Gotweb

Blob

Date:: Fri Oct 15 07:15:00 2021 UTC
Message:: reuse temporary files which were not used by got_object_raw_open()
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/tree.h>
21 #include <sys/uio.h>
22 #include <sys/stat.h>
23 
24 #include <stdint.h>
25 #include <imsg.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <sha1.h>
30 #include <limits.h>
31 #include <zlib.h>
32 
33 #include "got_error.h"
34 #include "got_cancel.h"
35 #include "got_object.h"
36 #include "got_path.h"
37 #include "got_reference.h"
38 #include "got_repository_admin.h"
39 #include "got_opentemp.h"
40 
41 #include "got_lib_deltify.h"
42 #include "got_lib_delta.h"
43 #include "got_lib_object.h"
44 #include "got_lib_object_idset.h"
45 #include "got_lib_object_cache.h"
46 #include "got_lib_deflate.h"
47 #include "got_lib_pack.h"
48 #include "got_lib_privsep.h"
49 #include "got_lib_repository.h"
50 
51 #ifndef MAX
52 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
53 #endif
54 
55 struct got_pack_meta {
56 	struct got_object_id id;
57 	char	*path;
58 	int	obj_type;
59 	off_t	size;
60 	time_t	mtime;
61 
62 	/* The best delta we picked */
63 	struct got_pack_meta *head;
64 	struct got_pack_meta *prev;
65 	struct got_delta_instruction *deltas;
66 	int	ndeltas;
67 	int	nchain;
68 
69 	/* Only used for delta window */
70 	struct got_delta_table *dtab;
71 
72 	/* Only used for writing offset deltas */
73 	off_t	off;
74 };
75 
76 struct got_pack_metavec {
77 	struct got_pack_meta **meta;
78 	int nmeta;
79 	int metasz;
80 };
81 
82 static const struct got_error *
83 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
84     const char *path, int obj_type, time_t mtime)
85 {
86 	const struct got_error *err = NULL;
87 	struct got_pack_meta *m;
88 
89 	*new = NULL;
90 
91 	m = calloc(1, sizeof(*m));
92 	if (m == NULL)
93 		return got_error_from_errno("calloc");
94 
95 	memcpy(&m->id, id, sizeof(m->id));
96 
97 	m->path = strdup(path);
98 	if (m->path == NULL) {
99 		err = got_error_from_errno("strdup");
100 		free(m);
101 		return err;
102 	}
103 
104 	m->obj_type = obj_type;
105 	m->mtime = mtime;
106 	*new = m;
107 	return NULL;
108 }
109 
110 static void
111 clear_meta(struct got_pack_meta *meta)
112 {
113 	if (meta == NULL)
114 		return;
115 	free(meta->deltas);
116 	meta->deltas = NULL;
117 	free(meta->path);
118 	meta->path = NULL;
119 }
120 
121 static void
122 free_nmeta(struct got_pack_meta **meta, int nmeta)
123 {
124 	int i;
125 
126 	for (i = 0; i < nmeta; i++)
127 		clear_meta(meta[i]);
128 	free(meta);
129 }
130 
131 static int
132 delta_order_cmp(const void *pa, const void *pb)
133 {
134 	struct got_pack_meta *a, *b;
135 	int cmp;
136 
137 	a = *(struct got_pack_meta **)pa;
138 	b = *(struct got_pack_meta **)pb;
139 
140 	if (a->obj_type != b->obj_type)
141 		return a->obj_type - b->obj_type;
142 	cmp = strcmp(a->path, b->path);
143 	if (cmp != 0)
144 		return cmp;
145 	if (a->mtime != b->mtime)
146 		return a->mtime - b->mtime;
147 	return got_object_id_cmp(&a->id, &b->id);
148 }
149 
150 static int
151 delta_size(struct got_delta_instruction *deltas, int ndeltas)
152 {
153 	int i, size = 32;
154 	for (i = 0; i < ndeltas; i++) {
155 		if (deltas[i].copy)
156 			size += GOT_DELTA_SIZE_SHIFT;
157 		else
158 			size += deltas[i].len + 1;
159 	}
160 	return size;
161 }
162 
163 
164 static const struct got_error *
165 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
166     struct got_repository *repo,
167     got_pack_progress_cb progress_cb, void *progress_arg,
168     got_cancel_cb cancel_cb, void *cancel_arg)
169 {
170 	const struct got_error *err = NULL;
171 	struct got_pack_meta *m = NULL, *base = NULL;
172 	struct got_raw_object *raw = NULL, *base_raw = NULL;
173 	struct got_delta_instruction *deltas;
174 	int i, j, size, ndeltas, best;
175 	const int max_base_candidates = 10;
176 	int outfd = -1;
177 
178 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
179 	for (i = 0; i < nmeta; i++) {
180 		if (cancel_cb) {
181 			err = (*cancel_cb)(cancel_arg);
182 			if (err)
183 				break;
184 		}
185 		if (progress_cb) {
186 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
187 			if (err)
188 				goto done;
189 		}
190 		m = meta[i];
191 		m->deltas = NULL;
192 		m->ndeltas = 0;
193 
194 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
195 		    m->obj_type == GOT_OBJ_TYPE_TAG)
196 			continue;
197 
198 		if (outfd == -1) {
199 			outfd = got_opentempfd();
200 			if (outfd == -1) {
201 				err = got_error_from_errno("got_opentempfd");
202 				goto done;
203 			}
204 		}
205 		err = got_object_raw_open(&raw, outfd, repo, &m->id, 8192);
206 		if (err)
207 			goto done;
208 		if (raw->data == NULL)
209 			outfd = -1; /* outfd is now raw->f */
210 		m->size = raw->size;
211 
212 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
213 		    raw->size + raw->hdrlen);
214 		if (err)
215 			goto done;
216 
217 		if (i > max_base_candidates) {
218 			struct got_pack_meta *n = NULL;
219 			n = meta[i - (max_base_candidates + 1)];
220 			got_deltify_free(n->dtab);
221 			n->dtab = NULL;
222 		}
223 
224 		best = raw->size;
225 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
226 			if (cancel_cb) {
227 				err = (*cancel_cb)(cancel_arg);
228 				if (err)
229 					goto done;
230 			}
231 			base = meta[j];
232 			/* long chains make unpacking slow, avoid such bases */
233 			if (base->nchain >= 128 ||
234 			    base->obj_type != m->obj_type)
235 				continue;
236 
237 			if (outfd == -1) {
238 				outfd = got_opentempfd();
239 				if (outfd == -1) {
240 					err = got_error_from_errno(
241 					    "got_opentempfd");
242 					goto done;
243 				}
244 			}
245 			err = got_object_raw_open(&base_raw, outfd, repo,
246 			    &base->id, 8192);
247 			if (err)
248 				goto done;
249 			if (base_raw->data == NULL)
250 				outfd = -1; /* outfd is now base_raw->f */
251 			err = got_deltify(&deltas, &ndeltas,
252 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
253 			    base->dtab, base_raw->f, base_raw->hdrlen,
254 			    base_raw->size + base_raw->hdrlen);
255 			got_object_raw_close(base_raw);
256 			base_raw = NULL;
257 			if (err)
258 				goto done;
259 
260 			size = delta_size(deltas, ndeltas);
261 			if (size + 32 < best){
262 				/*
263 				 * if we already picked a best delta,
264 				 * replace it.
265 				 */
266 				free(m->deltas);
267 				best = size;
268 				m->deltas = deltas;
269 				m->ndeltas = ndeltas;
270 				m->nchain = base->nchain + 1;
271 				m->prev = base;
272 				m->head = base->head;
273 				if (m->head == NULL)
274 					m->head = base;
275 			} else {
276 				free(deltas);
277 				deltas = NULL;
278 				ndeltas = 0;
279 			}
280 		}
281 
282 		got_object_raw_close(raw);
283 		raw = NULL;
284 	}
285 done:
286 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
287 		got_deltify_free(meta[i]->dtab);
288 		meta[i]->dtab = NULL;
289 	}
290 	if (raw)
291 		got_object_raw_close(raw);
292 	if (base_raw)
293 		got_object_raw_close(base_raw);
294 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
295 		err = got_error_from_errno("close");
296 	return err;
297 }
298 
299 static const struct got_error *
300 search_packidx(int *found, struct got_object_id *id,
301     struct got_repository *repo)
302 {
303 	const struct got_error *err = NULL;
304 	struct got_packidx *packidx = NULL;
305 	int idx;
306 
307 	*found = 0;
308 
309 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
310 	if (err == NULL)
311 		*found = 1; /* object is already packed */
312 	else if (err->code == GOT_ERR_NO_OBJ)
313 		err = NULL;
314 	return err;
315 }
316 
317 static const int obj_types[] = {
318 	GOT_OBJ_TYPE_ANY,
319 	GOT_OBJ_TYPE_COMMIT,
320 	GOT_OBJ_TYPE_TREE,
321 	GOT_OBJ_TYPE_BLOB,
322 	GOT_OBJ_TYPE_TAG,
323 	GOT_OBJ_TYPE_OFFSET_DELTA,
324 	GOT_OBJ_TYPE_REF_DELTA
325 };
326 
327 static const struct got_error *
328 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
329     struct got_object_id *id, const char *path, int obj_type,
330     time_t mtime, int loose_obj_only, struct got_repository *repo)
331 {
332 	const struct got_error *err;
333 	struct got_pack_meta *m;
334 
335 	if (loose_obj_only) {
336 		int is_packed;
337 		err = search_packidx(&is_packed, id, repo);
338 		if (err)
339 			return err;
340 		if (is_packed)
341 			return NULL;
342 	}
343 
344 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
345 	if (err)
346 		return err;
347 
348 	if (v == NULL)
349 		return NULL;
350 
351 	err = alloc_meta(&m, id, path, obj_type, mtime);
352 	if (err)
353 		goto done;
354 
355 	if (v->nmeta == v->metasz){
356 		size_t newsize = 2 * v->metasz;
357 		struct got_pack_meta **new;
358 		new = reallocarray(v->meta, newsize, sizeof(*new));
359 		if (new == NULL) {
360 			err = got_error_from_errno("reallocarray");
361 			goto done;
362 		}
363 		v->meta = new;
364 		v->metasz = newsize; 
365 	}
366 done:
367 	if (err) {
368 		clear_meta(m);
369 		free(m);
370 	} else
371 		v->meta[v->nmeta++] = m;
372 
373 	return err;
374 }
375 
376 static const struct got_error *
377 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
378     struct got_object_idset *idset, struct got_object_id *tree_id,
379     const char *dpath, time_t mtime, struct got_repository *repo,
380     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
381 {
382 	const struct got_error *err;
383 	struct got_tree_object *tree;
384 	char *p = NULL;
385 	int i;
386 
387 	err = got_object_open_as_tree(&tree, repo, tree_id);
388 	if (err)
389 		return err;
390 
391 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
392 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
393 		struct got_object_id *id = got_tree_entry_get_id(e);
394 		mode_t mode = got_tree_entry_get_mode(e);
395 
396 		if (cancel_cb) {
397 			err = (*cancel_cb)(cancel_arg);
398 			if (err)
399 				break;
400 		}
401 
402 		if (got_object_tree_entry_is_submodule(e) ||
403 		    got_object_idset_contains(idset, id))
404 			continue;
405 		
406 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
407 		    got_tree_entry_get_name(e)) == -1) {
408 			err = got_error_from_errno("asprintf");
409 			break;
410 		}
411 
412 		if (S_ISDIR(mode)) {
413 			struct got_object_qid *qid;
414 			err = got_object_qid_alloc(&qid, id);
415 			if (err)
416 				break;
417 			STAILQ_INSERT_TAIL(ids, qid, entry);
418 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
419 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
420 			    mtime, loose_obj_only, repo);
421 			if (err)
422 				break;
423 		}
424 		free(p);
425 		p = NULL;
426 	}
427 
428 	got_object_tree_close(tree);
429 	free(p);
430 	return err;
431 }
432 
433 static const struct got_error *
434 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
435     struct got_object_id *tree_id, const char *dpath, time_t mtime,
436     int loose_obj_only, struct got_repository *repo,
437     got_cancel_cb cancel_cb, void *cancel_arg)
438 {
439 	const struct got_error *err = NULL;
440 	struct got_object_id_queue tree_ids;
441 	struct got_object_qid *qid;
442 
443 	if (got_object_idset_contains(idset, tree_id))
444 		return NULL;
445 
446 	err = got_object_qid_alloc(&qid, tree_id);
447 	if (err)
448 		return err;
449 
450 	STAILQ_INIT(&tree_ids);
451 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
452 
453 	while (!STAILQ_EMPTY(&tree_ids)) {
454 		if (cancel_cb) {
455 			err = (*cancel_cb)(cancel_arg);
456 			if (err)
457 				break;
458 		}
459 
460 		qid = STAILQ_FIRST(&tree_ids);
461 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
462 
463 		if (got_object_idset_contains(idset, qid->id)) {
464 			got_object_qid_free(qid);
465 			continue;
466 		}
467 
468 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
469 		    mtime, loose_obj_only, repo);
470 		if (err) {
471 			got_object_qid_free(qid);
472 			break;
473 		}
474 
475 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
476 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
477 		got_object_qid_free(qid);
478 		if (err)
479 			break;
480 	}
481 
482 	got_object_id_queue_free(&tree_ids);
483 	return err;
484 }
485 
486 static const struct got_error *
487 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
488     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
489     got_cancel_cb cancel_cb, void *cancel_arg)
490 {
491 	const struct got_error *err;
492 	struct got_commit_object *commit;
493 
494 	if (got_object_idset_contains(idset, id))
495 		return NULL;
496 
497 	if (loose_obj_only) {
498 		int is_packed;
499 		err = search_packidx(&is_packed, id, repo);
500 		if (err)
501 			return err;
502 		if (is_packed)
503 			return NULL;
504 	}
505 
506 	err = got_object_open_as_commit(&commit, repo, id);
507 	if (err)
508 		return err;
509 
510 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
511 	    got_object_commit_get_committer_time(commit),
512 	    loose_obj_only, repo);
513 	if (err)
514 		goto done;
515 
516 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
517 	    "", got_object_commit_get_committer_time(commit),
518 	    loose_obj_only, repo, cancel_cb, cancel_arg);
519 done:
520 	got_object_commit_close(commit);
521 	return err;
522 }
523 
524 static const struct got_error *
525 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
526     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
527     got_cancel_cb cancel_cb, void *cancel_arg)
528 {
529 	const struct got_error *err;
530 	struct got_tag_object *tag = NULL;
531 
532 	if (got_object_idset_contains(idset, id))
533 		return NULL;
534 
535 	if (loose_obj_only) {
536 		int is_packed;
537 		err = search_packidx(&is_packed, id, repo);
538 		if (err)
539 			return err;
540 		if (is_packed)
541 			return NULL;
542 	}
543 
544 	err = got_object_open_as_tag(&tag, repo, id);
545 	if (err)
546 		return err;
547 
548 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
549 	    got_object_tag_get_tagger_time(tag),
550 	    loose_obj_only, repo);
551 	if (err)
552 		goto done;
553 
554 	switch (got_object_tag_get_object_type(tag)) {
555 	case GOT_OBJ_TYPE_COMMIT:
556 		err = load_commit(v, idset,
557 		    got_object_tag_get_object_id(tag), repo,
558 		    loose_obj_only, cancel_cb, cancel_arg);
559 		break;
560 	case GOT_OBJ_TYPE_TREE:
561 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
562 		    "", got_object_tag_get_tagger_time(tag),
563 		    loose_obj_only, repo, cancel_cb, cancel_arg);
564 		break;
565 	default:
566 		break;
567 	}
568 
569 done:
570 	got_object_tag_close(tag);
571 	return err;
572 }
573 
574 enum findtwixt_color {
575 	COLOR_KEEP = 0,
576 	COLOR_DROP,
577 	COLOR_BLANK,
578 };
579 static const int findtwixt_colors[] = {
580 	COLOR_KEEP,
581 	COLOR_DROP,
582 	COLOR_BLANK
583 };
584 
585 static const struct got_error *
586 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
587     int color, struct got_repository *repo)
588 {
589 	const struct got_error *err;
590 	struct got_object_qid *qid;
591 
592 	err = got_object_qid_alloc(&qid, id);
593 	if (err)
594 		return err;
595 
596 	STAILQ_INSERT_TAIL(ids, qid, entry);
597 	qid->data = (void *)&findtwixt_colors[color];
598 	return NULL;
599 }
600 
601 static const struct got_error *
602 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
603     struct got_object_id *id, struct got_repository *repo,
604     got_cancel_cb cancel_cb, void *cancel_arg)
605 {
606 	const struct got_error *err = NULL;
607 	struct got_commit_object *commit;
608 	const struct got_object_id_queue *parents;
609 	struct got_object_id_queue ids;
610 	struct got_object_qid *qid;
611 
612 	STAILQ_INIT(&ids);
613 
614 	err = got_object_qid_alloc(&qid, id);
615 	if (err)
616 		return err;
617 	STAILQ_INSERT_HEAD(&ids, qid, entry);
618 
619 	while (!STAILQ_EMPTY(&ids)) {
620 		if (cancel_cb) {
621 			err = (*cancel_cb)(cancel_arg);
622 			if (err)
623 				break;
624 		}
625 
626 		qid = STAILQ_FIRST(&ids);
627 		STAILQ_REMOVE_HEAD(&ids, entry);
628 
629 		if (got_object_idset_contains(drop, qid->id)) {
630 			got_object_qid_free(qid);
631 			continue;
632 		}
633 
634 		err = got_object_idset_add(drop, qid->id,
635 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
636 		if (err) {
637 			got_object_qid_free(qid);
638 			break;
639 		}
640 
641 		if (!got_object_idset_contains(keep, qid->id)) {
642 			got_object_qid_free(qid);
643 			continue;
644 		}
645 
646 		err = got_object_open_as_commit(&commit, repo, qid->id);
647 		got_object_qid_free(qid);
648 		if (err)
649 			break;
650 
651 		parents = got_object_commit_get_parent_ids(commit);
652 		if (parents) {
653 			err = got_object_id_queue_copy(parents, &ids);
654 			if (err) {
655 				got_object_commit_close(commit);
656 				break;
657 			}
658 		}
659 		got_object_commit_close(commit);
660 	}
661 
662 	got_object_id_queue_free(&ids);
663 	return err;
664 }
665 
666 struct append_id_arg {
667 	struct got_object_id **array;
668 	int idx;
669 };
670 
671 static const struct got_error *
672 append_id(struct got_object_id *id, void *data, void *arg)
673 {
674 	struct append_id_arg *a = arg;
675 
676 	a->array[a->idx] = got_object_id_dup(id);
677 	if (a->array[a->idx] == NULL)
678 		return got_error_from_errno("got_object_id_dup");
679 
680 	a->idx++;
681 	return NULL;
682 }
683 
684 static const struct got_error *
685 findtwixt(struct got_object_id ***res, int *nres,
686     struct got_object_id **head, int nhead,
687     struct got_object_id **tail, int ntail,
688     struct got_repository *repo,
689     got_cancel_cb cancel_cb, void *cancel_arg)
690 {
691 	const struct got_error *err = NULL;
692 	struct got_object_id_queue ids;
693 	struct got_object_idset *keep, *drop;
694 	struct got_object_qid *qid;
695 	int i, ncolor, nkeep, obj_type;
696 
697 	STAILQ_INIT(&ids);
698 	*res = NULL;
699 	*nres = 0;
700 
701 	keep = got_object_idset_alloc();
702 	if (keep == NULL)
703 		return got_error_from_errno("got_object_idset_alloc");
704 
705 	drop = got_object_idset_alloc();
706 	if (drop == NULL) {
707 		err = got_error_from_errno("got_object_idset_alloc");
708 		goto done;
709 	}
710 
711 	for (i = 0; i < nhead; i++) {
712 		struct got_object_id *id = head[i];
713 		if (id == NULL)
714 			continue;
715 		err = got_object_get_type(&obj_type, repo, id);
716 		if (err)
717 			return err;
718 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
719 			continue;
720 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
721 		if (err)
722 			goto done;
723 	}		
724 	for (i = 0; i < ntail; i++) {
725 		struct got_object_id *id = tail[i];
726 		if (id == NULL)
727 			continue;
728 		err = got_object_get_type(&obj_type, repo, id);
729 		if (err)
730 			return err;
731 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
732 			continue;
733 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
734 		if (err)
735 			goto done;
736 	}
737 
738 	while (!STAILQ_EMPTY(&ids)) {
739 		int qcolor;
740 		qid = STAILQ_FIRST(&ids);
741 		qcolor = *((int *)qid->data);
742 
743 		if (got_object_idset_contains(drop, qid->id))
744 			ncolor = COLOR_DROP;
745 		else if (got_object_idset_contains(keep, qid->id))
746 			ncolor = COLOR_KEEP;
747 		else
748 			ncolor = COLOR_BLANK;
749 
750 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
751 		    qcolor == COLOR_KEEP)) {
752 			STAILQ_REMOVE_HEAD(&ids, entry);
753 			got_object_qid_free(qid);
754 			continue;
755 		}
756 
757 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
758 			err = drop_commit(keep, drop, qid->id, repo,
759 			    cancel_cb, cancel_arg);
760 			if (err)
761 				goto done;
762 		} else if (ncolor == COLOR_BLANK) {
763 			struct got_commit_object *commit;
764 			struct got_object_id *id;
765 			const struct got_object_id_queue *parents;
766 			struct got_object_qid *pid;
767 
768 			id = got_object_id_dup(qid->id);
769 			if (id == NULL) {
770 				err = got_error_from_errno("got_object_id_dup");
771 				goto done;
772 			}
773 			if (qcolor == COLOR_KEEP)
774 				err = got_object_idset_add(keep, id,
775 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
776 			else
777 				err = got_object_idset_add(drop, id,
778 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
779 			if (err) {
780 				free(id);
781 				goto done;
782 			}
783 
784 			err = got_object_open_as_commit(&commit, repo, id);
785 			if (err) {
786 				free(id);
787 				goto done;
788 			}
789 			parents = got_object_commit_get_parent_ids(commit);
790 			if (parents) {
791 				STAILQ_FOREACH(pid, parents, entry) {
792 					err = queue_commit_id(&ids, pid->id,
793 					    qcolor, repo);
794 					if (err) {
795 						free(id);
796 						goto done;
797 					}
798 				}
799 			}
800 			got_object_commit_close(commit);
801 			commit = NULL;
802 		} else {
803 			/* should not happen */
804 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
805 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
806 			goto done;
807 		}
808 
809 		STAILQ_REMOVE_HEAD(&ids, entry);
810 		got_object_qid_free(qid);
811 	}
812 
813 	nkeep = got_object_idset_num_elements(keep);
814 	if (nkeep > 0) {
815 		struct append_id_arg arg;
816 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
817 		if (arg.array == NULL) {
818 			err = got_error_from_errno("calloc");
819 			goto done;
820 		}
821 		arg.idx = 0;
822 		err = got_object_idset_for_each(keep, append_id, &arg);
823 		if (err) {
824 			free(arg.array);
825 			goto done;
826 		}
827 		*res = arg.array;
828 		*nres = nkeep;
829 	}
830 done:
831 	got_object_idset_free(keep);
832 	got_object_idset_free(drop);
833 	got_object_id_queue_free(&ids);
834 	return err;
835 }
836 
837 static const struct got_error *
838 read_meta(struct got_pack_meta ***meta, int *nmeta,
839     struct got_object_id **theirs, int ntheirs,
840     struct got_object_id **ours, int nours, struct got_repository *repo,
841     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
842     got_cancel_cb cancel_cb, void *cancel_arg)
843 {
844 	const struct got_error *err = NULL;
845 	struct got_object_id **ids = NULL;
846 	struct got_object_idset *idset;
847 	int i, nobj = 0, obj_type;
848 	struct got_pack_metavec v;
849 
850 	*meta = NULL;
851 	*nmeta = 0;
852 
853 	idset = got_object_idset_alloc();
854 	if (idset == NULL)
855 		return got_error_from_errno("got_object_idset_alloc");
856 
857 	v.nmeta = 0;
858 	v.metasz = 64;
859 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
860 	if (v.meta == NULL) {
861 		err = got_error_from_errno("calloc");
862 		goto done;
863 	}
864 
865 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
866 	    cancel_cb, cancel_arg);
867 	if (err || nobj == 0)
868 		goto done;
869 
870 	for (i = 0; i < ntheirs; i++) {
871 		struct got_object_id *id = theirs[i];
872 		if (id == NULL)
873 			continue;
874 		err = got_object_get_type(&obj_type, repo, id);
875 		if (err)
876 			return err;
877 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
878 			continue;
879 		err = load_commit(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 < ntheirs; i++) {
892 		struct got_object_id *id = theirs[i];
893 		int *cached_type;
894 		if (id == NULL)
895 			continue;
896 		cached_type = got_object_idset_get(idset, id);
897 		if (cached_type == NULL) {
898 			err = got_object_get_type(&obj_type, repo, id);
899 			if (err)
900 				goto done;
901 		} else
902 			obj_type = *cached_type;
903 		if (obj_type != GOT_OBJ_TYPE_TAG)
904 			continue;
905 		err = load_tag(NULL, idset, id, repo,
906 		    loose_obj_only, cancel_cb, cancel_arg);
907 		if (err)
908 			goto done;
909 		if (progress_cb) {
910 			err = progress_cb(progress_arg, 0L, nours,
911 			    v.nmeta, 0, 0);
912 			if (err)
913 				goto done;
914 		}
915 	}
916 
917 	for (i = 0; i < nobj; i++) {
918 		err = load_commit(&v, idset, ids[i], 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 	for (i = 0; i < nours; i++) {
931 		struct got_object_id *id = ours[i];
932 		int *cached_type;
933 		if (id == NULL)
934 			continue;
935 		cached_type = got_object_idset_get(idset, id);
936 		if (cached_type == NULL) {
937 			err = got_object_get_type(&obj_type, repo, id);
938 			if (err)
939 				goto done;
940 		} else
941 			obj_type = *cached_type;
942 		if (obj_type != GOT_OBJ_TYPE_TAG)
943 			continue;
944 		err = load_tag(&v, idset, id, repo,
945 		    loose_obj_only, cancel_cb, cancel_arg);
946 		if (err)
947 			goto done;
948 		if (progress_cb) {
949 			err = progress_cb(progress_arg, 0L, nours,
950 			    v.nmeta, 0, 0);
951 			if (err)
952 				goto done;
953 		}
954 	}
955 
956 done:
957 	for (i = 0; i < nobj; i++) {
958 		free(ids[i]);
959 	}
960 	free(ids);
961 	got_object_idset_free(idset);
962 	if (err == NULL) {
963 		*meta = v.meta;
964 		*nmeta = v.nmeta;
965 	} else
966 		free(v.meta);
967 
968 	return err;
969 }
970 
971 const struct got_error *
972 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
973 {
974 	size_t n;
975 
976 	SHA1Update(ctx, buf, len);
977 	n = fwrite(buf, 1, len, f);
978 	if (n != len)
979 		return got_ferror(f, GOT_ERR_IO);
980 	return NULL;
981 }
982 
983 static void
984 putbe32(char *b, uint32_t n)
985 {
986 	b[0] = n >> 24;
987 	b[1] = n >> 16;
988 	b[2] = n >> 8;
989 	b[3] = n >> 0;
990 }
991 
992 static int
993 write_order_cmp(const void *pa, const void *pb)
994 {
995 	struct got_pack_meta *a, *b, *ahd, *bhd;
996 
997 	a = *(struct got_pack_meta **)pa;
998 	b = *(struct got_pack_meta **)pb;
999 	ahd = (a->head == NULL) ? a : a->head;
1000 	bhd = (b->head == NULL) ? b : b->head;
1001 	if (ahd->mtime != bhd->mtime)
1002 		return bhd->mtime - ahd->mtime;
1003 	if (ahd != bhd)
1004 		return (uintptr_t)bhd - (uintptr_t)ahd;
1005 	if (a->nchain != b->nchain)
1006 		return a->nchain - b->nchain;
1007 	return a->mtime - b->mtime;
1008 }
1009 
1010 static const struct got_error *
1011 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1012 {
1013 	size_t i;
1014 
1015 	*hdrlen = 0;
1016 
1017 	hdr[0] = obj_type << 4;
1018 	hdr[0] |= len & 0xf;
1019 	len >>= 4;
1020 	for (i = 1; len != 0; i++){
1021 		if (i >= bufsize)
1022 			return got_error(GOT_ERR_NO_SPACE);
1023 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1024 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1025 		len >>= GOT_DELTA_SIZE_SHIFT;
1026 	}
1027 
1028 	*hdrlen = i;
1029 	return NULL;
1030 }
1031 
1032 static const struct got_error *
1033 encodedelta(struct got_pack_meta *m, struct got_raw_object *o,
1034     off_t base_size, FILE *f)
1035 {
1036 	unsigned char buf[16], *bp;
1037 	int i, j;
1038 	off_t n;
1039 	size_t w;
1040 	struct got_delta_instruction *d;
1041 
1042 	/* base object size */
1043 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1044 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1045 	for (i = 1; n > 0; i++) {
1046 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1047 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1048 		n >>= GOT_DELTA_SIZE_SHIFT;
1049 	}
1050 	w = fwrite(buf, 1, i, f);
1051 	if (w != i)
1052 		return got_ferror(f, GOT_ERR_IO);
1053 
1054 	/* target object size */
1055 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1056 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1057 	for (i = 1; n > 0; i++) {
1058 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1059 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1060 		n >>= GOT_DELTA_SIZE_SHIFT;
1061 	}
1062 	w = fwrite(buf, 1, i, f);
1063 	if (w != i)
1064 		return got_ferror(f, GOT_ERR_IO);
1065 
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 			w = fwrite(buf, 1, bp - buf, f);
1091 			if (w != bp - buf)
1092 				return got_ferror(f, GOT_ERR_IO);
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 				w = fwrite(buf, 1, 1, f);
1102 				if (w != 1)
1103 					return got_ferror(f, GOT_ERR_IO);
1104 				r = fread(content, 1, buf[0], o->f);
1105 				if (r != buf[0])
1106 					return got_ferror(o->f, GOT_ERR_IO);
1107 				w = fwrite(content, 1, buf[0], f);
1108 				if (w != buf[0])
1109 					return got_ferror(f, GOT_ERR_IO);
1110 				n += buf[0];
1111 			}
1112 		}
1113 	}
1114 
1115 	return NULL;
1116 }
1117 
1118 static int
1119 packoff(char *hdr, off_t off)
1120 {
1121 	int i, j;
1122 	char rbuf[8];
1123 
1124 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1125 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1126 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1127 		    GOT_DELTA_SIZE_MORE;
1128 	}
1129 
1130 	j = 0;
1131 	while (i > 0)
1132 		hdr[j++] = rbuf[--i];
1133 	return j;
1134 }
1135 
1136 static const struct got_error *
1137 genpack(uint8_t *pack_sha1, FILE *packfile,
1138     struct got_pack_meta **meta, int nmeta, int nours,
1139     int use_offset_deltas, struct got_repository *repo,
1140     got_pack_progress_cb progress_cb, void *progress_arg,
1141     got_cancel_cb cancel_cb, void *cancel_arg)
1142 {
1143 	const struct got_error *err = NULL;
1144 	int i, nh;
1145 	off_t nd;
1146 	SHA1_CTX ctx;
1147 	struct got_pack_meta *m;
1148 	struct got_raw_object *raw = NULL;
1149 	FILE *delta_file = NULL;
1150 	char buf[32];
1151 	size_t outlen, n;
1152 	struct got_deflate_checksum csum;
1153 	off_t packfile_size = 0;
1154 	int outfd = -1;
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 		if (outfd == -1) {
1182 			outfd = got_opentempfd();
1183 			if (outfd == -1) {
1184 				err = got_error_from_errno("got_opentempfd");
1185 				goto done;
1186 			}
1187 		}
1188 		err = got_object_raw_open(&raw, outfd, repo, &m->id, 8192);
1189 		if (err)
1190 			goto done;
1191 		if (raw->data == NULL)
1192 			outfd = -1; /* outfd is now raw->f */
1193 		if (m->deltas == NULL) {
1194 			err = packhdr(&nh, buf, sizeof(buf),
1195 			    m->obj_type, raw->size);
1196 			if (err)
1197 				goto done;
1198 			err = hwrite(packfile, buf, nh, &ctx);
1199 			if (err)
1200 				goto done;
1201 			packfile_size += nh;
1202 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1203 				err = got_error_from_errno("fseeko");
1204 				goto done;
1205 			}
1206 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1207 			    &csum);
1208 			if (err)
1209 				goto done;
1210 			packfile_size += outlen;
1211 		} else {
1212 			if (delta_file == NULL) {
1213 				delta_file = got_opentemp();
1214 				if (delta_file == NULL) {
1215 					err = got_error_from_errno(
1216 					    "got_opentemp");
1217 					goto done;
1218 				}
1219 			}
1220 			if (ftruncate(fileno(delta_file), 0L) == -1) {
1221 				err = got_error_from_errno("ftruncate");
1222 				goto done;
1223 			}
1224 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1225 				err = got_error_from_errno("fseeko");
1226 				goto done;
1227 			}
1228 			err = encodedelta(m, raw, m->prev->size, delta_file);
1229 			if (err)
1230 				goto done;
1231 			nd = ftello(delta_file);
1232 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1233 				err = got_error_from_errno("fseeko");
1234 				goto done;
1235 			}
1236 			if (use_offset_deltas && m->prev->off != 0) {
1237 				err = packhdr(&nh, buf, sizeof(buf),
1238 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1239 				if (err)
1240 					goto done;
1241 				nh += packoff(buf + nh,
1242 				    m->off - m->prev->off);
1243 				err = hwrite(packfile, buf, nh, &ctx);
1244 				if (err)
1245 					goto done;
1246 				packfile_size += nh;
1247 			} else {
1248 				err = packhdr(&nh, buf, sizeof(buf),
1249 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1250 				err = hwrite(packfile, buf, nh, &ctx);
1251 				if (err)
1252 					goto done;
1253 				packfile_size += nh;
1254 				err = hwrite(packfile, m->prev->id.sha1,
1255 				    sizeof(m->prev->id.sha1), &ctx);
1256 				packfile_size += sizeof(m->prev->id.sha1);
1257 				if (err)
1258 					goto done;
1259 			}
1260 			err = got_deflate_to_file(&outlen, delta_file,
1261 			    packfile, &csum);
1262 			if (err)
1263 				goto done;
1264 			packfile_size += outlen;
1265 		}
1266 		got_object_raw_close(raw);
1267 		raw = NULL;
1268 	}
1269 	SHA1Final(pack_sha1, &ctx);
1270 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1271 	if (n != SHA1_DIGEST_LENGTH)
1272 		err = got_ferror(packfile, GOT_ERR_IO);
1273 	packfile_size += SHA1_DIGEST_LENGTH;
1274 	packfile_size += sizeof(struct got_packfile_hdr);
1275 	err = progress_cb(progress_arg, packfile_size, nours,
1276 	    nmeta, nmeta, nmeta);
1277 	if (err)
1278 		goto done;
1279 done:
1280 	if (delta_file && fclose(delta_file) == EOF && err == NULL)
1281 		err = got_error_from_errno("fclose");
1282 	if (raw)
1283 		got_object_raw_close(raw);
1284 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1285 		err = got_error_from_errno("close");
1286 	return err;
1287 }
1288 
1289 const struct got_error *
1290 got_pack_create(uint8_t *packsha1, FILE *packfile,
1291     struct got_object_id **theirs, int ntheirs,
1292     struct got_object_id **ours, int nours,
1293     struct got_repository *repo, int loose_obj_only, int allow_empty,
1294     got_pack_progress_cb progress_cb, void *progress_arg,
1295     got_cancel_cb cancel_cb, void *cancel_arg)
1296 {
1297 	const struct got_error *err;
1298 	struct got_pack_meta **meta;
1299 	int nmeta;
1300 
1301 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1302 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1303 	if (err)
1304 		return err;
1305 
1306 	if (nmeta == 0 && !allow_empty) {
1307 		err = got_error(GOT_ERR_CANNOT_PACK);
1308 		goto done;
1309 	}
1310 	if (nmeta > 0) {
1311 		err = pick_deltas(meta, nmeta, nours, repo,
1312 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1313 		if (err)
1314 			goto done;
1315 	}
1316 
1317 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1318 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1319 	if (err)
1320 		goto done;
1321 done:
1322 	free_nmeta(meta, nmeta);
1323 	return err;
1324 }