Got Mirror

Blob

Date:: Fri Oct 15 10:34:44 2021 UTC
Message:: while packing, store encoded deltas in a temporary file instead of in memory
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 MIN
52 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
53 #endif
54 
55 #ifndef MAX
56 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
57 #endif
58 
59 struct got_pack_meta {
60 	struct got_object_id id;
61 	char	*path;
62 	int	obj_type;
63 	off_t	size;
64 	time_t	mtime;
65 
66 	/* The best delta we picked */
67 	struct got_pack_meta *head;
68 	struct got_pack_meta *prev;
69 	off_t	delta_offset;	/* offset in delta cache file */
70 	off_t	delta_len;	/* length in delta cache file */
71 	int	nchain;
72 
73 	/* Only used for delta window */
74 	struct got_delta_table *dtab;
75 
76 	/* Only used for writing offset deltas */
77 	off_t	off;
78 };
79 
80 struct got_pack_metavec {
81 	struct got_pack_meta **meta;
82 	int nmeta;
83 	int metasz;
84 };
85 
86 static const struct got_error *
87 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
88     const char *path, int obj_type, time_t mtime)
89 {
90 	const struct got_error *err = NULL;
91 	struct got_pack_meta *m;
92 
93 	*new = NULL;
94 
95 	m = calloc(1, sizeof(*m));
96 	if (m == NULL)
97 		return got_error_from_errno("calloc");
98 
99 	memcpy(&m->id, id, sizeof(m->id));
100 
101 	m->path = strdup(path);
102 	if (m->path == NULL) {
103 		err = got_error_from_errno("strdup");
104 		free(m);
105 		return err;
106 	}
107 
108 	m->obj_type = obj_type;
109 	m->mtime = mtime;
110 	*new = m;
111 	return NULL;
112 }
113 
114 static void
115 clear_meta(struct got_pack_meta *meta)
116 {
117 	if (meta == NULL)
118 		return;
119 	free(meta->path);
120 	meta->path = NULL;
121 }
122 
123 static void
124 free_nmeta(struct got_pack_meta **meta, int nmeta)
125 {
126 	int i;
127 
128 	for (i = 0; i < nmeta; i++)
129 		clear_meta(meta[i]);
130 	free(meta);
131 }
132 
133 static int
134 delta_order_cmp(const void *pa, const void *pb)
135 {
136 	struct got_pack_meta *a, *b;
137 	int cmp;
138 
139 	a = *(struct got_pack_meta **)pa;
140 	b = *(struct got_pack_meta **)pb;
141 
142 	if (a->obj_type != b->obj_type)
143 		return a->obj_type - b->obj_type;
144 	cmp = strcmp(a->path, b->path);
145 	if (cmp != 0)
146 		return cmp;
147 	if (a->mtime != b->mtime)
148 		return a->mtime - b->mtime;
149 	return got_object_id_cmp(&a->id, &b->id);
150 }
151 
152 static int
153 delta_size(struct got_delta_instruction *deltas, int ndeltas)
154 {
155 	int i, size = 32;
156 	for (i = 0; i < ndeltas; i++) {
157 		if (deltas[i].copy)
158 			size += GOT_DELTA_SIZE_SHIFT;
159 		else
160 			size += deltas[i].len + 1;
161 	}
162 	return size;
163 }
164 
165 static const struct got_error *
166 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
167     struct got_delta_instruction *deltas, int ndeltas,
168     off_t base_size, FILE *f);
169 
170 static const struct got_error *
171 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
172     FILE *delta_cache, struct got_repository *repo,
173     got_pack_progress_cb progress_cb, void *progress_arg,
174     got_cancel_cb cancel_cb, void *cancel_arg)
175 {
176 	const struct got_error *err = NULL;
177 	struct got_pack_meta *m = NULL, *base = NULL;
178 	struct got_raw_object *raw = NULL, *base_raw = NULL;
179 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
180 	int i, j, size, best_size, ndeltas, best_ndeltas;
181 	const int max_base_candidates = 10;
182 	int outfd = -1;
183 
184 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
185 	for (i = 0; i < nmeta; i++) {
186 		if (cancel_cb) {
187 			err = (*cancel_cb)(cancel_arg);
188 			if (err)
189 				break;
190 		}
191 		if (progress_cb) {
192 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
193 			if (err)
194 				goto done;
195 		}
196 		m = meta[i];
197 
198 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
199 		    m->obj_type == GOT_OBJ_TYPE_TAG)
200 			continue;
201 
202 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
203 		if (err)
204 			goto done;
205 		m->size = raw->size;
206 
207 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
208 		    raw->size + raw->hdrlen);
209 		if (err)
210 			goto done;
211 
212 		if (i > max_base_candidates) {
213 			struct got_pack_meta *n = NULL;
214 			n = meta[i - (max_base_candidates + 1)];
215 			got_deltify_free(n->dtab);
216 			n->dtab = NULL;
217 		}
218 
219 		best_size = raw->size;
220 		best_ndeltas = 0;
221 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
222 			if (cancel_cb) {
223 				err = (*cancel_cb)(cancel_arg);
224 				if (err)
225 					goto done;
226 			}
227 			base = meta[j];
228 			/* long chains make unpacking slow, avoid such bases */
229 			if (base->nchain >= 32 ||
230 			    base->obj_type != m->obj_type)
231 				continue;
232 
233 			err = got_object_raw_open(&base_raw, &outfd, repo,
234 			    &base->id);
235 			if (err)
236 				goto done;
237 			err = got_deltify(&deltas, &ndeltas,
238 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
239 			    base->dtab, base_raw->f, base_raw->hdrlen,
240 			    base_raw->size + base_raw->hdrlen);
241 			got_object_raw_close(base_raw);
242 			base_raw = NULL;
243 			if (err)
244 				goto done;
245 
246 			size = delta_size(deltas, ndeltas);
247 			if (size + 32 < best_size){
248 				/*
249 				 * if we already picked a best delta,
250 				 * replace it.
251 				 */
252 				best_size = size;
253 				free(best_deltas);
254 				best_deltas = deltas;
255 				best_ndeltas = ndeltas;
256 				deltas = NULL;
257 				m->nchain = base->nchain + 1;
258 				m->prev = base;
259 				m->head = base->head;
260 				if (m->head == NULL)
261 					m->head = base;
262 			} else {
263 				free(deltas);
264 				deltas = NULL;
265 				ndeltas = 0;
266 			}
267 		}
268 
269 		if (best_ndeltas > 0) {
270 			m->delta_offset = ftello(delta_cache);
271 			err = encode_delta(m, raw, best_deltas,
272 			    best_ndeltas, m->prev->size, delta_cache);
273 			free(best_deltas);
274 			best_deltas = NULL;
275 			best_ndeltas = 0;
276 			if (err)
277 				goto done;
278 			m->delta_len = ftello(delta_cache) - m->delta_offset;
279 		}
280 
281 		got_object_raw_close(raw);
282 		raw = NULL;
283 	}
284 done:
285 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
286 		got_deltify_free(meta[i]->dtab);
287 		meta[i]->dtab = NULL;
288 	}
289 	if (raw)
290 		got_object_raw_close(raw);
291 	if (base_raw)
292 		got_object_raw_close(base_raw);
293 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
294 		err = got_error_from_errno("close");
295 	free(deltas);
296 	free(best_deltas);
297 	return err;
298 }
299 
300 static const struct got_error *
301 search_packidx(int *found, struct got_object_id *id,
302     struct got_repository *repo)
303 {
304 	const struct got_error *err = NULL;
305 	struct got_packidx *packidx = NULL;
306 	int idx;
307 
308 	*found = 0;
309 
310 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
311 	if (err == NULL)
312 		*found = 1; /* object is already packed */
313 	else if (err->code == GOT_ERR_NO_OBJ)
314 		err = NULL;
315 	return err;
316 }
317 
318 static const int obj_types[] = {
319 	GOT_OBJ_TYPE_ANY,
320 	GOT_OBJ_TYPE_COMMIT,
321 	GOT_OBJ_TYPE_TREE,
322 	GOT_OBJ_TYPE_BLOB,
323 	GOT_OBJ_TYPE_TAG,
324 	GOT_OBJ_TYPE_OFFSET_DELTA,
325 	GOT_OBJ_TYPE_REF_DELTA
326 };
327 
328 static const struct got_error *
329 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
330     struct got_object_id *id, const char *path, int obj_type,
331     time_t mtime, int loose_obj_only, struct got_repository *repo)
332 {
333 	const struct got_error *err;
334 	struct got_pack_meta *m;
335 
336 	if (loose_obj_only) {
337 		int is_packed;
338 		err = search_packidx(&is_packed, id, repo);
339 		if (err)
340 			return err;
341 		if (is_packed)
342 			return NULL;
343 	}
344 
345 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
346 	if (err)
347 		return err;
348 
349 	if (v == NULL)
350 		return NULL;
351 
352 	err = alloc_meta(&m, id, path, obj_type, mtime);
353 	if (err)
354 		goto done;
355 
356 	if (v->nmeta == v->metasz){
357 		size_t newsize = 2 * v->metasz;
358 		struct got_pack_meta **new;
359 		new = reallocarray(v->meta, newsize, sizeof(*new));
360 		if (new == NULL) {
361 			err = got_error_from_errno("reallocarray");
362 			goto done;
363 		}
364 		v->meta = new;
365 		v->metasz = newsize; 
366 	}
367 done:
368 	if (err) {
369 		clear_meta(m);
370 		free(m);
371 	} else
372 		v->meta[v->nmeta++] = m;
373 
374 	return err;
375 }
376 
377 static const struct got_error *
378 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
379     struct got_object_idset *idset, struct got_object_id *tree_id,
380     const char *dpath, time_t mtime, struct got_repository *repo,
381     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
382 {
383 	const struct got_error *err;
384 	struct got_tree_object *tree;
385 	char *p = NULL;
386 	int i;
387 
388 	err = got_object_open_as_tree(&tree, repo, tree_id);
389 	if (err)
390 		return err;
391 
392 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
393 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
394 		struct got_object_id *id = got_tree_entry_get_id(e);
395 		mode_t mode = got_tree_entry_get_mode(e);
396 
397 		if (cancel_cb) {
398 			err = (*cancel_cb)(cancel_arg);
399 			if (err)
400 				break;
401 		}
402 
403 		if (got_object_tree_entry_is_submodule(e) ||
404 		    got_object_idset_contains(idset, id))
405 			continue;
406 		
407 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
408 		    got_tree_entry_get_name(e)) == -1) {
409 			err = got_error_from_errno("asprintf");
410 			break;
411 		}
412 
413 		if (S_ISDIR(mode)) {
414 			struct got_object_qid *qid;
415 			err = got_object_qid_alloc(&qid, id);
416 			if (err)
417 				break;
418 			STAILQ_INSERT_TAIL(ids, qid, entry);
419 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
420 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
421 			    mtime, loose_obj_only, repo);
422 			if (err)
423 				break;
424 		}
425 		free(p);
426 		p = NULL;
427 	}
428 
429 	got_object_tree_close(tree);
430 	free(p);
431 	return err;
432 }
433 
434 static const struct got_error *
435 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
436     struct got_object_id *tree_id, const char *dpath, time_t mtime,
437     int loose_obj_only, struct got_repository *repo,
438     got_cancel_cb cancel_cb, void *cancel_arg)
439 {
440 	const struct got_error *err = NULL;
441 	struct got_object_id_queue tree_ids;
442 	struct got_object_qid *qid;
443 
444 	if (got_object_idset_contains(idset, tree_id))
445 		return NULL;
446 
447 	err = got_object_qid_alloc(&qid, tree_id);
448 	if (err)
449 		return err;
450 
451 	STAILQ_INIT(&tree_ids);
452 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
453 
454 	while (!STAILQ_EMPTY(&tree_ids)) {
455 		if (cancel_cb) {
456 			err = (*cancel_cb)(cancel_arg);
457 			if (err)
458 				break;
459 		}
460 
461 		qid = STAILQ_FIRST(&tree_ids);
462 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
463 
464 		if (got_object_idset_contains(idset, qid->id)) {
465 			got_object_qid_free(qid);
466 			continue;
467 		}
468 
469 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
470 		    mtime, loose_obj_only, repo);
471 		if (err) {
472 			got_object_qid_free(qid);
473 			break;
474 		}
475 
476 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
477 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
478 		got_object_qid_free(qid);
479 		if (err)
480 			break;
481 	}
482 
483 	got_object_id_queue_free(&tree_ids);
484 	return err;
485 }
486 
487 static const struct got_error *
488 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
489     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
490     got_cancel_cb cancel_cb, void *cancel_arg)
491 {
492 	const struct got_error *err;
493 	struct got_commit_object *commit;
494 
495 	if (got_object_idset_contains(idset, id))
496 		return NULL;
497 
498 	if (loose_obj_only) {
499 		int is_packed;
500 		err = search_packidx(&is_packed, id, repo);
501 		if (err)
502 			return err;
503 		if (is_packed)
504 			return NULL;
505 	}
506 
507 	err = got_object_open_as_commit(&commit, repo, id);
508 	if (err)
509 		return err;
510 
511 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
512 	    got_object_commit_get_committer_time(commit),
513 	    loose_obj_only, repo);
514 	if (err)
515 		goto done;
516 
517 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
518 	    "", got_object_commit_get_committer_time(commit),
519 	    loose_obj_only, repo, cancel_cb, cancel_arg);
520 done:
521 	got_object_commit_close(commit);
522 	return err;
523 }
524 
525 static const struct got_error *
526 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
527     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
528     got_cancel_cb cancel_cb, void *cancel_arg)
529 {
530 	const struct got_error *err;
531 	struct got_tag_object *tag = NULL;
532 
533 	if (got_object_idset_contains(idset, id))
534 		return NULL;
535 
536 	if (loose_obj_only) {
537 		int is_packed;
538 		err = search_packidx(&is_packed, id, repo);
539 		if (err)
540 			return err;
541 		if (is_packed)
542 			return NULL;
543 	}
544 
545 	err = got_object_open_as_tag(&tag, repo, id);
546 	if (err)
547 		return err;
548 
549 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
550 	    got_object_tag_get_tagger_time(tag),
551 	    loose_obj_only, repo);
552 	if (err)
553 		goto done;
554 
555 	switch (got_object_tag_get_object_type(tag)) {
556 	case GOT_OBJ_TYPE_COMMIT:
557 		err = load_commit(v, idset,
558 		    got_object_tag_get_object_id(tag), repo,
559 		    loose_obj_only, cancel_cb, cancel_arg);
560 		break;
561 	case GOT_OBJ_TYPE_TREE:
562 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
563 		    "", got_object_tag_get_tagger_time(tag),
564 		    loose_obj_only, repo, cancel_cb, cancel_arg);
565 		break;
566 	default:
567 		break;
568 	}
569 
570 done:
571 	got_object_tag_close(tag);
572 	return err;
573 }
574 
575 enum findtwixt_color {
576 	COLOR_KEEP = 0,
577 	COLOR_DROP,
578 	COLOR_BLANK,
579 };
580 static const int findtwixt_colors[] = {
581 	COLOR_KEEP,
582 	COLOR_DROP,
583 	COLOR_BLANK
584 };
585 
586 static const struct got_error *
587 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
588     int color, struct got_repository *repo)
589 {
590 	const struct got_error *err;
591 	struct got_object_qid *qid;
592 
593 	err = got_object_qid_alloc(&qid, id);
594 	if (err)
595 		return err;
596 
597 	STAILQ_INSERT_TAIL(ids, qid, entry);
598 	qid->data = (void *)&findtwixt_colors[color];
599 	return NULL;
600 }
601 
602 static const struct got_error *
603 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
604     struct got_object_id *id, struct got_repository *repo,
605     got_cancel_cb cancel_cb, void *cancel_arg)
606 {
607 	const struct got_error *err = NULL;
608 	struct got_commit_object *commit;
609 	const struct got_object_id_queue *parents;
610 	struct got_object_id_queue ids;
611 	struct got_object_qid *qid;
612 
613 	STAILQ_INIT(&ids);
614 
615 	err = got_object_qid_alloc(&qid, id);
616 	if (err)
617 		return err;
618 	STAILQ_INSERT_HEAD(&ids, qid, entry);
619 
620 	while (!STAILQ_EMPTY(&ids)) {
621 		if (cancel_cb) {
622 			err = (*cancel_cb)(cancel_arg);
623 			if (err)
624 				break;
625 		}
626 
627 		qid = STAILQ_FIRST(&ids);
628 		STAILQ_REMOVE_HEAD(&ids, entry);
629 
630 		if (got_object_idset_contains(drop, qid->id)) {
631 			got_object_qid_free(qid);
632 			continue;
633 		}
634 
635 		err = got_object_idset_add(drop, qid->id,
636 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
637 		if (err) {
638 			got_object_qid_free(qid);
639 			break;
640 		}
641 
642 		if (!got_object_idset_contains(keep, qid->id)) {
643 			got_object_qid_free(qid);
644 			continue;
645 		}
646 
647 		err = got_object_open_as_commit(&commit, repo, qid->id);
648 		got_object_qid_free(qid);
649 		if (err)
650 			break;
651 
652 		parents = got_object_commit_get_parent_ids(commit);
653 		if (parents) {
654 			err = got_object_id_queue_copy(parents, &ids);
655 			if (err) {
656 				got_object_commit_close(commit);
657 				break;
658 			}
659 		}
660 		got_object_commit_close(commit);
661 	}
662 
663 	got_object_id_queue_free(&ids);
664 	return err;
665 }
666 
667 struct append_id_arg {
668 	struct got_object_id **array;
669 	int idx;
670 };
671 
672 static const struct got_error *
673 append_id(struct got_object_id *id, void *data, void *arg)
674 {
675 	struct append_id_arg *a = arg;
676 
677 	a->array[a->idx] = got_object_id_dup(id);
678 	if (a->array[a->idx] == NULL)
679 		return got_error_from_errno("got_object_id_dup");
680 
681 	a->idx++;
682 	return NULL;
683 }
684 
685 static const struct got_error *
686 findtwixt(struct got_object_id ***res, int *nres,
687     struct got_object_id **head, int nhead,
688     struct got_object_id **tail, int ntail,
689     struct got_repository *repo,
690     got_cancel_cb cancel_cb, void *cancel_arg)
691 {
692 	const struct got_error *err = NULL;
693 	struct got_object_id_queue ids;
694 	struct got_object_idset *keep, *drop;
695 	struct got_object_qid *qid;
696 	int i, ncolor, nkeep, obj_type;
697 
698 	STAILQ_INIT(&ids);
699 	*res = NULL;
700 	*nres = 0;
701 
702 	keep = got_object_idset_alloc();
703 	if (keep == NULL)
704 		return got_error_from_errno("got_object_idset_alloc");
705 
706 	drop = got_object_idset_alloc();
707 	if (drop == NULL) {
708 		err = got_error_from_errno("got_object_idset_alloc");
709 		goto done;
710 	}
711 
712 	for (i = 0; i < nhead; i++) {
713 		struct got_object_id *id = head[i];
714 		if (id == NULL)
715 			continue;
716 		err = got_object_get_type(&obj_type, repo, id);
717 		if (err)
718 			return err;
719 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
720 			continue;
721 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
722 		if (err)
723 			goto done;
724 	}		
725 	for (i = 0; i < ntail; i++) {
726 		struct got_object_id *id = tail[i];
727 		if (id == NULL)
728 			continue;
729 		err = got_object_get_type(&obj_type, repo, id);
730 		if (err)
731 			return err;
732 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
733 			continue;
734 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
735 		if (err)
736 			goto done;
737 	}
738 
739 	while (!STAILQ_EMPTY(&ids)) {
740 		int qcolor;
741 		qid = STAILQ_FIRST(&ids);
742 		qcolor = *((int *)qid->data);
743 
744 		if (got_object_idset_contains(drop, qid->id))
745 			ncolor = COLOR_DROP;
746 		else if (got_object_idset_contains(keep, qid->id))
747 			ncolor = COLOR_KEEP;
748 		else
749 			ncolor = COLOR_BLANK;
750 
751 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
752 		    qcolor == COLOR_KEEP)) {
753 			STAILQ_REMOVE_HEAD(&ids, entry);
754 			got_object_qid_free(qid);
755 			continue;
756 		}
757 
758 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
759 			err = drop_commit(keep, drop, qid->id, repo,
760 			    cancel_cb, cancel_arg);
761 			if (err)
762 				goto done;
763 		} else if (ncolor == COLOR_BLANK) {
764 			struct got_commit_object *commit;
765 			struct got_object_id *id;
766 			const struct got_object_id_queue *parents;
767 			struct got_object_qid *pid;
768 
769 			id = got_object_id_dup(qid->id);
770 			if (id == NULL) {
771 				err = got_error_from_errno("got_object_id_dup");
772 				goto done;
773 			}
774 			if (qcolor == COLOR_KEEP)
775 				err = got_object_idset_add(keep, id,
776 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
777 			else
778 				err = got_object_idset_add(drop, id,
779 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
780 			if (err) {
781 				free(id);
782 				goto done;
783 			}
784 
785 			err = got_object_open_as_commit(&commit, repo, id);
786 			if (err) {
787 				free(id);
788 				goto done;
789 			}
790 			parents = got_object_commit_get_parent_ids(commit);
791 			if (parents) {
792 				STAILQ_FOREACH(pid, parents, entry) {
793 					err = queue_commit_id(&ids, pid->id,
794 					    qcolor, repo);
795 					if (err) {
796 						free(id);
797 						goto done;
798 					}
799 				}
800 			}
801 			got_object_commit_close(commit);
802 			commit = NULL;
803 		} else {
804 			/* should not happen */
805 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
806 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
807 			goto done;
808 		}
809 
810 		STAILQ_REMOVE_HEAD(&ids, entry);
811 		got_object_qid_free(qid);
812 	}
813 
814 	nkeep = got_object_idset_num_elements(keep);
815 	if (nkeep > 0) {
816 		struct append_id_arg arg;
817 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
818 		if (arg.array == NULL) {
819 			err = got_error_from_errno("calloc");
820 			goto done;
821 		}
822 		arg.idx = 0;
823 		err = got_object_idset_for_each(keep, append_id, &arg);
824 		if (err) {
825 			free(arg.array);
826 			goto done;
827 		}
828 		*res = arg.array;
829 		*nres = nkeep;
830 	}
831 done:
832 	got_object_idset_free(keep);
833 	got_object_idset_free(drop);
834 	got_object_id_queue_free(&ids);
835 	return err;
836 }
837 
838 static const struct got_error *
839 read_meta(struct got_pack_meta ***meta, int *nmeta,
840     struct got_object_id **theirs, int ntheirs,
841     struct got_object_id **ours, int nours, struct got_repository *repo,
842     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
843     got_cancel_cb cancel_cb, void *cancel_arg)
844 {
845 	const struct got_error *err = NULL;
846 	struct got_object_id **ids = NULL;
847 	struct got_object_idset *idset;
848 	int i, nobj = 0, obj_type;
849 	struct got_pack_metavec v;
850 
851 	*meta = NULL;
852 	*nmeta = 0;
853 
854 	idset = got_object_idset_alloc();
855 	if (idset == NULL)
856 		return got_error_from_errno("got_object_idset_alloc");
857 
858 	v.nmeta = 0;
859 	v.metasz = 64;
860 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
861 	if (v.meta == NULL) {
862 		err = got_error_from_errno("calloc");
863 		goto done;
864 	}
865 
866 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
867 	    cancel_cb, cancel_arg);
868 	if (err || nobj == 0)
869 		goto done;
870 
871 	for (i = 0; i < ntheirs; i++) {
872 		struct got_object_id *id = theirs[i];
873 		if (id == NULL)
874 			continue;
875 		err = got_object_get_type(&obj_type, repo, id);
876 		if (err)
877 			return err;
878 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
879 			continue;
880 		err = load_commit(NULL, idset, id, repo,
881 		    loose_obj_only, cancel_cb, cancel_arg);
882 		if (err)
883 			goto done;
884 		if (progress_cb) {
885 			err = progress_cb(progress_arg, 0L, nours,
886 			    v.nmeta, 0, 0);
887 			if (err)
888 				goto done;
889 		}
890 	}
891 
892 	for (i = 0; i < ntheirs; i++) {
893 		struct got_object_id *id = theirs[i];
894 		int *cached_type;
895 		if (id == NULL)
896 			continue;
897 		cached_type = got_object_idset_get(idset, id);
898 		if (cached_type == NULL) {
899 			err = got_object_get_type(&obj_type, repo, id);
900 			if (err)
901 				goto done;
902 		} else
903 			obj_type = *cached_type;
904 		if (obj_type != GOT_OBJ_TYPE_TAG)
905 			continue;
906 		err = load_tag(NULL, idset, id, repo,
907 		    loose_obj_only, cancel_cb, cancel_arg);
908 		if (err)
909 			goto done;
910 		if (progress_cb) {
911 			err = progress_cb(progress_arg, 0L, nours,
912 			    v.nmeta, 0, 0);
913 			if (err)
914 				goto done;
915 		}
916 	}
917 
918 	for (i = 0; i < nobj; i++) {
919 		err = load_commit(&v, idset, ids[i], repo,
920 		    loose_obj_only, cancel_cb, cancel_arg);
921 		if (err)
922 			goto done;
923 		if (progress_cb) {
924 			err = progress_cb(progress_arg, 0L, nours,
925 			    v.nmeta, 0, 0);
926 			if (err)
927 				goto done;
928 		}
929 	}
930 
931 	for (i = 0; i < nours; i++) {
932 		struct got_object_id *id = ours[i];
933 		int *cached_type;
934 		if (id == NULL)
935 			continue;
936 		cached_type = got_object_idset_get(idset, id);
937 		if (cached_type == NULL) {
938 			err = got_object_get_type(&obj_type, repo, id);
939 			if (err)
940 				goto done;
941 		} else
942 			obj_type = *cached_type;
943 		if (obj_type != GOT_OBJ_TYPE_TAG)
944 			continue;
945 		err = load_tag(&v, idset, id, repo,
946 		    loose_obj_only, cancel_cb, cancel_arg);
947 		if (err)
948 			goto done;
949 		if (progress_cb) {
950 			err = progress_cb(progress_arg, 0L, nours,
951 			    v.nmeta, 0, 0);
952 			if (err)
953 				goto done;
954 		}
955 	}
956 
957 done:
958 	for (i = 0; i < nobj; i++) {
959 		free(ids[i]);
960 	}
961 	free(ids);
962 	got_object_idset_free(idset);
963 	if (err == NULL) {
964 		*meta = v.meta;
965 		*nmeta = v.nmeta;
966 	} else
967 		free(v.meta);
968 
969 	return err;
970 }
971 
972 const struct got_error *
973 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
974 {
975 	size_t n;
976 
977 	SHA1Update(ctx, buf, len);
978 	n = fwrite(buf, 1, len, f);
979 	if (n != len)
980 		return got_ferror(f, GOT_ERR_IO);
981 	return NULL;
982 }
983 
984 static void
985 putbe32(char *b, uint32_t n)
986 {
987 	b[0] = n >> 24;
988 	b[1] = n >> 16;
989 	b[2] = n >> 8;
990 	b[3] = n >> 0;
991 }
992 
993 static int
994 write_order_cmp(const void *pa, const void *pb)
995 {
996 	struct got_pack_meta *a, *b, *ahd, *bhd;
997 
998 	a = *(struct got_pack_meta **)pa;
999 	b = *(struct got_pack_meta **)pb;
1000 	ahd = (a->head == NULL) ? a : a->head;
1001 	bhd = (b->head == NULL) ? b : b->head;
1002 	if (ahd->mtime != bhd->mtime)
1003 		return bhd->mtime - ahd->mtime;
1004 	if (ahd != bhd)
1005 		return (uintptr_t)bhd - (uintptr_t)ahd;
1006 	if (a->nchain != b->nchain)
1007 		return a->nchain - b->nchain;
1008 	return a->mtime - b->mtime;
1009 }
1010 
1011 static const struct got_error *
1012 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1013 {
1014 	size_t i;
1015 
1016 	*hdrlen = 0;
1017 
1018 	hdr[0] = obj_type << 4;
1019 	hdr[0] |= len & 0xf;
1020 	len >>= 4;
1021 	for (i = 1; len != 0; i++){
1022 		if (i >= bufsize)
1023 			return got_error(GOT_ERR_NO_SPACE);
1024 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1025 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1026 		len >>= GOT_DELTA_SIZE_SHIFT;
1027 	}
1028 
1029 	*hdrlen = i;
1030 	return NULL;
1031 }
1032 
1033 static const struct got_error *
1034 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
1035     struct got_delta_instruction *deltas, int ndeltas,
1036     off_t base_size, FILE *f)
1037 {
1038 	unsigned char buf[16], *bp;
1039 	int i, j;
1040 	off_t n;
1041 	size_t w;
1042 	struct got_delta_instruction *d;
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 	w = fwrite(buf, 1, i, f);
1053 	if (w != i)
1054 		return got_ferror(f, GOT_ERR_IO);
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 	w = fwrite(buf, 1, i, f);
1065 	if (w != i)
1066 		return got_ferror(f, GOT_ERR_IO);
1067 
1068 	for (j = 0; j < ndeltas; j++) {
1069 		d = &deltas[j];
1070 		if (d->copy) {
1071 			n = d->offset;
1072 			bp = &buf[1];
1073 			buf[0] = GOT_DELTA_BASE_COPY;
1074 			for (i = 0; i < 4; i++) {
1075 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1076 				buf[0] |= 1 << i;
1077 				*bp++ = n & 0xff;
1078 				n >>= 8;
1079 				if (n == 0)
1080 					break;
1081 			}
1082 
1083 			n = d->len;
1084 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1085 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1086 				for (i = 0; i < 3 && n > 0; i++) {
1087 					buf[0] |= 1 << (i + 4);
1088 					*bp++ = n & 0xff;
1089 					n >>= 8;
1090 				}
1091 			}
1092 			w = fwrite(buf, 1, bp - buf, f);
1093 			if (w != bp - buf)
1094 				return got_ferror(f, GOT_ERR_IO);
1095 		} else {
1096 			char content[128];
1097 			size_t r;
1098 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1099 				return got_error_from_errno("fseeko");
1100 			n = 0;
1101 			while (n != d->len) {
1102 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1103 				w = fwrite(buf, 1, 1, f);
1104 				if (w != 1)
1105 					return got_ferror(f, GOT_ERR_IO);
1106 				r = fread(content, 1, buf[0], o->f);
1107 				if (r != buf[0])
1108 					return got_ferror(o->f, GOT_ERR_IO);
1109 				w = fwrite(content, 1, buf[0], f);
1110 				if (w != buf[0])
1111 					return got_ferror(f, GOT_ERR_IO);
1112 				n += buf[0];
1113 			}
1114 		}
1115 	}
1116 
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, FILE *delta_cache,
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;
1147 	SHA1_CTX ctx;
1148 	struct got_pack_meta *m;
1149 	struct got_raw_object *raw = NULL;
1150 	FILE *delta_file = NULL;
1151 	char buf[32];
1152 	size_t outlen, n;
1153 	struct got_deflate_checksum csum;
1154 	off_t packfile_size = 0;
1155 	int outfd = -1;
1156 
1157 	SHA1Init(&ctx);
1158 	csum.output_sha1 = &ctx;
1159 	csum.output_crc = NULL;
1160 
1161 	err = hwrite(packfile, "PACK", 4, &ctx);
1162 	if (err)
1163 		return err;
1164 	putbe32(buf, GOT_PACKFILE_VERSION);
1165 	err = hwrite(packfile, buf, 4, &ctx);
1166 	if (err)
1167 		goto done;
1168 	putbe32(buf, nmeta);
1169 	err = hwrite(packfile, buf, 4, &ctx);
1170 	if (err)
1171 		goto done;
1172 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1173 	for (i = 0; i < nmeta; i++) {
1174 		if (progress_cb) {
1175 			err = progress_cb(progress_arg, packfile_size, nours,
1176 			    nmeta, nmeta, i);
1177 			if (err)
1178 				goto done;
1179 		}
1180 		m = meta[i];
1181 		m->off = ftello(packfile);
1182 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
1183 		if (err)
1184 			goto done;
1185 		if (m->delta_len == 0) {
1186 			err = packhdr(&nh, buf, sizeof(buf),
1187 			    m->obj_type, raw->size);
1188 			if (err)
1189 				goto done;
1190 			err = hwrite(packfile, buf, nh, &ctx);
1191 			if (err)
1192 				goto done;
1193 			packfile_size += nh;
1194 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1195 				err = got_error_from_errno("fseeko");
1196 				goto done;
1197 			}
1198 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1199 			    &csum);
1200 			if (err)
1201 				goto done;
1202 			packfile_size += outlen;
1203 		} else {
1204 			off_t remain;
1205 			if (delta_file == NULL) {
1206 				delta_file = got_opentemp();
1207 				if (delta_file == NULL) {
1208 					err = got_error_from_errno(
1209 					    "got_opentemp");
1210 					goto done;
1211 				}
1212 			}
1213 			if (ftruncate(fileno(delta_file), 0L) == -1) {
1214 				err = got_error_from_errno("ftruncate");
1215 				goto done;
1216 			}
1217 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1218 				err = got_error_from_errno("fseeko");
1219 				goto done;
1220 			}
1221 			if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1222 			    == -1) {
1223 				err = got_error_from_errno("fseeko");
1224 				goto done;
1225 			}
1226 			remain = m->delta_len;
1227 			while (remain > 0) {
1228 				char delta_buf[8192];
1229 				size_t r, w, n;
1230 				n = MIN(remain, sizeof(delta_buf));
1231 				r = fread(delta_buf, 1, n, delta_cache);
1232 				if (r != n) {
1233 					err = got_ferror(delta_cache,
1234 					    GOT_ERR_IO);
1235 					goto done;
1236 				}
1237 				w = fwrite(delta_buf, 1, n, delta_file);
1238 				if (w != n) {
1239 					err = got_ferror(delta_file,
1240 					    GOT_ERR_IO);
1241 					goto done;
1242 				}
1243 				remain -= n;
1244 			}
1245 			if (use_offset_deltas && m->prev->off != 0) {
1246 				err = packhdr(&nh, buf, sizeof(buf),
1247 				    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1248 				if (err)
1249 					goto done;
1250 				nh += packoff(buf + nh,
1251 				    m->off - m->prev->off);
1252 				err = hwrite(packfile, buf, nh, &ctx);
1253 				if (err)
1254 					goto done;
1255 				packfile_size += nh;
1256 			} else {
1257 				err = packhdr(&nh, buf, sizeof(buf),
1258 				    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1259 				err = hwrite(packfile, buf, nh, &ctx);
1260 				if (err)
1261 					goto done;
1262 				packfile_size += nh;
1263 				err = hwrite(packfile, m->prev->id.sha1,
1264 				    sizeof(m->prev->id.sha1), &ctx);
1265 				packfile_size += sizeof(m->prev->id.sha1);
1266 				if (err)
1267 					goto done;
1268 			}
1269 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1270 				err = got_error_from_errno("fseeko");
1271 				goto done;
1272 			}
1273 			err = got_deflate_to_file(&outlen, delta_file,
1274 			    packfile, &csum);
1275 			if (err)
1276 				goto done;
1277 			packfile_size += outlen;
1278 		}
1279 		got_object_raw_close(raw);
1280 		raw = NULL;
1281 	}
1282 	SHA1Final(pack_sha1, &ctx);
1283 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1284 	if (n != SHA1_DIGEST_LENGTH)
1285 		err = got_ferror(packfile, GOT_ERR_IO);
1286 	packfile_size += SHA1_DIGEST_LENGTH;
1287 	packfile_size += sizeof(struct got_packfile_hdr);
1288 	err = progress_cb(progress_arg, packfile_size, nours,
1289 	    nmeta, nmeta, nmeta);
1290 	if (err)
1291 		goto done;
1292 done:
1293 	if (delta_file && fclose(delta_file) == EOF && err == NULL)
1294 		err = got_error_from_errno("fclose");
1295 	if (raw)
1296 		got_object_raw_close(raw);
1297 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1298 		err = got_error_from_errno("close");
1299 	return err;
1300 }
1301 
1302 const struct got_error *
1303 got_pack_create(uint8_t *packsha1, FILE *packfile,
1304     struct got_object_id **theirs, int ntheirs,
1305     struct got_object_id **ours, int nours,
1306     struct got_repository *repo, int loose_obj_only, int allow_empty,
1307     got_pack_progress_cb progress_cb, void *progress_arg,
1308     got_cancel_cb cancel_cb, void *cancel_arg)
1309 {
1310 	const struct got_error *err;
1311 	struct got_pack_meta **meta;
1312 	int nmeta;
1313 	FILE *delta_cache = NULL;
1314 
1315 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1316 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1317 	if (err)
1318 		return err;
1319 
1320 	if (nmeta == 0 && !allow_empty) {
1321 		err = got_error(GOT_ERR_CANNOT_PACK);
1322 		goto done;
1323 	}
1324 
1325 	delta_cache = got_opentemp();
1326 	if (delta_cache == NULL) {
1327 		err = got_error_from_errno("got_opentemp");
1328 		goto done;
1329 	}
1330 
1331 	if (nmeta > 0) {
1332 		err = pick_deltas(meta, nmeta, nours, delta_cache, repo,
1333 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1334 		if (err)
1335 			goto done;
1336 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1337 			err = got_error_from_errno("fseeko");
1338 			goto done;
1339 		}
1340 	}
1341 
1342 	err = genpack(packsha1, packfile, delta_cache, meta, nmeta, nours, 1,
1343 	    repo, progress_cb, progress_arg, cancel_cb, cancel_arg);
1344 	if (err)
1345 		goto done;
1346 done:
1347 	free_nmeta(meta, nmeta);
1348 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1349 		err = got_error_from_errno("fclose");
1350 	return err;
1351 }