op public repos

Blob

Date:: Wed Sep 22 13:32:37 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/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_submodule(e) ||
376 		    got_object_idset_contains(idset, id))
377 			continue;
378 		
379 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
380 		    got_tree_entry_get_name(e)) == -1) {
381 			err = got_error_from_errno("asprintf");
382 			break;
383 		}
384 
385 		if (S_ISDIR(mode)) {
386 			struct got_object_qid *qid;
387 			err = got_object_qid_alloc(&qid, id);
388 			if (err)
389 				break;
390 			STAILQ_INSERT_TAIL(ids, qid, entry);
391 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
392 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
393 			    mtime, loose_obj_only, repo);
394 			if (err)
395 				break;
396 		}
397 		free(p);
398 		p = NULL;
399 	}
400 
401 	got_object_tree_close(tree);
402 	free(p);
403 	return err;
404 }
405 
406 static const struct got_error *
407 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
408     struct got_object_id *tree_id, const char *dpath, time_t mtime,
409     int loose_obj_only, struct got_repository *repo,
410     got_cancel_cb cancel_cb, void *cancel_arg)
411 {
412 	const struct got_error *err = NULL;
413 	struct got_object_id_queue tree_ids;
414 	struct got_object_qid *qid;
415 
416 	if (got_object_idset_contains(idset, tree_id))
417 		return NULL;
418 
419 	err = got_object_qid_alloc(&qid, tree_id);
420 	if (err)
421 		return err;
422 
423 	STAILQ_INIT(&tree_ids);
424 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
425 
426 	while (!STAILQ_EMPTY(&tree_ids)) {
427 		if (cancel_cb) {
428 			err = (*cancel_cb)(cancel_arg);
429 			if (err)
430 				break;
431 		}
432 
433 		qid = STAILQ_FIRST(&tree_ids);
434 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
435 
436 		if (got_object_idset_contains(idset, qid->id)) {
437 			got_object_qid_free(qid);
438 			continue;
439 		}
440 
441 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
442 		    mtime, loose_obj_only, repo);
443 		if (err) {
444 			got_object_qid_free(qid);
445 			break;
446 		}
447 
448 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
449 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
450 		got_object_qid_free(qid);
451 		if (err)
452 			break;
453 	}
454 
455 	got_object_id_queue_free(&tree_ids);
456 	return err;
457 }
458 
459 static const struct got_error *
460 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
461     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
462     got_cancel_cb cancel_cb, void *cancel_arg)
463 {
464 	const struct got_error *err;
465 	struct got_commit_object *commit;
466 
467 	if (got_object_idset_contains(idset, id))
468 		return NULL;
469 
470 	if (loose_obj_only) {
471 		int is_packed;
472 		err = search_packidx(&is_packed, id, repo);
473 		if (err)
474 			return err;
475 		if (is_packed)
476 			return NULL;
477 	}
478 
479 	err = got_object_open_as_commit(&commit, repo, id);
480 	if (err)
481 		return err;
482 
483 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
484 	    got_object_commit_get_committer_time(commit),
485 	    loose_obj_only, repo);
486 	if (err)
487 		goto done;
488 
489 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
490 	    "", got_object_commit_get_committer_time(commit),
491 	    loose_obj_only, repo, cancel_cb, cancel_arg);
492 done:
493 	got_object_commit_close(commit);
494 	return err;
495 }
496 
497 static const struct got_error *
498 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
499     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
500     got_cancel_cb cancel_cb, void *cancel_arg)
501 {
502 	const struct got_error *err;
503 	struct got_tag_object *tag = NULL;
504 
505 	if (got_object_idset_contains(idset, id))
506 		return NULL;
507 
508 	if (loose_obj_only) {
509 		int is_packed;
510 		err = search_packidx(&is_packed, id, repo);
511 		if (err)
512 			return err;
513 		if (is_packed)
514 			return NULL;
515 	}
516 
517 	err = got_object_open_as_tag(&tag, repo, id);
518 	if (err)
519 		return err;
520 
521 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
522 	    got_object_tag_get_tagger_time(tag),
523 	    loose_obj_only, repo);
524 	if (err)
525 		goto done;
526 
527 	switch (got_object_tag_get_object_type(tag)) {
528 	case GOT_OBJ_TYPE_COMMIT:
529 		err = load_commit(v, idset,
530 		    got_object_tag_get_object_id(tag), repo,
531 		    loose_obj_only, cancel_cb, cancel_arg);
532 		break;
533 	case GOT_OBJ_TYPE_TREE:
534 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
535 		    "", got_object_tag_get_tagger_time(tag),
536 		    loose_obj_only, repo, cancel_cb, cancel_arg);
537 		break;
538 	default:
539 		break;
540 	}
541 
542 done:
543 	got_object_tag_close(tag);
544 	return err;
545 }
546 
547 enum findtwixt_color {
548 	COLOR_KEEP = 0,
549 	COLOR_DROP,
550 	COLOR_BLANK,
551 };
552 static const int findtwixt_colors[] = {
553 	COLOR_KEEP,
554 	COLOR_DROP,
555 	COLOR_BLANK
556 };
557 
558 static const struct got_error *
559 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
560     int color, struct got_repository *repo)
561 {
562 	const struct got_error *err;
563 	struct got_object_qid *qid;
564 
565 	err = got_object_qid_alloc(&qid, id);
566 	if (err)
567 		return err;
568 
569 	STAILQ_INSERT_TAIL(ids, qid, entry);
570 	qid->data = (void *)&findtwixt_colors[color];
571 	return NULL;
572 }
573 
574 static const struct got_error *
575 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
576     struct got_object_id *id, struct got_repository *repo,
577     got_cancel_cb cancel_cb, void *cancel_arg)
578 {
579 	const struct got_error *err = NULL;
580 	struct got_commit_object *commit;
581 	const struct got_object_id_queue *parents;
582 	struct got_object_id_queue ids;
583 	struct got_object_qid *qid;
584 
585 	STAILQ_INIT(&ids);
586 
587 	err = got_object_qid_alloc(&qid, id);
588 	if (err)
589 		return err;
590 	STAILQ_INSERT_HEAD(&ids, qid, entry);
591 
592 	while (!STAILQ_EMPTY(&ids)) {
593 		if (cancel_cb) {
594 			err = (*cancel_cb)(cancel_arg);
595 			if (err)
596 				break;
597 		}
598 
599 		qid = STAILQ_FIRST(&ids);
600 		STAILQ_REMOVE_HEAD(&ids, entry);
601 
602 		if (got_object_idset_contains(drop, qid->id)) {
603 			got_object_qid_free(qid);
604 			continue;
605 		}
606 
607 		err = got_object_idset_add(drop, qid->id,
608 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
609 		if (err) {
610 			got_object_qid_free(qid);
611 			break;
612 		}
613 
614 		if (!got_object_idset_contains(keep, qid->id)) {
615 			got_object_qid_free(qid);
616 			continue;
617 		}
618 
619 		err = got_object_open_as_commit(&commit, repo, qid->id);
620 		got_object_qid_free(qid);
621 		if (err)
622 			break;
623 
624 		parents = got_object_commit_get_parent_ids(commit);
625 		if (parents) {
626 			err = got_object_id_queue_copy(parents, &ids);
627 			if (err) {
628 				got_object_commit_close(commit);
629 				break;
630 			}
631 		}
632 		got_object_commit_close(commit);
633 	}
634 
635 	got_object_id_queue_free(&ids);
636 	return err;
637 }
638 
639 struct append_id_arg {
640 	struct got_object_id **array;
641 	int idx;
642 };
643 
644 static const struct got_error *
645 append_id(struct got_object_id *id, void *data, void *arg)
646 {
647 	struct append_id_arg *a = arg;
648 
649 	a->array[a->idx] = got_object_id_dup(id);
650 	if (a->array[a->idx] == NULL)
651 		return got_error_from_errno("got_object_id_dup");
652 
653 	a->idx++;
654 	return NULL;
655 }
656 
657 static const struct got_error *
658 findtwixt(struct got_object_id ***res, int *nres,
659     struct got_object_id **head, int nhead,
660     struct got_object_id **tail, int ntail,
661     struct got_repository *repo,
662     got_cancel_cb cancel_cb, void *cancel_arg)
663 {
664 	const struct got_error *err = NULL;
665 	struct got_object_id_queue ids;
666 	struct got_object_idset *keep, *drop;
667 	struct got_object_qid *qid;
668 	int i, ncolor, nkeep, obj_type;
669 
670 	STAILQ_INIT(&ids);
671 	*res = NULL;
672 	*nres = 0;
673 
674 	keep = got_object_idset_alloc();
675 	if (keep == NULL)
676 		return got_error_from_errno("got_object_idset_alloc");
677 
678 	drop = got_object_idset_alloc();
679 	if (drop == NULL) {
680 		err = got_error_from_errno("got_object_idset_alloc");
681 		goto done;
682 	}
683 
684 	for (i = 0; i < nhead; i++) {
685 		struct got_object_id *id = head[i];
686 		if (id == NULL)
687 			continue;
688 		err = got_object_get_type(&obj_type, repo, id);
689 		if (err)
690 			return err;
691 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
692 			continue;
693 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
694 		if (err)
695 			goto done;
696 	}		
697 	for (i = 0; i < ntail; i++) {
698 		struct got_object_id *id = tail[i];
699 		if (id == NULL)
700 			continue;
701 		err = got_object_get_type(&obj_type, repo, id);
702 		if (err)
703 			return err;
704 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
705 			continue;
706 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
707 		if (err)
708 			goto done;
709 	}
710 
711 	while (!STAILQ_EMPTY(&ids)) {
712 		int qcolor;
713 		qid = STAILQ_FIRST(&ids);
714 		qcolor = *((int *)qid->data);
715 
716 		if (got_object_idset_contains(drop, qid->id))
717 			ncolor = COLOR_DROP;
718 		else if (got_object_idset_contains(keep, qid->id))
719 			ncolor = COLOR_KEEP;
720 		else
721 			ncolor = COLOR_BLANK;
722 
723 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
724 		    qcolor == COLOR_KEEP)) {
725 			STAILQ_REMOVE_HEAD(&ids, entry);
726 			got_object_qid_free(qid);
727 			continue;
728 		}
729 
730 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
731 			err = drop_commit(keep, drop, qid->id, repo,
732 			    cancel_cb, cancel_arg);
733 			if (err)
734 				goto done;
735 		} else if (ncolor == COLOR_BLANK) {
736 			struct got_commit_object *commit;
737 			struct got_object_id *id;
738 			const struct got_object_id_queue *parents;
739 			struct got_object_qid *pid;
740 
741 			id = got_object_id_dup(qid->id);
742 			if (id == NULL) {
743 				err = got_error_from_errno("got_object_id_dup");
744 				goto done;
745 			}
746 			if (qcolor == COLOR_KEEP)
747 				err = got_object_idset_add(keep, id,
748 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
749 			else
750 				err = got_object_idset_add(drop, id,
751 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
752 			if (err) {
753 				free(id);
754 				goto done;
755 			}
756 
757 			err = got_object_open_as_commit(&commit, repo, id);
758 			if (err) {
759 				free(id);
760 				goto done;
761 			}
762 			parents = got_object_commit_get_parent_ids(commit);
763 			if (parents) {
764 				STAILQ_FOREACH(pid, parents, entry) {
765 					err = queue_commit_id(&ids, pid->id,
766 					    qcolor, repo);
767 					if (err) {
768 						free(id);
769 						goto done;
770 					}
771 				}
772 			}
773 			got_object_commit_close(commit);
774 			commit = NULL;
775 		} else {
776 			/* should not happen */
777 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
778 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
779 			goto done;
780 		}
781 
782 		STAILQ_REMOVE_HEAD(&ids, entry);
783 		got_object_qid_free(qid);
784 	}
785 
786 	nkeep = got_object_idset_num_elements(keep);
787 	if (nkeep > 0) {
788 		struct append_id_arg arg;
789 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
790 		if (arg.array == NULL) {
791 			err = got_error_from_errno("calloc");
792 			goto done;
793 		}
794 		arg.idx = 0;
795 		err = got_object_idset_for_each(keep, append_id, &arg);
796 		if (err) {
797 			free(arg.array);
798 			goto done;
799 		}
800 		*res = arg.array;
801 		*nres = nkeep;
802 	}
803 done:
804 	got_object_idset_free(keep);
805 	got_object_idset_free(drop);
806 	got_object_id_queue_free(&ids);
807 	return err;
808 }
809 
810 static const struct got_error *
811 read_meta(struct got_pack_meta ***meta, int *nmeta,
812     struct got_object_id **theirs, int ntheirs,
813     struct got_object_id **ours, int nours, struct got_repository *repo,
814     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
815     got_cancel_cb cancel_cb, void *cancel_arg)
816 {
817 	const struct got_error *err = NULL;
818 	struct got_object_id **ids = NULL;
819 	struct got_object_idset *idset;
820 	int i, nobj = 0, obj_type;
821 	struct got_pack_metavec v;
822 
823 	*meta = NULL;
824 	*nmeta = 0;
825 
826 	idset = got_object_idset_alloc();
827 	if (idset == NULL)
828 		return got_error_from_errno("got_object_idset_alloc");
829 
830 	v.nmeta = 0;
831 	v.metasz = 64;
832 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
833 	if (v.meta == NULL) {
834 		err = got_error_from_errno("reallocarray");
835 		goto done;
836 	}
837 
838 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
839 	    cancel_cb, cancel_arg);
840 	if (err || nobj == 0)
841 		goto done;
842 
843 	for (i = 0; i < ntheirs; i++) {
844 		struct got_object_id *id = theirs[i];
845 		if (id == NULL)
846 			continue;
847 		err = got_object_get_type(&obj_type, repo, id);
848 		if (err)
849 			return err;
850 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
851 			continue;
852 		err = load_commit(NULL, idset, id, repo,
853 		    loose_obj_only, cancel_cb, cancel_arg);
854 		if (err)
855 			goto done;
856 		if (progress_cb) {
857 			err = progress_cb(progress_arg, 0L, nours,
858 			    v.nmeta, 0, 0);
859 			if (err)
860 				goto done;
861 		}
862 	}
863 
864 	for (i = 0; i < ntheirs; i++) {
865 		struct got_object_id *id = theirs[i];
866 		int *cached_type;
867 		if (id == NULL)
868 			continue;
869 		cached_type = got_object_idset_get(idset, id);
870 		if (cached_type == NULL) {
871 			err = got_object_get_type(&obj_type, repo, id);
872 			if (err)
873 				goto done;
874 		} else
875 			obj_type = *cached_type;
876 		if (obj_type != GOT_OBJ_TYPE_TAG)
877 			continue;
878 		err = load_tag(NULL, idset, id, repo,
879 		    loose_obj_only, cancel_cb, cancel_arg);
880 		if (err)
881 			goto done;
882 		if (progress_cb) {
883 			err = progress_cb(progress_arg, 0L, nours,
884 			    v.nmeta, 0, 0);
885 			if (err)
886 				goto done;
887 		}
888 	}
889 
890 	for (i = 0; i < nobj; i++) {
891 		err = load_commit(&v, idset, ids[i], repo,
892 		    loose_obj_only, cancel_cb, cancel_arg);
893 		if (err)
894 			goto done;
895 		if (progress_cb) {
896 			err = progress_cb(progress_arg, 0L, nours,
897 			    v.nmeta, 0, 0);
898 			if (err)
899 				goto done;
900 		}
901 	}
902 
903 	for (i = 0; i < nours; i++) {
904 		struct got_object_id *id = ours[i];
905 		int *cached_type;
906 		if (id == NULL)
907 			continue;
908 		cached_type = got_object_idset_get(idset, id);
909 		if (cached_type == NULL) {
910 			err = got_object_get_type(&obj_type, repo, id);
911 			if (err)
912 				goto done;
913 		} else
914 			obj_type = *cached_type;
915 		if (obj_type != GOT_OBJ_TYPE_TAG)
916 			continue;
917 		err = load_tag(&v, idset, id, repo,
918 		    loose_obj_only, cancel_cb, cancel_arg);
919 		if (err)
920 			goto done;
921 		if (progress_cb) {
922 			err = progress_cb(progress_arg, 0L, nours,
923 			    v.nmeta, 0, 0);
924 			if (err)
925 				goto done;
926 		}
927 	}
928 
929 done:
930 	for (i = 0; i < nobj; i++) {
931 		free(ids[i]);
932 	}
933 	free(ids);
934 	got_object_idset_free(idset);
935 	if (err == NULL) {
936 		*meta = v.meta;
937 		*nmeta = v.nmeta;
938 	} else
939 		free(v.meta);
940 
941 	return err;
942 }
943 
944 const struct got_error *
945 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
946 {
947 	size_t n;
948 
949 	SHA1Update(ctx, buf, len);
950 	n = fwrite(buf, 1, len, f);
951 	if (n != len)
952 		return got_ferror(f, GOT_ERR_IO);
953 	return NULL;
954 }
955 
956 static void
957 putbe32(char *b, uint32_t n)
958 {
959 	b[0] = n >> 24;
960 	b[1] = n >> 16;
961 	b[2] = n >> 8;
962 	b[3] = n >> 0;
963 }
964 
965 static int
966 write_order_cmp(const void *pa, const void *pb)
967 {
968 	struct got_pack_meta *a, *b, *ahd, *bhd;
969 
970 	a = *(struct got_pack_meta **)pa;
971 	b = *(struct got_pack_meta **)pb;
972 	ahd = (a->head == NULL) ? a : a->head;
973 	bhd = (b->head == NULL) ? b : b->head;
974 	if (ahd->mtime != bhd->mtime)
975 		return bhd->mtime - ahd->mtime;
976 	if (ahd != bhd)
977 		return (uintptr_t)bhd - (uintptr_t)ahd;
978 	if (a->nchain != b->nchain)
979 		return a->nchain - b->nchain;
980 	return a->mtime - b->mtime;
981 }
982 
983 static const struct got_error *
984 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
985 {
986 	size_t i;
987 
988 	*hdrlen = 0;
989 
990 	hdr[0] = obj_type << 4;
991 	hdr[0] |= len & 0xf;
992 	len >>= 4;
993 	for (i = 1; len != 0; i++){
994 		if (i >= bufsize)
995 			return got_error(GOT_ERR_NO_SPACE);
996 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
997 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
998 		len >>= GOT_DELTA_SIZE_SHIFT;
999 	}
1000 
1001 	*hdrlen = i;
1002 	return NULL;
1003 }
1004 
1005 static const struct got_error *
1006 append(char **p, int *len, int *sz, void *seg, int nseg)
1007 {
1008 	char *n;
1009 
1010 	if (*len + nseg >= *sz) {
1011 		while (*len + nseg >= *sz)
1012 			*sz += *sz / 2;
1013 		n = realloc(*p, *sz);
1014 		if (n == NULL)
1015 			return got_error_from_errno("realloc");
1016 		*p = n;
1017 	}
1018 	memcpy(*p + *len, seg, nseg);
1019 	*len += nseg;
1020 	return NULL;
1021 }
1022 
1023 
1024 static const struct got_error *
1025 encodedelta(int *nd, struct got_pack_meta *m, struct got_raw_object *o,
1026     off_t base_size, char **pp)
1027 {
1028 	const struct got_error *err = NULL;
1029 	char *p;
1030 	unsigned char buf[16], *bp;
1031 	int len, sz, i, j;
1032 	off_t n;
1033 	struct got_delta_instruction *d;
1034 
1035 	*pp = NULL;
1036 	*nd = 0;
1037 
1038 	sz = 128;
1039 	len = 0;
1040 	p = malloc(sz);
1041 	if (p == NULL)
1042 		return got_error_from_errno("malloc");
1043 
1044 	/* base object size */
1045 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1046 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1047 	for (i = 1; n > 0; i++) {
1048 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1049 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1050 		n >>= GOT_DELTA_SIZE_SHIFT;
1051 	}
1052 	err = append(&p, &len, &sz, buf, i);
1053 	if (err)
1054 		return err;
1055 
1056 	/* target object size */
1057 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1058 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1059 	for (i = 1; n > 0; i++) {
1060 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1061 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1062 		n >>= GOT_DELTA_SIZE_SHIFT;
1063 	}
1064 	err = append(&p, &len, &sz, buf, i);
1065 	if (err)
1066 		return err;
1067 	for (j = 0; j < m->ndeltas; j++) {
1068 		d = &m->deltas[j];
1069 		if (d->copy) {
1070 			n = d->offset;
1071 			bp = &buf[1];
1072 			buf[0] = GOT_DELTA_BASE_COPY;
1073 			for (i = 0; i < 4; i++) {
1074 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1075 				buf[0] |= 1 << i;
1076 				*bp++ = n & 0xff;
1077 				n >>= 8;
1078 				if (n == 0)
1079 					break;
1080 			}
1081 
1082 			n = d->len;
1083 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1084 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1085 				for (i = 0; i < 3 && n > 0; i++) {
1086 					buf[0] |= 1 << (i + 4);
1087 					*bp++ = n & 0xff;
1088 					n >>= 8;
1089 				}
1090 			}
1091 			err = append(&p, &len, &sz, buf, bp - buf);
1092 			if (err)
1093 				return err;
1094 		} else {
1095 			char content[128];
1096 			size_t r;
1097 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1098 				return got_error_from_errno("fseeko");
1099 			n = 0;
1100 			while (n != d->len) {
1101 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1102 				err = append(&p, &len, &sz, buf, 1);
1103 				if (err)
1104 					return err;
1105 				r = fread(content, 1, buf[0], o->f);
1106 				if (r != buf[0])
1107 					return got_ferror(o->f, GOT_ERR_IO);
1108 				err = append(&p, &len, &sz, content, buf[0]);
1109 				if (err)
1110 					return err;
1111 				n += buf[0];
1112 			}
1113 		}
1114 	}
1115 	*pp = p;
1116 	*nd = len;
1117 	return NULL;
1118 }
1119 
1120 static int
1121 packoff(char *hdr, off_t off)
1122 {
1123 	int i, j;
1124 	char rbuf[8];
1125 
1126 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1127 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1128 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1129 		    GOT_DELTA_SIZE_MORE;
1130 	}
1131 
1132 	j = 0;
1133 	while (i > 0)
1134 		hdr[j++] = rbuf[--i];
1135 	return j;
1136 }
1137 
1138 static const struct got_error *
1139 genpack(uint8_t *pack_sha1, FILE *packfile,
1140     struct got_pack_meta **meta, int nmeta, int nours,
1141     int use_offset_deltas, struct got_repository *repo,
1142     got_pack_progress_cb progress_cb, void *progress_arg,
1143     got_cancel_cb cancel_cb, void *cancel_arg)
1144 {
1145 	const struct got_error *err = NULL;
1146 	int i, nh, nd;
1147 	SHA1_CTX ctx;
1148 	struct got_pack_meta *m;
1149 	struct got_raw_object *raw;
1150 	char *p = NULL, buf[32];
1151 	size_t outlen, n;
1152 	struct got_deflate_checksum csum;
1153 	off_t packfile_size = 0;
1154 
1155 	SHA1Init(&ctx);
1156 	csum.output_sha1 = &ctx;
1157 	csum.output_crc = NULL;
1158 
1159 	err = hwrite(packfile, "PACK", 4, &ctx);
1160 	if (err)
1161 		return err;
1162 	putbe32(buf, GOT_PACKFILE_VERSION);
1163 	err = hwrite(packfile, buf, 4, &ctx);
1164 	if (err)
1165 		goto done;
1166 	putbe32(buf, nmeta);
1167 	err = hwrite(packfile, buf, 4, &ctx);
1168 	if (err)
1169 		goto done;
1170 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1171 	for (i = 0; i < nmeta; i++) {
1172 		if (progress_cb) {
1173 			err = progress_cb(progress_arg, packfile_size, nours,
1174 			    nmeta, nmeta, i);
1175 			if (err)
1176 				goto done;
1177 		}
1178 		m = meta[i];
1179 		m->off = ftello(packfile);
1180 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
1181 		if (err)
1182 			goto done;
1183 		if (m->deltas == NULL) {
1184 			err = packhdr(&nh, buf, sizeof(buf),
1185 			    m->obj_type, raw->size);
1186 			if (err)
1187 				goto done;
1188 			err = hwrite(packfile, buf, nh, &ctx);
1189 			if (err)
1190 				goto done;
1191 			packfile_size += nh;
1192 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1193 				err = got_error_from_errno("fseeko");
1194 				goto done;
1195 			}
1196 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1197 			    &csum);
1198 			if (err)
1199 				goto done;
1200 			packfile_size += outlen;
1201 		} else {
1202 			FILE *delta_file;
1203 			struct got_raw_object *base_raw;
1204 			err = got_object_raw_open(&base_raw, repo,
1205 			    &m->prev->id, 8192);
1206 			if (err)
1207 				goto done;
1208 			err = encodedelta(&nd, m, raw, base_raw->size, &p);
1209 			if (err)
1210 				goto done;
1211 			got_object_raw_close(base_raw);
1212 			if (use_offset_deltas && m->prev->off != 0) {
1213 				err = packhdr(&nh, buf, sizeof(buf),
1214 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1215 				if (err)
1216 					goto done;
1217 				nh += packoff(buf + nh,
1218 				    m->off - m->prev->off);
1219 				err = hwrite(packfile, buf, nh, &ctx);
1220 				if (err)
1221 					goto done;
1222 				packfile_size += nh;
1223 			} else {
1224 				err = packhdr(&nh, buf, sizeof(buf),
1225 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1226 				err = hwrite(packfile, buf, nh, &ctx);
1227 				if (err)
1228 					goto done;
1229 				packfile_size += nh;
1230 				err = hwrite(packfile, m->prev->id.sha1,
1231 				    sizeof(m->prev->id.sha1), &ctx);
1232 				packfile_size += sizeof(m->prev->id.sha1);
1233 				if (err)
1234 					goto done;
1235 			}
1236 			/* XXX need got_deflate_from_mem() */
1237 			delta_file = fmemopen(p, nd, "r");
1238 			if (delta_file == NULL) {
1239 				err = got_error_from_errno("fmemopen");
1240 				goto done;
1241 			}
1242 			err = got_deflate_to_file(&outlen, delta_file,
1243 			    packfile, &csum);
1244 			fclose(delta_file);
1245 			if (err)
1246 				goto done;
1247 			packfile_size += outlen;
1248 			free(p);
1249 			p = NULL;
1250 		}
1251 		got_object_raw_close(raw);
1252 		raw = NULL;
1253 	}
1254 	SHA1Final(pack_sha1, &ctx);
1255 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1256 	if (n != SHA1_DIGEST_LENGTH)
1257 		err = got_ferror(packfile, GOT_ERR_IO);
1258 	packfile_size += SHA1_DIGEST_LENGTH;
1259 	packfile_size += sizeof(struct got_packfile_hdr);
1260 	err = progress_cb(progress_arg, packfile_size, nours,
1261 	    nmeta, nmeta, nmeta);
1262 	if (err)
1263 		goto done;
1264 done:
1265 	free(p);
1266 	return err;
1267 }
1268 
1269 const struct got_error *
1270 got_pack_create(uint8_t *packsha1, FILE *packfile,
1271     struct got_object_id **theirs, int ntheirs,
1272     struct got_object_id **ours, int nours,
1273     struct got_repository *repo, int loose_obj_only, int allow_empty,
1274     got_pack_progress_cb progress_cb, void *progress_arg,
1275     got_cancel_cb cancel_cb, void *cancel_arg)
1276 {
1277 	const struct got_error *err;
1278 	struct got_pack_meta **meta;
1279 	int nmeta;
1280 
1281 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1282 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1283 	if (err)
1284 		return err;
1285 
1286 	if (nmeta == 0 && !allow_empty) {
1287 		err = got_error(GOT_ERR_CANNOT_PACK);
1288 		goto done;
1289 	}
1290 	if (nmeta > 0) {
1291 		err = pick_deltas(meta, nmeta, nours, repo,
1292 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1293 		if (err)
1294 			goto done;
1295 	}
1296 
1297 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1298 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1299 	if (err)
1300 		goto done;
1301 done:
1302 	free_nmeta(meta, nmeta);
1303 	return err;
1304 }