel_cb, cancel_arg);
841 	if (err || nobj == 0)
842 		goto done;
843 
844 	for (i = 0; i < ntheirs; i++) {
845 		struct got_object_id *id = theirs[i];
846 		if (id == NULL)
847 			continue;
848 		err = got_object_get_type(&obj_type, repo, id);
849 		if (err)
850 			return err;
851 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
852 			continue;
853 		err = load_commit(NULL, idset, id, repo,
854 		    loose_obj_only, cancel_cb, cancel_arg);
855 		if (err)
856 			goto done;
857 		if (progress_cb) {
858 			err = progress_cb(progress_arg, 0L, nours,
859 			    v.nmeta, 0, 0);
860 			if (err)
861 				goto done;
862 		}
863 	}
864 
865 	for (i = 0; i < ntheirs; i++) {
866 		struct got_object_id *id = theirs[i];
867 		int *cached_type;
868 		if (id == NULL)
869 			continue;
870 		cached_type = got_object_idset_get(idset, id);
871 		if (cached_type == NULL) {
872 			err = got_object_get_type(&obj_type, repo, id);
873 			if (err)
874 				goto done;
875 		} else
876 			obj_type = *cached_type;
877 		if (obj_type != GOT_OBJ_TYPE_TAG)
878 			continue;
879 		err = load_tag(NULL, idset, id, repo,
880 		    loose_obj_only, cancel_cb, cancel_arg);
881 		if (err)
882 			goto done;
883 		if (progress_cb) {
884 			err = progress_cb(progress_arg, 0L, nours,
885 			    v.nmeta, 0, 0);
886 			if (err)
887 				goto done;
888 		}
889 	}
890 
891 	for (i = 0; i < nobj; i++) {
892 		err = load_commit(&v, idset, ids[i], repo,
893 		    loose_obj_only, cancel_cb, cancel_arg);
894 		if (err)
895 			goto done;
896 		if (progress_cb) {
897 			err = progress_cb(progress_arg, 0L, nours,
898 			    v.nmeta, 0, 0);
899 			if (err)
900 				goto done;
901 		}
902 	}
903 
904 	for (i = 0; i < nours; i++) {
905 		struct got_object_id *id = ours[i];
906 		int *cached_type;
907 		if (id == NULL)
908 			continue;
909 		cached_type = got_object_idset_get(idset, id);
910 		if (cached_type == NULL) {
911 			err = got_object_get_type(&obj_type, repo, id);
912 			if (err)
913 				goto done;
914 		} else
915 			obj_type = *cached_type;
916 		if (obj_type != GOT_OBJ_TYPE_TAG)
917 			continue;
918 		err = load_tag(&v, idset, id, repo,
919 		    loose_obj_only, cancel_cb, cancel_arg);
920 		if (err)
921 			goto done;
922 		if (progress_cb) {
923 			err = progress_cb(progress_arg, 0L, nours,
924 			    v.nmeta, 0, 0);
925 			if (err)
926 				goto done;
927 		}
928 	}
929 
930 done:
931 	for (i = 0; i < nobj; i++) {
932 		free(ids[i]);
933 	}
934 	free(ids);
935 	got_object_idset_free(idset);
936 	if (err == NULL) {
937 		*meta = v.meta;
938 		*nmeta = v.nmeta;
939 	} else
940 		free(v.meta);
941 
942 	return err;
943 }
944 
945 const struct got_error *
946 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
947 {
948 	size_t n;
949 
950 	SHA1Update(ctx, buf, len);
951 	n = fwrite(buf, 1, len, f);
952 	if (n != len)
953 		return got_ferror(f, GOT_ERR_IO);
954 	return NULL;
955 }
956 
957 static void
958 putbe32(char *b, uint32_t n)
959 {
960 	b[0] = n >> 24;
961 	b[1] = n >> 16;
962 	b[2] = n >> 8;
963 	b[3] = n >> 0;
964 }
965 
966 static int
967 write_order_cmp(const void *pa, const void *pb)
968 {
969 	struct got_pack_meta *a, *b, *ahd, *bhd;
970 
971 	a = *(struct got_pack_meta **)pa;
972 	b = *(struct got_pack_meta **)pb;
973 	ahd = (a->head == NULL) ? a : a->head;
974 	bhd = (b->head == NULL) ? b : b->head;
975 	if (ahd->mtime != bhd->mtime)
976 		return bhd->mtime - ahd->mtime;
977 	if (ahd != bhd)
978 		return (uintptr_t)bhd - (uintptr_t)ahd;
979 	if (a->nchain != b->nchain)
980 		return a->nchain - b->nchain;
981 	return a->mtime - b->mtime;
982 }
983 
984 static const struct got_error *
985 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
986 {
987 	size_t i;
988 
989 	*hdrlen = 0;
990 
991 	hdr[0] = obj_type << 4;
992 	hdr[0] |= len & 0xf;
993 	len >>= 4;
994 	for (i = 1; len != 0; i++){
995 		if (i >= bufsize)
996 			return got_error(GOT_ERR_NO_SPACE);
997 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
998 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
999 		len >>= GOT_DELTA_SIZE_SHIFT;
1000 	}
1001 
1002 	*hdrlen = i;
1003 	return NULL;
1004 }
1005 
1006 static const struct got_error *
1007 append(char **p, int *len, int *sz, void *seg, int nseg)
1008 {
1009 	char *n;
1010 
1011 	if (*len + nseg >= *sz) {
1012 		while (*len + nseg >= *sz)
1013 			*sz += *sz / 2;
1014 		n = realloc(*p, *sz);
1015 		if (n == NULL)
1016 			return got_error_from_errno("realloc");
1017 		*p = n;
1018 	}
1019 	memcpy(*p + *len, seg, nseg);
1020 	*len += nseg;
1021 	return NULL;
1022 }
1023 
1024 
1025 static const struct got_error *
1026 encodedelta(int *nd, struct got_pack_meta *m, struct got_raw_object *o,
1027     off_t base_size, char **pp)
1028 {
1029 	const struct got_error *err = NULL;
1030 	char *p;
1031 	unsigned char buf[16], *bp;
1032 	int len, sz, i, j;
1033 	off_t n;
1034 	struct got_delta_instruction *d;
1035 
1036 	*pp = NULL;
1037 	*nd = 0;
1038 
1039 	sz = 128;
1040 	len = 0;
1041 	p = malloc(sz);
1042 	if (p == NULL)
1043 		return got_error_from_errno("malloc");
1044 
1045 	/* base object size */
1046 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1047 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1048 	for (i = 1; n > 0; i++) {
1049 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1050 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1051 		n >>= GOT_DELTA_SIZE_SHIFT;
1052 	}
1053 	err = append(&p, &len, &sz, buf, i);
1054 	if (err)
1055 		return err;
1056 
1057 	/* target object size */
1058 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1059 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1060 	for (i = 1; n > 0; i++) {
1061 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1062 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1063 		n >>= GOT_DELTA_SIZE_SHIFT;
1064 	}
1065 	err = append(&p, &len, &sz, buf, i);
1066 	if (err)
1067 		return err;
1068 	for (j = 0; j < m->ndeltas; j++) {
1069 		d = &m->deltas[j];
1070 		if (d->copy) {
1071 			n = d->offset;
1072 			bp = &buf[1];
1073 			buf[0] = GOT_DELTA_BASE_COPY;
1074 			for (i = 0; i < 4; i++) {
1075 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1076 				buf[0] |= 1 << i;
1077 				*bp++ = n & 0xff;
1078 				n >>= 8;
1079 				if (n == 0)
1080 					break;
1081 			}
1082 
1083 			n = d->len;
1084 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1085 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1086 				for (i = 0; i < 3 && n > 0; i++) {
1087 					buf[0] |= 1 << (i + 4);
1088 					*bp++ = n & 0xff;
1089 					n >>= 8;
1090 				}
1091 			}
1092 			err = append(&p, &len, &sz, buf, bp - buf);
1093 			if (err)
1094 				return err;
1095 		} else {
1096 			char content[128];
1097 			size_t r;
1098 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1099 				return got_error_from_errno("fseeko");
1100 			n = 0;
1101 			while (n != d->len) {
1102 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1103 				err = append(&p, &len, &sz, buf, 1);
1104 				if (err)
1105 					return err;
1106 				r = fread(content, 1, buf[0], o->f);
1107 				if (r != buf[0])
1108 					return got_ferror(o->f, GOT_ERR_IO);
1109 				err = append(&p, &len, &sz, content, buf[0]);
1110 				if (err)
1111 					return err;
1112 				n += buf[0];
1113 			}
1114 		}
1115 	}
1116 	*pp = p;
1117 	*nd = len;
1118 	return NULL;
1119 }
1120 
1121 static int
1122 packoff(char *hdr, off_t off)
1123 {
1124 	int i, j;
1125 	char rbuf[8];
1126 
1127 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1128 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1129 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1130 		    GOT_DELTA_SIZE_MORE;
1131 	}
1132 
1133 	j = 0;
1134 	while (i > 0)
1135 		hdr[j++] = rbuf[--i];
1136 	return j;
1137 }
1138 
1139 static const struct got_error *
1140 genpack(uint8_t *pack_sha1, FILE *packfile,
1141     struct got_pack_meta **meta, int nmeta, int nours,
1142     int use_offset_deltas, struct got_repository *repo,
1143     got_pack_progress_cb progress_cb, void *progress_arg,
1144     got_cancel_cb cancel_cb, void *cancel_arg)
1145 {
1146 	const struct got_error *err = NULL;
1147 	int i, nh, nd;
1148 	SHA1_CTX ctx;
1149 	struct got_pack_meta *m;
1150 	struct got_raw_object *raw;
1151 	char *p = NULL, buf[32];
1152 	size_t outlen, n;
1153 	struct got_deflate_checksum csum;
1154 	off_t packfile_size = 0;
1155 
1156 	SHA1Init(&ctx);
1157 	csum.output_sha1 = &ctx;
1158 	csum.output_crc = NULL;
1159 
1160 	err = hwrite(packfile, "PACK", 4, &ctx);
1161 	if (err)
1162 		return err;
1163 	putbe32(buf, GOT_PACKFILE_VERSION);
1164 	err = hwrite(packfile, buf, 4, &ctx);
1165 	if (err)
1166 		goto done;
1167 	putbe32(buf, nmeta);
1168 	err = hwrite(packfile, buf, 4, &ctx);
1169 	if (err)
1170 		goto done;
1171 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1172 	for (i = 0; i < nmeta; i++) {
1173 		if (progress_cb) {
1174 			err = progress_cb(progress_arg, packfile_size, nours,
1175 			    nmeta, nmeta, i);
1176 			if (err)
1177 				goto done;
1178 		}
1179 		m = meta[i];
1180 		m->off = ftello(packfile);
1181 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
1182 		if (err)
1183 			goto done;
1184 		if (m->deltas == NULL) {
1185 			err = packhdr(&nh, buf, sizeof(buf),
1186 			    m->obj_type, raw->size);
1187 			if (err)
1188 				goto done;
1189 			err = hwrite(packfile, buf, nh, &ctx);
1190 			if (err)
1191 				goto done;
1192 			packfile_size += nh;
1193 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1194 				err = got_error_from_errno("fseeko");
1195 				goto done;
1196 			}
1197 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1198 			    &csum);
1199 			if (err)
1200 				goto done;
1201 			packfile_size += outlen;
1202 		} else {
1203 			FILE *delta_file;
1204 			struct got_raw_object *base_raw;
1205 			err = got_object_raw_open(&base_raw, repo,
1206 			    &m->prev->id, 8192);
1207 			if (err)
1208 				goto done;
1209 			err = encodedelta(&nd, m, raw, base_raw->size, &p);
1210 			if (err)
1211 				goto done;
1212 			got_object_raw_close(base_raw);
1213 			if (use_offset_deltas && m->prev->off != 0) {
1214 				err = packhdr(&nh, buf, sizeof(buf),
1215 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1216 				if (err)
1217 					goto done;
1218 				nh += packoff(buf + nh,
1219 				    m->off - m->prev->off);
1220 				err = hwrite(packfile, buf, nh, &ctx);
1221 				if (err)
1222 					goto done;
1223 				packfile_size += nh;
1224 			} else {
1225 				err = packhdr(&nh, buf, sizeof(buf),
1226 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1227 				err = hwrite(packfile, buf, nh, &ctx);
1228 				if (err)
1229 					goto done;
1230 				packfile_size += nh;
1231 				err = hwrite(packfile, m->prev->id.sha1,
1232 				    sizeof(m->prev->id.sha1), &ctx);
1233 				packfile_size += sizeof(m->prev->id.sha1);
1234 				if (err)
1235 					goto done;
1236 			}
1237 			/* XXX need got_deflate_from_mem() */
1238 			delta_file = fmemopen(p, nd, "r");
1239 			if (delta_file == NULL) {
1240 				err = got_error_from_errno("fmemopen");
1241 				goto done;
1242 			}
1243 			err = got_deflate_to_file(&outlen, delta_file,
1244 			    packfile, &csum);
1245 			fclose(delta_file);
1246 			if (err)
1247 				goto done;
1248 			packfile_size += outlen;
1249 			free(p);
1250 			p = NULL;
1251 		}
1252 		got_object_raw_close(raw);
1253 		raw = NULL;
1254 	}
1255 	SHA1Final(pack_sha1, &ctx);
1256 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1257 	if (n != SHA1_DIGEST_LENGTH)
1258 		err = got_ferror(packfile, GOT_ERR_IO);
1259 	packfile_size += SHA1_DIGEST_LENGTH;
1260 	packfile_size += sizeof(struct got_packfile_hdr);
1261 	err = progress_cb(progress_arg, packfile_size, nours,
1262 	    nmeta, nmeta, nmeta);
1263 	if (err)
1264 		goto done;
1265 done:
1266 	free(p);
1267 	return err;
1268 }
1269 
1270 const struct got_error *
1271 got_pack_create(uint8_t *packsha1, FILE *packfile,
1272     struct got_object_id **theirs, int ntheirs,
1273     struct got_object_id **ours, int nours,
1274     struct got_repository *repo, int loose_obj_only, int allow_empty,
1275     got_pack_progress_cb progress_cb, void *progress_arg,
1276     got_cancel_cb cancel_cb, void *cancel_arg)
1277 {
1278 	const struct got_error *err;
1279 	struct got_pack_meta **meta;
1280 	int nmeta;
1281 
1282 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1283 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1284 	if (err)
1285 		return err;
1286 
1287 	if (nmeta == 0 && !allow_empty) {
1288 		err = got_error(GOT_ERR_CANNOT_PACK);
1289 		goto done;
1290 	}
1291 	if (nmeta > 0) {
1292 		err = pick_deltas(meta, nmeta, nours, repo,
1293 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1294 		if (err)
1295 			goto done;
1296 	}
1297 
1298 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1299 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1300 	if (err)
1301 		goto done;
1302 done:
1303 	free_nmeta(meta, nmeta);
1304 	return err;
1305 }