Gotweb

Blob

Date:: Tue Sep 21 20:34:22 2021 UTC
Message:: portable: initial Linux compilation This commit modifies the GoT main branch to be able to compile it under linux.
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 #include <sys/uio.h>
20 #include <sys/stat.h>
21 
22 #include <stdint.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <sha1.h>
27 #include <limits.h>
28 #include <zlib.h>
29 
30 #include "got_error.h"
31 #include "got_cancel.h"
32 #include "got_object.h"
33 #include "got_reference.h"
34 #include "got_repository_admin.h"
35 
36 #include "got_lib_deltify.h"
37 #include "got_lib_delta.h"
38 #include "got_lib_object.h"
39 #include "got_lib_object_idset.h"
40 #include "got_lib_object_cache.h"
41 #include "got_lib_deflate.h"
42 #include "got_lib_pack.h"
43 #include "got_lib_privsep.h"
44 #include "got_lib_repository.h"
45 
46 #ifndef MAX
47 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
48 #endif
49 
50 struct got_pack_meta {
51 	struct got_object_id id;
52 	char	*path;
53 	int	obj_type;
54 	time_t	mtime;
55 
56 	/* The best delta we picked */
57 	struct got_pack_meta *head;
58 	struct got_pack_meta *prev;
59 	struct got_delta_instruction *deltas;
60 	int	ndeltas;
61 	int	nchain;
62 
63 	/* Only used for delta window */
64 	struct got_delta_table *dtab;
65 
66 	/* Only used for writing offset deltas */
67 	off_t	off;
68 };
69 
70 struct got_pack_metavec {
71 	struct got_pack_meta **meta;
72 	int nmeta;
73 	int metasz;
74 };
75 
76 static const struct got_error *
77 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
78     const char *path, int obj_type, time_t mtime)
79 {
80 	const struct got_error *err = NULL;
81 	struct got_pack_meta *m;
82 
83 	*new = NULL;
84 
85 	m = calloc(1, sizeof(*m));
86 	if (m == NULL)
87 		return got_error_from_errno("malloc");
88 
89 	memcpy(&m->id, id, sizeof(m->id));
90 
91 	m->path = strdup(path);
92 	if (m->path == NULL) {
93 		err = got_error_from_errno("strdup");
94 		free(m);
95 		return err;
96 	}
97 
98 	m->obj_type = obj_type;
99 	m->mtime = mtime;
100 	*new = m;
101 	return NULL;
102 }
103 
104 static void
105 clear_meta(struct got_pack_meta *meta)
106 {
107 	if (meta == NULL)
108 		return;
109 	free(meta->deltas);
110 	meta->deltas = NULL;
111 	free(meta->path);
112 	meta->path = NULL;
113 }
114 
115 static void
116 free_nmeta(struct got_pack_meta **meta, int nmeta)
117 {
118 	int i;
119 
120 	for (i = 0; i < nmeta; i++)
121 		clear_meta(meta[i]);
122 	free(meta);
123 }
124 
125 static int
126 delta_order_cmp(const void *pa, const void *pb)
127 {
128 	struct got_pack_meta *a, *b;
129 	int cmp;
130 
131 	a = *(struct got_pack_meta **)pa;
132 	b = *(struct got_pack_meta **)pb;
133 
134 	if (a->obj_type != b->obj_type)
135 		return a->obj_type - b->obj_type;
136 	cmp = strcmp(a->path, b->path);
137 	if (cmp != 0)
138 		return cmp;
139 	if (a->mtime != b->mtime)
140 		return a->mtime - b->mtime;
141 	return got_object_id_cmp(&a->id, &b->id);
142 }
143 
144 static int
145 delta_size(struct got_delta_instruction *deltas, int ndeltas)
146 {
147 	int i, size = 32;
148 	for (i = 0; i < ndeltas; i++) {
149 		if (deltas[i].copy)
150 			size += GOT_DELTA_SIZE_SHIFT;
151 		else
152 			size += deltas[i].len + 1;
153 	}
154 	return size;
155 }
156 
157 
158 static const struct got_error *
159 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
160     struct got_repository *repo,
161     got_pack_progress_cb progress_cb, void *progress_arg,
162     got_cancel_cb cancel_cb, void *cancel_arg)
163 {
164 	const struct got_error *err = NULL;
165 	struct got_pack_meta *m = NULL, *base = NULL;
166 	struct got_raw_object *raw = NULL, *base_raw = NULL;
167 	struct got_delta_instruction *deltas;
168 	int i, j, size, ndeltas, best;
169 	const int max_base_candidates = 10;
170 
171 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
172 	for (i = 0; i < nmeta; i++) {
173 		if (cancel_cb) {
174 			err = (*cancel_cb)(cancel_arg);
175 			if (err)
176 				break;
177 		}
178 		if (progress_cb) {
179 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
180 			if (err)
181 				goto done;
182 		}
183 		m = meta[i];
184 		m->deltas = NULL;
185 		m->ndeltas = 0;
186 
187 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
188 		    m->obj_type == GOT_OBJ_TYPE_TAG)
189 			continue;
190 
191 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
192 		if (err)
193 			goto done;
194 
195 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
196 		    raw->size + raw->hdrlen);
197 		if (err)
198 			goto done;
199 
200 		if (i > max_base_candidates) {
201 			struct got_pack_meta *n = NULL;
202 			n = meta[i - (max_base_candidates + 1)];
203 			got_deltify_free(n->dtab);
204 			n->dtab = NULL;
205 		}
206 
207 		best = raw->size;
208 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
209 			if (cancel_cb) {
210 				err = (*cancel_cb)(cancel_arg);
211 				if (err)
212 					goto done;
213 			}
214 			base = meta[j];
215 			/* long chains make unpacking slow, avoid such bases */
216 			if (base->nchain >= 128 ||
217 			    base->obj_type != m->obj_type)
218 				continue;
219 
220 			err = got_object_raw_open(&base_raw, repo, &base->id,
221 			    8192);
222 			if (err)
223 				goto done;
224 			err = got_deltify(&deltas, &ndeltas,
225 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
226 			    base->dtab, base_raw->f, base_raw->hdrlen,
227 			    base_raw->size + base_raw->hdrlen);
228 			got_object_raw_close(base_raw);
229 			base_raw = NULL;
230 			if (err)
231 				goto done;
232 
233 			size = delta_size(deltas, ndeltas);
234 			if (size + 32 < best){
235 				/*
236 				 * if we already picked a best delta,
237 				 * replace it.
238 				 */
239 				free(m->deltas);
240 				best = size;
241 				m->deltas = deltas;
242 				m->ndeltas = ndeltas;
243 				m->nchain = base->nchain + 1;
244 				m->prev = base;
245 				m->head = base->head;
246 				if (m->head == NULL)
247 					m->head = base;
248 			} else {
249 				free(deltas);
250 				deltas = NULL;
251 				ndeltas = 0;
252 			}
253 		}
254 
255 		got_object_raw_close(raw);
256 		raw = NULL;
257 	}
258 done:
259 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
260 		got_deltify_free(meta[i]->dtab);
261 		meta[i]->dtab = NULL;
262 	}
263 	if (raw)
264 		got_object_raw_close(raw);
265 	if (base_raw)
266 		got_object_raw_close(base_raw);
267 	return err;
268 }
269 
270 static const struct got_error *
271 search_packidx(int *found, struct got_object_id *id,
272     struct got_repository *repo)
273 {
274 	const struct got_error *err = NULL;
275 	struct got_packidx *packidx = NULL;
276 	int idx;
277 
278 	*found = 0;
279 
280 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
281 	if (err == NULL)
282 		*found = 1; /* object is already packed */
283 	else if (err->code == GOT_ERR_NO_OBJ)
284 		err = NULL;
285 	return err;
286 }
287 
288 static const int obj_types[] = {
289 	GOT_OBJ_TYPE_ANY,
290 	GOT_OBJ_TYPE_COMMIT,
291 	GOT_OBJ_TYPE_TREE,
292 	GOT_OBJ_TYPE_BLOB,
293 	GOT_OBJ_TYPE_TAG,
294 	GOT_OBJ_TYPE_OFFSET_DELTA,
295 	GOT_OBJ_TYPE_REF_DELTA
296 };
297 
298 static const struct got_error *
299 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
300     struct got_object_id *id, const char *path, int obj_type,
301     time_t mtime, int loose_obj_only, struct got_repository *repo)
302 {
303 	const struct got_error *err;
304 	struct got_pack_meta *m;
305 
306 	if (loose_obj_only) {
307 		int is_packed;
308 		err = search_packidx(&is_packed, id, repo);
309 		if (err)
310 			return err;
311 		if (is_packed)
312 			return NULL;
313 	}
314 
315 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
316 	if (err)
317 		return err;
318 
319 	if (v == NULL)
320 		return NULL;
321 
322 	err = alloc_meta(&m, id, path, obj_type, mtime);
323 	if (err)
324 		goto done;
325 
326 	if (v->nmeta == v->metasz){
327 		size_t newsize = 2 * v->metasz;
328 		struct got_pack_meta **new;
329 		new = reallocarray(v->meta, newsize, sizeof(*new));
330 		if (new == NULL) {
331 			err = got_error_from_errno("reallocarray");
332 			goto done;
333 		}
334 		v->meta = new;
335 		v->metasz = newsize; 
336 	}
337 done:
338 	if (err) {
339 		clear_meta(m);
340 		free(m);
341 	} else
342 		v->meta[v->nmeta++] = m;
343 
344 	return err;
345 }
346 
347 static const struct got_error *
348 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
349     struct got_object_idset *idset, struct got_object_id *tree_id,
350     const char *dpath, time_t mtime, struct got_repository *repo,
351     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
352 {
353 	const struct got_error *err;
354 	struct got_tree_object *tree;
355 	char *p = NULL;
356 	int i;
357 
358 	err = got_object_open_as_tree(&tree, repo, tree_id);
359 	if (err)
360 		return err;
361 
362 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
363 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
364 		struct got_object_id *id = got_tree_entry_get_id(e);
365 		mode_t mode = got_tree_entry_get_mode(e);
366 
367 		if (cancel_cb) {
368 			err = (*cancel_cb)(cancel_arg);
369 			if (err)
370 				break;
371 		}
372 
373 		if (got_object_tree_entry_is_symlink(e) ||
374 		    got_object_tree_entry_is_submodule(e) ||
375 		    got_object_idset_contains(idset, id))
376 			continue;
377 		
378 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
379 		    got_tree_entry_get_name(e)) == -1) {
380 			err = got_error_from_errno("asprintf");
381 			break;
382 		}
383 
384 		if (S_ISDIR(mode)) {
385 			struct got_object_qid *qid;
386 			err = got_object_qid_alloc(&qid, id);
387 			if (err)
388 				break;
389 			STAILQ_INSERT_TAIL(ids, qid, entry);
390 		} else if (S_ISREG(mode)) {
391 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
392 			    mtime, loose_obj_only, repo);
393 			if (err)
394 				break;
395 		}
396 		free(p);
397 		p = NULL;
398 	}
399 
400 	got_object_tree_close(tree);
401 	free(p);
402 	return err;
403 }
404 
405 static const struct got_error *
406 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
407     struct got_object_id *tree_id, const char *dpath, time_t mtime,
408     int loose_obj_only, struct got_repository *repo,
409     got_cancel_cb cancel_cb, void *cancel_arg)
410 {
411 	const struct got_error *err = NULL;
412 	struct got_object_id_queue tree_ids;
413 	struct got_object_qid *qid;
414 
415 	if (got_object_idset_contains(idset, tree_id))
416 		return NULL;
417 
418 	err = got_object_qid_alloc(&qid, tree_id);
419 	if (err)
420 		return err;
421 
422 	STAILQ_INIT(&tree_ids);
423 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
424 
425 	while (!STAILQ_EMPTY(&tree_ids)) {
426 		if (cancel_cb) {
427 			err = (*cancel_cb)(cancel_arg);
428 			if (err)
429 				break;
430 		}
431 
432 		qid = STAILQ_FIRST(&tree_ids);
433 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
434 
435 		if (got_object_idset_contains(idset, qid->id)) {
436 			got_object_qid_free(qid);
437 			continue;
438 		}
439 
440 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
441 		    mtime, loose_obj_only, repo);
442 		if (err) {
443 			got_object_qid_free(qid);
444 			break;
445 		}
446 
447 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
448 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
449 		got_object_qid_free(qid);
450 		if (err)
451 			break;
452 	}
453 
454 	got_object_id_queue_free(&tree_ids);
455 	return err;
456 }
457 
458 static const struct got_error *
459 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
460     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
461     got_cancel_cb cancel_cb, void *cancel_arg)
462 {
463 	const struct got_error *err;
464 	struct got_commit_object *commit;
465 
466 	if (got_object_idset_contains(idset, id))
467 		return NULL;
468 
469 	if (loose_obj_only) {
470 		int is_packed;
471 		err = search_packidx(&is_packed, id, repo);
472 		if (err)
473 			return err;
474 		if (is_packed)
475 			return NULL;
476 	}
477 
478 	err = got_object_open_as_commit(&commit, repo, id);
479 	if (err)
480 		return err;
481 
482 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
483 	    got_object_commit_get_committer_time(commit),
484 	    loose_obj_only, repo);
485 	if (err)
486 		goto done;
487 
488 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
489 	    "", got_object_commit_get_committer_time(commit),
490 	    loose_obj_only, repo, cancel_cb, cancel_arg);
491 done:
492 	got_object_commit_close(commit);
493 	return err;
494 }
495 
496 static const struct got_error *
497 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
498     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
499     got_cancel_cb cancel_cb, void *cancel_arg)
500 {
501 	const struct got_error *err;
502 	struct got_tag_object *tag = NULL;
503 
504 	if (got_object_idset_contains(idset, id))
505 		return NULL;
506 
507 	if (loose_obj_only) {
508 		int is_packed;
509 		err = search_packidx(&is_packed, id, repo);
510 		if (err)
511 			return err;
512 		if (is_packed)
513 			return NULL;
514 	}
515 
516 	err = got_object_open_as_tag(&tag, repo, id);
517 	if (err)
518 		return err;
519 
520 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
521 	    got_object_tag_get_tagger_time(tag),
522 	    loose_obj_only, repo);
523 	if (err)
524 		goto done;
525 
526 	switch (got_object_tag_get_object_type(tag)) {
527 	case GOT_OBJ_TYPE_COMMIT:
528 		err = load_commit(v, idset,
529 		    got_object_tag_get_object_id(tag), repo,
530 		    loose_obj_only, cancel_cb, cancel_arg);
531 		break;
532 	case GOT_OBJ_TYPE_TREE:
533 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
534 		    "", got_object_tag_get_tagger_time(tag),
535 		    loose_obj_only, repo, cancel_cb, cancel_arg);
536 		break;
537 	default:
538 		break;
539 	}
540 
541 done:
542 	got_object_tag_close(tag);
543 	return err;
544 }
545 
546 enum findtwixt_color {
547 	COLOR_KEEP = 0,
548 	COLOR_DROP,
549 	COLOR_BLANK,
550 };
551 static const int findtwixt_colors[] = {
552 	COLOR_KEEP,
553 	COLOR_DROP,
554 	COLOR_BLANK
555 };
556 
557 static const struct got_error *
558 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
559     int color, struct got_repository *repo)
560 {
561 	const struct got_error *err;
562 	struct got_object_qid *qid;
563 
564 	err = got_object_qid_alloc(&qid, id);
565 	if (err)
566 		return err;
567 
568 	STAILQ_INSERT_TAIL(ids, qid, entry);
569 	qid->data = (void *)&findtwixt_colors[color];
570 	return NULL;
571 }
572 
573 static const struct got_error *
574 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
575     struct got_object_id *id, struct got_repository *repo,
576     got_cancel_cb cancel_cb, void *cancel_arg)
577 {
578 	const struct got_error *err = NULL;
579 	struct got_commit_object *commit;
580 	const struct got_object_id_queue *parents;
581 	struct got_object_id_queue ids;
582 	struct got_object_qid *qid;
583 
584 	STAILQ_INIT(&ids);
585 
586 	err = got_object_qid_alloc(&qid, id);
587 	if (err)
588 		return err;
589 	STAILQ_INSERT_HEAD(&ids, qid, entry);
590 
591 	while (!STAILQ_EMPTY(&ids)) {
592 		if (cancel_cb) {
593 			err = (*cancel_cb)(cancel_arg);
594 			if (err)
595 				break;
596 		}
597 
598 		qid = STAILQ_FIRST(&ids);
599 		STAILQ_REMOVE_HEAD(&ids, entry);
600 
601 		if (got_object_idset_contains(drop, qid->id)) {
602 			got_object_qid_free(qid);
603 			continue;
604 		}
605 
606 		err = got_object_idset_add(drop, qid->id,
607 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
608 		if (err) {
609 			got_object_qid_free(qid);
610 			break;
611 		}
612 
613 		if (!got_object_idset_contains(keep, qid->id)) {
614 			got_object_qid_free(qid);
615 			continue;
616 		}
617 
618 		err = got_object_open_as_commit(&commit, repo, qid->id);
619 		got_object_qid_free(qid);
620 		if (err)
621 			break;
622 
623 		parents = got_object_commit_get_parent_ids(commit);
624 		if (parents) {
625 			err = got_object_id_queue_copy(parents, &ids);
626 			if (err) {
627 				got_object_commit_close(commit);
628 				break;
629 			}
630 		}
631 		got_object_commit_close(commit);
632 	}
633 
634 	got_object_id_queue_free(&ids);
635 	return err;
636 }
637 
638 struct append_id_arg {
639 	struct got_object_id **array;
640 	int idx;
641 };
642 
643 static const struct got_error *
644 append_id(struct got_object_id *id, void *data, void *arg)
645 {
646 	struct append_id_arg *a = arg;
647 
648 	a->array[a->idx] = got_object_id_dup(id);
649 	if (a->array[a->idx] == NULL)
650 		return got_error_from_errno("got_object_id_dup");
651 
652 	a->idx++;
653 	return NULL;
654 }
655 
656 static const struct got_error *
657 findtwixt(struct got_object_id ***res, int *nres,
658     struct got_object_id **head, int nhead,
659     struct got_object_id **tail, int ntail,
660     struct got_repository *repo,
661     got_cancel_cb cancel_cb, void *cancel_arg)
662 {
663 	const struct got_error *err = NULL;
664 	struct got_object_id_queue ids;
665 	struct got_object_idset *keep, *drop;
666 	struct got_object_qid *qid;
667 	int i, ncolor, nkeep, obj_type;
668 
669 	STAILQ_INIT(&ids);
670 	*res = NULL;
671 	*nres = 0;
672 
673 	keep = got_object_idset_alloc();
674 	if (keep == NULL)
675 		return got_error_from_errno("got_object_idset_alloc");
676 
677 	drop = got_object_idset_alloc();
678 	if (drop == NULL) {
679 		err = got_error_from_errno("got_object_idset_alloc");
680 		goto done;
681 	}
682 
683 	for (i = 0; i < nhead; i++) {
684 		struct got_object_id *id = head[i];
685 		if (id == NULL)
686 			continue;
687 		err = got_object_get_type(&obj_type, repo, id);
688 		if (err)
689 			return err;
690 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
691 			continue;
692 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
693 		if (err)
694 			goto done;
695 	}		
696 	for (i = 0; i < ntail; i++) {
697 		struct got_object_id *id = tail[i];
698 		if (id == NULL)
699 			continue;
700 		err = got_object_get_type(&obj_type, repo, id);
701 		if (err)
702 			return err;
703 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
704 			continue;
705 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
706 		if (err)
707 			goto done;
708 	}
709 
710 	while (!STAILQ_EMPTY(&ids)) {
711 		int qcolor;
712 		qid = STAILQ_FIRST(&ids);
713 		qcolor = *((int *)qid->data);
714 
715 		if (got_object_idset_contains(drop, qid->id))
716 			ncolor = COLOR_DROP;
717 		else if (got_object_idset_contains(keep, qid->id))
718 			ncolor = COLOR_KEEP;
719 		else
720 			ncolor = COLOR_BLANK;
721 
722 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
723 		    qcolor == COLOR_KEEP)) {
724 			STAILQ_REMOVE_HEAD(&ids, entry);
725 			got_object_qid_free(qid);
726 			continue;
727 		}
728 
729 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
730 			err = drop_commit(keep, drop, qid->id, repo,
731 			    cancel_cb, cancel_arg);
732 			if (err)
733 				goto done;
734 		} else if (ncolor == COLOR_BLANK) {
735 			struct got_commit_object *commit;
736 			struct got_object_id *id;
737 			const struct got_object_id_queue *parents;
738 			struct got_object_qid *pid;
739 
740 			id = got_object_id_dup(qid->id);
741 			if (id == NULL) {
742 				err = got_error_from_errno("got_object_id_dup");
743 				goto done;
744 			}
745 			if (qcolor == COLOR_KEEP)
746 				err = got_object_idset_add(keep, id,
747 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
748 			else
749 				err = got_object_idset_add(drop, id,
750 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
751 			if (err) {
752 				free(id);
753 				goto done;
754 			}
755 
756 			err = got_object_open_as_commit(&commit, repo, id);
757 			if (err) {
758 				free(id);
759 				goto done;
760 			}
761 			parents = got_object_commit_get_parent_ids(commit);
762 			if (parents) {
763 				STAILQ_FOREACH(pid, parents, entry) {
764 					err = queue_commit_id(&ids, pid->id,
765 					    qcolor, repo);
766 					if (err) {
767 						free(id);
768 						goto done;
769 					}
770 				}
771 			}
772 			got_object_commit_close(commit);
773 			commit = NULL;
774 		} else {
775 			/* should not happen */
776 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
777 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
778 			goto done;
779 		}
780 
781 		STAILQ_REMOVE_HEAD(&ids, entry);
782 		got_object_qid_free(qid);
783 	}
784 
785 	nkeep = got_object_idset_num_elements(keep);
786 	if (nkeep > 0) {
787 		struct append_id_arg arg;
788 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
789 		if (arg.array == NULL) {
790 			err = got_error_from_errno("calloc");
791 			goto done;
792 		}
793 		arg.idx = 0;
794 		err = got_object_idset_for_each(keep, append_id, &arg);
795 		if (err) {
796 			free(arg.array);
797 			goto done;
798 		}
799 		*res = arg.array;
800 		*nres = nkeep;
801 	}
802 done:
803 	got_object_idset_free(keep);
804 	got_object_idset_free(drop);
805 	got_object_id_queue_free(&ids);
806 	return err;
807 }
808 
809 static const struct got_error *
810 read_meta(struct got_pack_meta ***meta, int *nmeta,
811     struct got_object_id **theirs, int ntheirs,
812     struct got_object_id **ours, int nours, struct got_repository *repo,
813     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
814     got_cancel_cb cancel_cb, void *cancel_arg)
815 {
816 	const struct got_error *err = NULL;
817 	struct got_object_id **ids = NULL;
818 	struct got_object_idset *idset;
819 	int i, nobj = 0, obj_type;
820 	struct got_pack_metavec v;
821 
822 	*meta = NULL;
823 	*nmeta = 0;
824 
825 	idset = got_object_idset_alloc();
826 	if (idset == NULL)
827 		return got_error_from_errno("got_object_idset_alloc");
828 
829 	v.nmeta = 0;
830 	v.metasz = 64;
831 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
832 	if (v.meta == NULL) {
833 		err = got_error_from_errno("reallocarray");
834 		goto done;
835 	}
836 
837 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
838 	    cancel_cb, cancel_arg);
839 	if (err || nobj == 0)
840 		goto done;
841 
842 	for (i = 0; i < ntheirs; i++) {
843 		struct got_object_id *id = theirs[i];
844 		if (id == NULL)
845 			continue;
846 		err = got_object_get_type(&obj_type, repo, id);
847 		if (err)
848 			return err;
849 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
850 			continue;
851 		err = load_commit(NULL, idset, id, repo,
852 		    loose_obj_only, cancel_cb, cancel_arg);
853 		if (err)
854 			goto done;
855 		if (progress_cb) {
856 			err = progress_cb(progress_arg, 0L, nours,
857 			    v.nmeta, 0, 0);
858 			if (err)
859 				goto done;
860 		}
861 	}
862 
863 	for (i = 0; i < ntheirs; i++) {
864 		struct got_object_id *id = theirs[i];
865 		int *cached_type;
866 		if (id == NULL)
867 			continue;
868 		cached_type = got_object_idset_get(idset, id);
869 		if (cached_type == NULL) {
870 			err = got_object_get_type(&obj_type, repo, id);
871 			if (err)
872 				goto done;
873 		} else
874 			obj_type = *cached_type;
875 		if (obj_type != GOT_OBJ_TYPE_TAG)
876 			continue;
877 		err = load_tag(NULL, idset, id, repo,
878 		    loose_obj_only, cancel_cb, cancel_arg);
879 		if (err)
880 			goto done;
881 		if (progress_cb) {
882 			err = progress_cb(progress_arg, 0L, nours,
883 			    v.nmeta, 0, 0);
884 			if (err)
885 				goto done;
886 		}
887 	}
888 
889 	for (i = 0; i < nobj; i++) {
890 		err = load_commit(&v, idset, ids[i], repo,
891 		    loose_obj_only, cancel_cb, cancel_arg);
892 		if (err)
893 			goto done;
894 		if (progress_cb) {
895 			err = progress_cb(progress_arg, 0L, nours,
896 			    v.nmeta, 0, 0);
897 			if (err)
898 				goto done;
899 		}
900 	}
901 
902 	for (i = 0; i < nours; i++) {
903 		struct got_object_id *id = ours[i];
904 		int *cached_type;
905 		if (id == NULL)
906 			continue;
907 		cached_type = got_object_idset_get(idset, id);
908 		if (cached_type == NULL) {
909 			err = got_object_get_type(&obj_type, repo, id);
910 			if (err)
911 				goto done;
912 		} else
913 			obj_type = *cached_type;
914 		if (obj_type != GOT_OBJ_TYPE_TAG)
915 			continue;
916 		err = load_tag(&v, idset, id, repo,
917 		    loose_obj_only, cancel_cb, cancel_arg);
918 		if (err)
919 			goto done;
920 		if (progress_cb) {
921 			err = progress_cb(progress_arg, 0L, nours,
922 			    v.nmeta, 0, 0);
923 			if (err)
924 				goto done;
925 		}
926 	}
927 
928 done:
929 	for (i = 0; i < nobj; i++) {
930 		free(ids[i]);
931 	}
932 	free(ids);
933 	got_object_idset_free(idset);
934 	if (err == NULL) {
935 		*meta = v.meta;
936 		*nmeta = v.nmeta;
937 	} else
938 		free(v.meta);
939 
940 	return err;
941 }
942 
943 const struct got_error *
944 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
945 {
946 	size_t n;
947 
948 	SHA1Update(ctx, buf, len);
949 	n = fwrite(buf, 1, len, f);
950 	if (n != len)
951 		return got_ferror(f, GOT_ERR_IO);
952 	return NULL;
953 }
954 
955 static void
956 putbe32(char *b, uint32_t n)
957 {
958 	b[0] = n >> 24;
959 	b[1] = n >> 16;
960 	b[2] = n >> 8;
961 	b[3] = n >> 0;
962 }
963 
964 static int
965 write_order_cmp(const void *pa, const void *pb)
966 {
967 	struct got_pack_meta *a, *b, *ahd, *bhd;
968 
969 	a = *(struct got_pack_meta **)pa;
970 	b = *(struct got_pack_meta **)pb;
971 	ahd = (a->head == NULL) ? a : a->head;
972 	bhd = (b->head == NULL) ? b : b->head;
973 	if (ahd->mtime != bhd->mtime)
974 		return bhd->mtime - ahd->mtime;
975 	if (ahd != bhd)
976 		return (uintptr_t)bhd - (uintptr_t)ahd;
977 	if (a->nchain != b->nchain)
978 		return a->nchain - b->nchain;
979 	return a->mtime - b->mtime;
980 }
981 
982 static const struct got_error *
983 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
984 {
985 	size_t i;
986 
987 	*hdrlen = 0;
988 
989 	hdr[0] = obj_type << 4;
990 	hdr[0] |= len & 0xf;
991 	len >>= 4;
992 	for (i = 1; len != 0; i++){
993 		if (i >= bufsize)
994 			return got_error(GOT_ERR_NO_SPACE);
995 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
996 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
997 		len >>= GOT_DELTA_SIZE_SHIFT;
998 	}
999 
1000 	*hdrlen = i;
1001 	return NULL;
1002 }
1003 
1004 static const struct got_error *
1005 append(char **p, int *len, int *sz, void *seg, int nseg)
1006 {
1007 	char *n;
1008 
1009 	if (*len + nseg >= *sz) {
1010 		while (*len + nseg >= *sz)
1011 			*sz += *sz / 2;
1012 		n = realloc(*p, *sz);
1013 		if (n == NULL)
1014 			return got_error_from_errno("realloc");
1015 		*p = n;
1016 	}
1017 	memcpy(*p + *len, seg, nseg);
1018 	*len += nseg;
1019 	return NULL;
1020 }
1021 
1022 
1023 static const struct got_error *
1024 encodedelta(int *nd, struct got_pack_meta *m, struct got_raw_object *o,
1025     off_t base_size, char **pp)
1026 {
1027 	const struct got_error *err = NULL;
1028 	char *p;
1029 	unsigned char buf[16], *bp;
1030 	int len, sz, i, j;
1031 	off_t n;
1032 	struct got_delta_instruction *d;
1033 
1034 	*pp = NULL;
1035 	*nd = 0;
1036 
1037 	sz = 128;
1038 	len = 0;
1039 	p = malloc(sz);
1040 	if (p == NULL)
1041 		return got_error_from_errno("malloc");
1042 
1043 	/* base object size */
1044 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1045 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1046 	for (i = 1; n > 0; i++) {
1047 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1048 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1049 		n >>= GOT_DELTA_SIZE_SHIFT;
1050 	}
1051 	err = append(&p, &len, &sz, buf, i);
1052 	if (err)
1053 		return err;
1054 
1055 	/* target object size */
1056 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1057 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1058 	for (i = 1; n > 0; i++) {
1059 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1060 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1061 		n >>= GOT_DELTA_SIZE_SHIFT;
1062 	}
1063 	err = append(&p, &len, &sz, buf, i);
1064 	if (err)
1065 		return err;
1066 	for (j = 0; j < m->ndeltas; j++) {
1067 		d = &m->deltas[j];
1068 		if (d->copy) {
1069 			n = d->offset;
1070 			bp = &buf[1];
1071 			buf[0] = GOT_DELTA_BASE_COPY;
1072 			for (i = 0; i < 4; i++) {
1073 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1074 				buf[0] |= 1 << i;
1075 				*bp++ = n & 0xff;
1076 				n >>= 8;
1077 				if (n == 0)
1078 					break;
1079 			}
1080 
1081 			n = d->len;
1082 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1083 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1084 				for (i = 0; i < 3 && n > 0; i++) {
1085 					buf[0] |= 1 << (i + 4);
1086 					*bp++ = n & 0xff;
1087 					n >>= 8;
1088 				}
1089 			}
1090 			err = append(&p, &len, &sz, buf, bp - buf);
1091 			if (err)
1092 				return err;
1093 		} else {
1094 			char content[128];
1095 			size_t r;
1096 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1097 				return got_error_from_errno("fseeko");
1098 			n = 0;
1099 			while (n != d->len) {
1100 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1101 				err = append(&p, &len, &sz, buf, 1);
1102 				if (err)
1103 					return err;
1104 				r = fread(content, 1, buf[0], o->f);
1105 				if (r != buf[0])
1106 					return got_ferror(o->f, GOT_ERR_IO);
1107 				err = append(&p, &len, &sz, content, buf[0]);
1108 				if (err)
1109 					return err;
1110 				n += buf[0];
1111 			}
1112 		}
1113 	}
1114 	*pp = p;
1115 	*nd = len;
1116 	return NULL;
1117 }
1118 
1119 static int
1120 packoff(char *hdr, off_t off)
1121 {
1122 	int i, j;
1123 	char rbuf[8];
1124 
1125 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1126 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1127 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1128 		    GOT_DELTA_SIZE_MORE;
1129 	}
1130 
1131 	j = 0;
1132 	while (i > 0)
1133 		hdr[j++] = rbuf[--i];
1134 	return j;
1135 }
1136 
1137 static const struct got_error *
1138 genpack(uint8_t *pack_sha1, FILE *packfile,
1139     struct got_pack_meta **meta, int nmeta, int nours,
1140     int use_offset_deltas, struct got_repository *repo,
1141     got_pack_progress_cb progress_cb, void *progress_arg,
1142     got_cancel_cb cancel_cb, void *cancel_arg)
1143 {
1144 	const struct got_error *err = NULL;
1145 	int i, nh, nd;
1146 	SHA1_CTX ctx;
1147 	struct got_pack_meta *m;
1148 	struct got_raw_object *raw;
1149 	char *p = NULL, buf[32];
1150 	size_t outlen, n;
1151 	struct got_deflate_checksum csum;
1152 	off_t packfile_size = 0;
1153 
1154 	SHA1Init(&ctx);
1155 	csum.output_sha1 = &ctx;
1156 	csum.output_crc = NULL;
1157 
1158 	err = hwrite(packfile, "PACK", 4, &ctx);
1159 	if (err)
1160 		return err;
1161 	putbe32(buf, GOT_PACKFILE_VERSION);
1162 	err = hwrite(packfile, buf, 4, &ctx);
1163 	if (err)
1164 		goto done;
1165 	putbe32(buf, nmeta);
1166 	err = hwrite(packfile, buf, 4, &ctx);
1167 	if (err)
1168 		goto done;
1169 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1170 	for (i = 0; i < nmeta; i++) {
1171 		if (progress_cb) {
1172 			err = progress_cb(progress_arg, packfile_size, nours,
1173 			    nmeta, nmeta, i);
1174 			if (err)
1175 				goto done;
1176 		}
1177 		m = meta[i];
1178 		m->off = ftello(packfile);
1179 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
1180 		if (err)
1181 			goto done;
1182 		if (m->deltas == NULL) {
1183 			err = packhdr(&nh, buf, sizeof(buf),
1184 			    m->obj_type, raw->size);
1185 			if (err)
1186 				goto done;
1187 			err = hwrite(packfile, buf, nh, &ctx);
1188 			if (err)
1189 				goto done;
1190 			packfile_size += nh;
1191 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1192 				err = got_error_from_errno("fseeko");
1193 				goto done;
1194 			}
1195 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1196 			    &csum);
1197 			if (err)
1198 				goto done;
1199 			packfile_size += outlen;
1200 		} else {
1201 			FILE *delta_file;
1202 			struct got_raw_object *base_raw;
1203 			err = got_object_raw_open(&base_raw, repo,
1204 			    &m->prev->id, 8192);
1205 			if (err)
1206 				goto done;
1207 			err = encodedelta(&nd, m, raw, base_raw->size, &p);
1208 			if (err)
1209 				goto done;
1210 			got_object_raw_close(base_raw);
1211 			if (use_offset_deltas && m->prev->off != 0) {
1212 				err = packhdr(&nh, buf, sizeof(buf),
1213 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1214 				if (err)
1215 					goto done;
1216 				nh += packoff(buf + nh,
1217 				    m->off - m->prev->off);
1218 				err = hwrite(packfile, buf, nh, &ctx);
1219 				if (err)
1220 					goto done;
1221 				packfile_size += nh;
1222 			} else {
1223 				err = packhdr(&nh, buf, sizeof(buf),
1224 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1225 				err = hwrite(packfile, buf, nh, &ctx);
1226 				if (err)
1227 					goto done;
1228 				packfile_size += nh;
1229 				err = hwrite(packfile, m->prev->id.sha1,
1230 				    sizeof(m->prev->id.sha1), &ctx);
1231 				packfile_size += sizeof(m->prev->id.sha1);
1232 				if (err)
1233 					goto done;
1234 			}
1235 			/* XXX need got_deflate_from_mem() */
1236 			delta_file = fmemopen(p, nd, "r");
1237 			if (delta_file == NULL) {
1238 				err = got_error_from_errno("fmemopen");
1239 				goto done;
1240 			}
1241 			err = got_deflate_to_file(&outlen, delta_file,
1242 			    packfile, &csum);
1243 			fclose(delta_file);
1244 			if (err)
1245 				goto done;
1246 			packfile_size += outlen;
1247 			free(p);
1248 			p = NULL;
1249 		}
1250 		got_object_raw_close(raw);
1251 		raw = NULL;
1252 	}
1253 	SHA1Final(pack_sha1, &ctx);
1254 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1255 	if (n != SHA1_DIGEST_LENGTH)
1256 		err = got_ferror(packfile, GOT_ERR_IO);
1257 	packfile_size += SHA1_DIGEST_LENGTH;
1258 	packfile_size += sizeof(struct got_packfile_hdr);
1259 	err = progress_cb(progress_arg, packfile_size, nours,
1260 	    nmeta, nmeta, nmeta);
1261 	if (err)
1262 		goto done;
1263 done:
1264 	free(p);
1265 	return err;
1266 }
1267 
1268 const struct got_error *
1269 got_pack_create(uint8_t *packsha1, FILE *packfile,
1270     struct got_object_id **theirs, int ntheirs,
1271     struct got_object_id **ours, int nours,
1272     struct got_repository *repo, int loose_obj_only, int allow_empty,
1273     got_pack_progress_cb progress_cb, void *progress_arg,
1274     got_cancel_cb cancel_cb, void *cancel_arg)
1275 {
1276 	const struct got_error *err;
1277 	struct got_pack_meta **meta;
1278 	int nmeta;
1279 
1280 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1281 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1282 	if (err)
1283 		return err;
1284 
1285 	if (nmeta == 0 && !allow_empty) {
1286 		err = got_error(GOT_ERR_CANNOT_PACK);
1287 		goto done;
1288 	}
1289 	if (nmeta > 0) {
1290 		err = pick_deltas(meta, nmeta, nours, repo,
1291 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1292 		if (err)
1293 			goto done;
1294 	}
1295 
1296 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1297 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1298 	if (err)
1299 		goto done;
1300 done:
1301 	free_nmeta(meta, nmeta);
1302 	return err;
1303 }