Gotweb

Blob

Date:: Fri Oct 15 19:22:03 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 <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <limits.h>
29 #include <zlib.h>
30 
31 #include "got_error.h"
32 #include "got_cancel.h"
33 #include "got_object.h"
34 #include "got_path.h"
35 #include "got_reference.h"
36 #include "got_repository_admin.h"
37 #include "got_opentemp.h"
38 
39 #include "got_lib_deltify.h"
40 #include "got_lib_delta.h"
41 #include "got_lib_object.h"
42 #include "got_lib_object_idset.h"
43 #include "got_lib_object_cache.h"
44 #include "got_lib_deflate.h"
45 #include "got_lib_pack.h"
46 #include "got_lib_privsep.h"
47 #include "got_lib_repository.h"
48 
49 #ifndef MIN
50 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
51 #endif
52 
53 #ifndef MAX
54 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
55 #endif
56 
57 struct got_pack_meta {
58 	struct got_object_id id;
59 	char	*path;
60 	int	obj_type;
61 	off_t	size;
62 	time_t	mtime;
63 
64 	/* The best delta we picked */
65 	struct got_pack_meta *head;
66 	struct got_pack_meta *prev;
67 	off_t	delta_offset;	/* offset in delta cache file */
68 	off_t	delta_len;	/* length in delta cache file */
69 	int	nchain;
70 
71 	/* Only used for delta window */
72 	struct got_delta_table *dtab;
73 
74 	/* Only used for writing offset deltas */
75 	off_t	off;
76 };
77 
78 struct got_pack_metavec {
79 	struct got_pack_meta **meta;
80 	int nmeta;
81 	int metasz;
82 };
83 
84 static const struct got_error *
85 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
86     const char *path, int obj_type, time_t mtime)
87 {
88 	const struct got_error *err = NULL;
89 	struct got_pack_meta *m;
90 
91 	*new = NULL;
92 
93 	m = calloc(1, sizeof(*m));
94 	if (m == NULL)
95 		return got_error_from_errno("calloc");
96 
97 	memcpy(&m->id, id, sizeof(m->id));
98 
99 	m->path = strdup(path);
100 	if (m->path == NULL) {
101 		err = got_error_from_errno("strdup");
102 		free(m);
103 		return err;
104 	}
105 
106 	m->obj_type = obj_type;
107 	m->mtime = mtime;
108 	*new = m;
109 	return NULL;
110 }
111 
112 static void
113 clear_meta(struct got_pack_meta *meta)
114 {
115 	if (meta == NULL)
116 		return;
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 static const struct got_error *
164 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
165     struct got_delta_instruction *deltas, int ndeltas,
166     off_t base_size, FILE *f);
167 
168 static const struct got_error *
169 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
170     FILE *delta_cache, struct got_repository *repo,
171     got_pack_progress_cb progress_cb, void *progress_arg,
172     got_cancel_cb cancel_cb, void *cancel_arg)
173 {
174 	const struct got_error *err = NULL;
175 	struct got_pack_meta *m = NULL, *base = NULL;
176 	struct got_raw_object *raw = NULL, *base_raw = NULL;
177 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
178 	int i, j, size, best_size, ndeltas, best_ndeltas;
179 	const int max_base_candidates = 10;
180 	int outfd = -1;
181 
182 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
183 	for (i = 0; i < nmeta; i++) {
184 		if (cancel_cb) {
185 			err = (*cancel_cb)(cancel_arg);
186 			if (err)
187 				break;
188 		}
189 		if (progress_cb) {
190 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
191 			if (err)
192 				goto done;
193 		}
194 		m = meta[i];
195 
196 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
197 		    m->obj_type == GOT_OBJ_TYPE_TAG)
198 			continue;
199 
200 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
201 		if (err)
202 			goto done;
203 		m->size = raw->size;
204 
205 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
206 		    raw->size + raw->hdrlen);
207 		if (err)
208 			goto done;
209 
210 		if (i > max_base_candidates) {
211 			struct got_pack_meta *n = NULL;
212 			n = meta[i - (max_base_candidates + 1)];
213 			got_deltify_free(n->dtab);
214 			n->dtab = NULL;
215 		}
216 
217 		best_size = raw->size;
218 		best_ndeltas = 0;
219 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
220 			if (cancel_cb) {
221 				err = (*cancel_cb)(cancel_arg);
222 				if (err)
223 					goto done;
224 			}
225 			base = meta[j];
226 			/* long chains make unpacking slow, avoid such bases */
227 			if (base->nchain >= 32 ||
228 			    base->obj_type != m->obj_type)
229 				continue;
230 
231 			err = got_object_raw_open(&base_raw, &outfd, repo,
232 			    &base->id);
233 			if (err)
234 				goto done;
235 			err = got_deltify(&deltas, &ndeltas,
236 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
237 			    base->dtab, base_raw->f, base_raw->hdrlen,
238 			    base_raw->size + base_raw->hdrlen);
239 			got_object_raw_close(base_raw);
240 			base_raw = NULL;
241 			if (err)
242 				goto done;
243 
244 			size = delta_size(deltas, ndeltas);
245 			if (size + 32 < best_size){
246 				/*
247 				 * if we already picked a best delta,
248 				 * replace it.
249 				 */
250 				best_size = size;
251 				free(best_deltas);
252 				best_deltas = deltas;
253 				best_ndeltas = ndeltas;
254 				deltas = NULL;
255 				m->nchain = base->nchain + 1;
256 				m->prev = base;
257 				m->head = base->head;
258 				if (m->head == NULL)
259 					m->head = base;
260 			} else {
261 				free(deltas);
262 				deltas = NULL;
263 				ndeltas = 0;
264 			}
265 		}
266 
267 		if (best_ndeltas > 0) {
268 			m->delta_offset = ftello(delta_cache);
269 			err = encode_delta(m, raw, best_deltas,
270 			    best_ndeltas, m->prev->size, delta_cache);
271 			free(best_deltas);
272 			best_deltas = NULL;
273 			best_ndeltas = 0;
274 			if (err)
275 				goto done;
276 			m->delta_len = ftello(delta_cache) - m->delta_offset;
277 		}
278 
279 		got_object_raw_close(raw);
280 		raw = NULL;
281 	}
282 done:
283 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
284 		got_deltify_free(meta[i]->dtab);
285 		meta[i]->dtab = NULL;
286 	}
287 	if (raw)
288 		got_object_raw_close(raw);
289 	if (base_raw)
290 		got_object_raw_close(base_raw);
291 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
292 		err = got_error_from_errno("close");
293 	free(deltas);
294 	free(best_deltas);
295 	return err;
296 }
297 
298 static const struct got_error *
299 search_packidx(int *found, struct got_object_id *id,
300     struct got_repository *repo)
301 {
302 	const struct got_error *err = NULL;
303 	struct got_packidx *packidx = NULL;
304 	int idx;
305 
306 	*found = 0;
307 
308 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
309 	if (err == NULL)
310 		*found = 1; /* object is already packed */
311 	else if (err->code == GOT_ERR_NO_OBJ)
312 		err = NULL;
313 	return err;
314 }
315 
316 static const int obj_types[] = {
317 	GOT_OBJ_TYPE_ANY,
318 	GOT_OBJ_TYPE_COMMIT,
319 	GOT_OBJ_TYPE_TREE,
320 	GOT_OBJ_TYPE_BLOB,
321 	GOT_OBJ_TYPE_TAG,
322 	GOT_OBJ_TYPE_OFFSET_DELTA,
323 	GOT_OBJ_TYPE_REF_DELTA
324 };
325 
326 static const struct got_error *
327 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
328     struct got_object_id *id, const char *path, int obj_type,
329     time_t mtime, int loose_obj_only, struct got_repository *repo)
330 {
331 	const struct got_error *err;
332 	struct got_pack_meta *m;
333 
334 	if (loose_obj_only) {
335 		int is_packed;
336 		err = search_packidx(&is_packed, id, repo);
337 		if (err)
338 			return err;
339 		if (is_packed)
340 			return NULL;
341 	}
342 
343 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
344 	if (err)
345 		return err;
346 
347 	if (v == NULL)
348 		return NULL;
349 
350 	err = alloc_meta(&m, id, path, obj_type, mtime);
351 	if (err)
352 		goto done;
353 
354 	if (v->nmeta == v->metasz){
355 		size_t newsize = 2 * v->metasz;
356 		struct got_pack_meta **new;
357 		new = reallocarray(v->meta, newsize, sizeof(*new));
358 		if (new == NULL) {
359 			err = got_error_from_errno("reallocarray");
360 			goto done;
361 		}
362 		v->meta = new;
363 		v->metasz = newsize; 
364 	}
365 done:
366 	if (err) {
367 		clear_meta(m);
368 		free(m);
369 	} else
370 		v->meta[v->nmeta++] = m;
371 
372 	return err;
373 }
374 
375 static const struct got_error *
376 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
377     struct got_object_idset *idset, struct got_object_id *tree_id,
378     const char *dpath, time_t mtime, struct got_repository *repo,
379     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
380 {
381 	const struct got_error *err;
382 	struct got_tree_object *tree;
383 	char *p = NULL;
384 	int i;
385 
386 	err = got_object_open_as_tree(&tree, repo, tree_id);
387 	if (err)
388 		return err;
389 
390 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
391 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
392 		struct got_object_id *id = got_tree_entry_get_id(e);
393 		mode_t mode = got_tree_entry_get_mode(e);
394 
395 		if (cancel_cb) {
396 			err = (*cancel_cb)(cancel_arg);
397 			if (err)
398 				break;
399 		}
400 
401 		if (got_object_tree_entry_is_submodule(e) ||
402 		    got_object_idset_contains(idset, id))
403 			continue;
404 		
405 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
406 		    got_tree_entry_get_name(e)) == -1) {
407 			err = got_error_from_errno("asprintf");
408 			break;
409 		}
410 
411 		if (S_ISDIR(mode)) {
412 			struct got_object_qid *qid;
413 			err = got_object_qid_alloc(&qid, id);
414 			if (err)
415 				break;
416 			STAILQ_INSERT_TAIL(ids, qid, entry);
417 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
418 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
419 			    mtime, loose_obj_only, repo);
420 			if (err)
421 				break;
422 		}
423 		free(p);
424 		p = NULL;
425 	}
426 
427 	got_object_tree_close(tree);
428 	free(p);
429 	return err;
430 }
431 
432 static const struct got_error *
433 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
434     struct got_object_id *tree_id, const char *dpath, time_t mtime,
435     int loose_obj_only, struct got_repository *repo,
436     got_cancel_cb cancel_cb, void *cancel_arg)
437 {
438 	const struct got_error *err = NULL;
439 	struct got_object_id_queue tree_ids;
440 	struct got_object_qid *qid;
441 
442 	if (got_object_idset_contains(idset, tree_id))
443 		return NULL;
444 
445 	err = got_object_qid_alloc(&qid, tree_id);
446 	if (err)
447 		return err;
448 
449 	STAILQ_INIT(&tree_ids);
450 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
451 
452 	while (!STAILQ_EMPTY(&tree_ids)) {
453 		if (cancel_cb) {
454 			err = (*cancel_cb)(cancel_arg);
455 			if (err)
456 				break;
457 		}
458 
459 		qid = STAILQ_FIRST(&tree_ids);
460 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
461 
462 		if (got_object_idset_contains(idset, qid->id)) {
463 			got_object_qid_free(qid);
464 			continue;
465 		}
466 
467 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
468 		    mtime, loose_obj_only, repo);
469 		if (err) {
470 			got_object_qid_free(qid);
471 			break;
472 		}
473 
474 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
475 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
476 		got_object_qid_free(qid);
477 		if (err)
478 			break;
479 	}
480 
481 	got_object_id_queue_free(&tree_ids);
482 	return err;
483 }
484 
485 static const struct got_error *
486 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
487     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
488     got_cancel_cb cancel_cb, void *cancel_arg)
489 {
490 	const struct got_error *err;
491 	struct got_commit_object *commit;
492 
493 	if (got_object_idset_contains(idset, id))
494 		return NULL;
495 
496 	if (loose_obj_only) {
497 		int is_packed;
498 		err = search_packidx(&is_packed, id, repo);
499 		if (err)
500 			return err;
501 		if (is_packed)
502 			return NULL;
503 	}
504 
505 	err = got_object_open_as_commit(&commit, repo, id);
506 	if (err)
507 		return err;
508 
509 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
510 	    got_object_commit_get_committer_time(commit),
511 	    loose_obj_only, repo);
512 	if (err)
513 		goto done;
514 
515 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
516 	    "", got_object_commit_get_committer_time(commit),
517 	    loose_obj_only, repo, cancel_cb, cancel_arg);
518 done:
519 	got_object_commit_close(commit);
520 	return err;
521 }
522 
523 static const struct got_error *
524 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
525     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
526     got_cancel_cb cancel_cb, void *cancel_arg)
527 {
528 	const struct got_error *err;
529 	struct got_tag_object *tag = NULL;
530 
531 	if (got_object_idset_contains(idset, id))
532 		return NULL;
533 
534 	if (loose_obj_only) {
535 		int is_packed;
536 		err = search_packidx(&is_packed, id, repo);
537 		if (err)
538 			return err;
539 		if (is_packed)
540 			return NULL;
541 	}
542 
543 	err = got_object_open_as_tag(&tag, repo, id);
544 	if (err)
545 		return err;
546 
547 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
548 	    got_object_tag_get_tagger_time(tag),
549 	    loose_obj_only, repo);
550 	if (err)
551 		goto done;
552 
553 	switch (got_object_tag_get_object_type(tag)) {
554 	case GOT_OBJ_TYPE_COMMIT:
555 		err = load_commit(v, idset,
556 		    got_object_tag_get_object_id(tag), repo,
557 		    loose_obj_only, cancel_cb, cancel_arg);
558 		break;
559 	case GOT_OBJ_TYPE_TREE:
560 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
561 		    "", got_object_tag_get_tagger_time(tag),
562 		    loose_obj_only, repo, cancel_cb, cancel_arg);
563 		break;
564 	default:
565 		break;
566 	}
567 
568 done:
569 	got_object_tag_close(tag);
570 	return err;
571 }
572 
573 enum findtwixt_color {
574 	COLOR_KEEP = 0,
575 	COLOR_DROP,
576 	COLOR_BLANK,
577 };
578 static const int findtwixt_colors[] = {
579 	COLOR_KEEP,
580 	COLOR_DROP,
581 	COLOR_BLANK
582 };
583 
584 static const struct got_error *
585 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
586     int color, struct got_repository *repo)
587 {
588 	const struct got_error *err;
589 	struct got_object_qid *qid;
590 
591 	err = got_object_qid_alloc(&qid, id);
592 	if (err)
593 		return err;
594 
595 	STAILQ_INSERT_TAIL(ids, qid, entry);
596 	qid->data = (void *)&findtwixt_colors[color];
597 	return NULL;
598 }
599 
600 static const struct got_error *
601 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
602     struct got_object_id *id, struct got_repository *repo,
603     got_cancel_cb cancel_cb, void *cancel_arg)
604 {
605 	const struct got_error *err = NULL;
606 	struct got_commit_object *commit;
607 	const struct got_object_id_queue *parents;
608 	struct got_object_id_queue ids;
609 	struct got_object_qid *qid;
610 
611 	STAILQ_INIT(&ids);
612 
613 	err = got_object_qid_alloc(&qid, id);
614 	if (err)
615 		return err;
616 	STAILQ_INSERT_HEAD(&ids, qid, entry);
617 
618 	while (!STAILQ_EMPTY(&ids)) {
619 		if (cancel_cb) {
620 			err = (*cancel_cb)(cancel_arg);
621 			if (err)
622 				break;
623 		}
624 
625 		qid = STAILQ_FIRST(&ids);
626 		STAILQ_REMOVE_HEAD(&ids, entry);
627 
628 		if (got_object_idset_contains(drop, qid->id)) {
629 			got_object_qid_free(qid);
630 			continue;
631 		}
632 
633 		err = got_object_idset_add(drop, qid->id,
634 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
635 		if (err) {
636 			got_object_qid_free(qid);
637 			break;
638 		}
639 
640 		if (!got_object_idset_contains(keep, qid->id)) {
641 			got_object_qid_free(qid);
642 			continue;
643 		}
644 
645 		err = got_object_open_as_commit(&commit, repo, qid->id);
646 		got_object_qid_free(qid);
647 		if (err)
648 			break;
649 
650 		parents = got_object_commit_get_parent_ids(commit);
651 		if (parents) {
652 			err = got_object_id_queue_copy(parents, &ids);
653 			if (err) {
654 				got_object_commit_close(commit);
655 				break;
656 			}
657 		}
658 		got_object_commit_close(commit);
659 	}
660 
661 	got_object_id_queue_free(&ids);
662 	return err;
663 }
664 
665 struct append_id_arg {
666 	struct got_object_id **array;
667 	int idx;
668 };
669 
670 static const struct got_error *
671 append_id(struct got_object_id *id, void *data, void *arg)
672 {
673 	struct append_id_arg *a = arg;
674 
675 	a->array[a->idx] = got_object_id_dup(id);
676 	if (a->array[a->idx] == NULL)
677 		return got_error_from_errno("got_object_id_dup");
678 
679 	a->idx++;
680 	return NULL;
681 }
682 
683 static const struct got_error *
684 findtwixt(struct got_object_id ***res, int *nres,
685     struct got_object_id **head, int nhead,
686     struct got_object_id **tail, int ntail,
687     struct got_repository *repo,
688     got_cancel_cb cancel_cb, void *cancel_arg)
689 {
690 	const struct got_error *err = NULL;
691 	struct got_object_id_queue ids;
692 	struct got_object_idset *keep, *drop;
693 	struct got_object_qid *qid;
694 	int i, ncolor, nkeep, obj_type;
695 
696 	STAILQ_INIT(&ids);
697 	*res = NULL;
698 	*nres = 0;
699 
700 	keep = got_object_idset_alloc();
701 	if (keep == NULL)
702 		return got_error_from_errno("got_object_idset_alloc");
703 
704 	drop = got_object_idset_alloc();
705 	if (drop == NULL) {
706 		err = got_error_from_errno("got_object_idset_alloc");
707 		goto done;
708 	}
709 
710 	for (i = 0; i < nhead; i++) {
711 		struct got_object_id *id = head[i];
712 		if (id == NULL)
713 			continue;
714 		err = got_object_get_type(&obj_type, repo, id);
715 		if (err)
716 			return err;
717 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
718 			continue;
719 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
720 		if (err)
721 			goto done;
722 	}		
723 	for (i = 0; i < ntail; i++) {
724 		struct got_object_id *id = tail[i];
725 		if (id == NULL)
726 			continue;
727 		err = got_object_get_type(&obj_type, repo, id);
728 		if (err)
729 			return err;
730 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
731 			continue;
732 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
733 		if (err)
734 			goto done;
735 	}
736 
737 	while (!STAILQ_EMPTY(&ids)) {
738 		int qcolor;
739 		qid = STAILQ_FIRST(&ids);
740 		qcolor = *((int *)qid->data);
741 
742 		if (got_object_idset_contains(drop, qid->id))
743 			ncolor = COLOR_DROP;
744 		else if (got_object_idset_contains(keep, qid->id))
745 			ncolor = COLOR_KEEP;
746 		else
747 			ncolor = COLOR_BLANK;
748 
749 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
750 		    qcolor == COLOR_KEEP)) {
751 			STAILQ_REMOVE_HEAD(&ids, entry);
752 			got_object_qid_free(qid);
753 			continue;
754 		}
755 
756 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
757 			err = drop_commit(keep, drop, qid->id, repo,
758 			    cancel_cb, cancel_arg);
759 			if (err)
760 				goto done;
761 		} else if (ncolor == COLOR_BLANK) {
762 			struct got_commit_object *commit;
763 			struct got_object_id *id;
764 			const struct got_object_id_queue *parents;
765 			struct got_object_qid *pid;
766 
767 			id = got_object_id_dup(qid->id);
768 			if (id == NULL) {
769 				err = got_error_from_errno("got_object_id_dup");
770 				goto done;
771 			}
772 			if (qcolor == COLOR_KEEP)
773 				err = got_object_idset_add(keep, id,
774 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
775 			else
776 				err = got_object_idset_add(drop, id,
777 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
778 			if (err) {
779 				free(id);
780 				goto done;
781 			}
782 
783 			err = got_object_open_as_commit(&commit, repo, id);
784 			if (err) {
785 				free(id);
786 				goto done;
787 			}
788 			parents = got_object_commit_get_parent_ids(commit);
789 			if (parents) {
790 				STAILQ_FOREACH(pid, parents, entry) {
791 					err = queue_commit_id(&ids, pid->id,
792 					    qcolor, repo);
793 					if (err) {
794 						free(id);
795 						goto done;
796 					}
797 				}
798 			}
799 			got_object_commit_close(commit);
800 			commit = NULL;
801 		} else {
802 			/* should not happen */
803 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
804 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
805 			goto done;
806 		}
807 
808 		STAILQ_REMOVE_HEAD(&ids, entry);
809 		got_object_qid_free(qid);
810 	}
811 
812 	nkeep = got_object_idset_num_elements(keep);
813 	if (nkeep > 0) {
814 		struct append_id_arg arg;
815 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
816 		if (arg.array == NULL) {
817 			err = got_error_from_errno("calloc");
818 			goto done;
819 		}
820 		arg.idx = 0;
821 		err = got_object_idset_for_each(keep, append_id, &arg);
822 		if (err) {
823 			free(arg.array);
824 			goto done;
825 		}
826 		*res = arg.array;
827 		*nres = nkeep;
828 	}
829 done:
830 	got_object_idset_free(keep);
831 	got_object_idset_free(drop);
832 	got_object_id_queue_free(&ids);
833 	return err;
834 }
835 
836 static const struct got_error *
837 read_meta(struct got_pack_meta ***meta, int *nmeta,
838     struct got_object_id **theirs, int ntheirs,
839     struct got_object_id **ours, int nours, struct got_repository *repo,
840     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
841     got_cancel_cb cancel_cb, void *cancel_arg)
842 {
843 	const struct got_error *err = NULL;
844 	struct got_object_id **ids = NULL;
845 	struct got_object_idset *idset;
846 	int i, nobj = 0, obj_type;
847 	struct got_pack_metavec v;
848 
849 	*meta = NULL;
850 	*nmeta = 0;
851 
852 	idset = got_object_idset_alloc();
853 	if (idset == NULL)
854 		return got_error_from_errno("got_object_idset_alloc");
855 
856 	v.nmeta = 0;
857 	v.metasz = 64;
858 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
859 	if (v.meta == NULL) {
860 		err = got_error_from_errno("calloc");
861 		goto done;
862 	}
863 
864 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
865 	    cancel_cb, cancel_arg);
866 	if (err || nobj == 0)
867 		goto done;
868 
869 	for (i = 0; i < ntheirs; i++) {
870 		struct got_object_id *id = theirs[i];
871 		if (id == NULL)
872 			continue;
873 		err = got_object_get_type(&obj_type, repo, id);
874 		if (err)
875 			return err;
876 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
877 			continue;
878 		err = load_commit(NULL, idset, id, repo,
879 		    loose_obj_only, cancel_cb, cancel_arg);
880 		if (err)
881 			goto done;
882 		if (progress_cb) {
883 			err = progress_cb(progress_arg, 0L, nours,
884 			    v.nmeta, 0, 0);
885 			if (err)
886 				goto done;
887 		}
888 	}
889 
890 	for (i = 0; i < ntheirs; i++) {
891 		struct got_object_id *id = theirs[i];
892 		int *cached_type;
893 		if (id == NULL)
894 			continue;
895 		cached_type = got_object_idset_get(idset, id);
896 		if (cached_type == NULL) {
897 			err = got_object_get_type(&obj_type, repo, id);
898 			if (err)
899 				goto done;
900 		} else
901 			obj_type = *cached_type;
902 		if (obj_type != GOT_OBJ_TYPE_TAG)
903 			continue;
904 		err = load_tag(NULL, idset, id, repo,
905 		    loose_obj_only, cancel_cb, cancel_arg);
906 		if (err)
907 			goto done;
908 		if (progress_cb) {
909 			err = progress_cb(progress_arg, 0L, nours,
910 			    v.nmeta, 0, 0);
911 			if (err)
912 				goto done;
913 		}
914 	}
915 
916 	for (i = 0; i < nobj; i++) {
917 		err = load_commit(&v, idset, ids[i], repo,
918 		    loose_obj_only, cancel_cb, cancel_arg);
919 		if (err)
920 			goto done;
921 		if (progress_cb) {
922 			err = progress_cb(progress_arg, 0L, nours,
923 			    v.nmeta, 0, 0);
924 			if (err)
925 				goto done;
926 		}
927 	}
928 
929 	for (i = 0; i < nours; i++) {
930 		struct got_object_id *id = ours[i];
931 		int *cached_type;
932 		if (id == NULL)
933 			continue;
934 		cached_type = got_object_idset_get(idset, id);
935 		if (cached_type == NULL) {
936 			err = got_object_get_type(&obj_type, repo, id);
937 			if (err)
938 				goto done;
939 		} else
940 			obj_type = *cached_type;
941 		if (obj_type != GOT_OBJ_TYPE_TAG)
942 			continue;
943 		err = load_tag(&v, idset, id, repo,
944 		    loose_obj_only, cancel_cb, cancel_arg);
945 		if (err)
946 			goto done;
947 		if (progress_cb) {
948 			err = progress_cb(progress_arg, 0L, nours,
949 			    v.nmeta, 0, 0);
950 			if (err)
951 				goto done;
952 		}
953 	}
954 
955 done:
956 	for (i = 0; i < nobj; i++) {
957 		free(ids[i]);
958 	}
959 	free(ids);
960 	got_object_idset_free(idset);
961 	if (err == NULL) {
962 		*meta = v.meta;
963 		*nmeta = v.nmeta;
964 	} else
965 		free(v.meta);
966 
967 	return err;
968 }
969 
970 const struct got_error *
971 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
972 {
973 	size_t n;
974 
975 	SHA1Update(ctx, buf, len);
976 	n = fwrite(buf, 1, len, f);
977 	if (n != len)
978 		return got_ferror(f, GOT_ERR_IO);
979 	return NULL;
980 }
981 
982 static void
983 putbe32(char *b, uint32_t n)
984 {
985 	b[0] = n >> 24;
986 	b[1] = n >> 16;
987 	b[2] = n >> 8;
988 	b[3] = n >> 0;
989 }
990 
991 static int
992 write_order_cmp(const void *pa, const void *pb)
993 {
994 	struct got_pack_meta *a, *b, *ahd, *bhd;
995 
996 	a = *(struct got_pack_meta **)pa;
997 	b = *(struct got_pack_meta **)pb;
998 	ahd = (a->head == NULL) ? a : a->head;
999 	bhd = (b->head == NULL) ? b : b->head;
1000 	if (ahd->mtime != bhd->mtime)
1001 		return bhd->mtime - ahd->mtime;
1002 	if (ahd != bhd)
1003 		return (uintptr_t)bhd - (uintptr_t)ahd;
1004 	if (a->nchain != b->nchain)
1005 		return a->nchain - b->nchain;
1006 	return a->mtime - b->mtime;
1007 }
1008 
1009 static const struct got_error *
1010 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1011 {
1012 	size_t i;
1013 
1014 	*hdrlen = 0;
1015 
1016 	hdr[0] = obj_type << 4;
1017 	hdr[0] |= len & 0xf;
1018 	len >>= 4;
1019 	for (i = 1; len != 0; i++){
1020 		if (i >= bufsize)
1021 			return got_error(GOT_ERR_NO_SPACE);
1022 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1023 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1024 		len >>= GOT_DELTA_SIZE_SHIFT;
1025 	}
1026 
1027 	*hdrlen = i;
1028 	return NULL;
1029 }
1030 
1031 static const struct got_error *
1032 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
1033     struct got_delta_instruction *deltas, int ndeltas,
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 < ndeltas; j++) {
1067 		d = &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, FILE *delta_cache,
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 	SHA1_CTX ctx;
1146 	struct got_pack_meta *m;
1147 	struct got_raw_object *raw = NULL;
1148 	FILE *delta_file = NULL;
1149 	char buf[32];
1150 	size_t outlen, n;
1151 	struct got_deflate_checksum csum;
1152 	off_t packfile_size = 0;
1153 	int outfd = -1;
1154 
1155 	SHA1Init(&ctx);
1156 	csum.output_sha1 = &ctx;
1157 	csum.output_crc = NULL;
1158 
1159 	err = hwrite(packfile, "PACK", 4, &ctx);
1160 	if (err)
1161 		return err;
1162 	putbe32(buf, GOT_PACKFILE_VERSION);
1163 	err = hwrite(packfile, buf, 4, &ctx);
1164 	if (err)
1165 		goto done;
1166 	putbe32(buf, nmeta);
1167 	err = hwrite(packfile, buf, 4, &ctx);
1168 	if (err)
1169 		goto done;
1170 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1171 	for (i = 0; i < nmeta; i++) {
1172 		if (progress_cb) {
1173 			err = progress_cb(progress_arg, packfile_size, nours,
1174 			    nmeta, nmeta, i);
1175 			if (err)
1176 				goto done;
1177 		}
1178 		m = meta[i];
1179 		m->off = ftello(packfile);
1180 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
1181 		if (err)
1182 			goto done;
1183 		if (m->delta_len == 0) {
1184 			err = packhdr(&nh, buf, sizeof(buf),
1185 			    m->obj_type, raw->size);
1186 			if (err)
1187 				goto done;
1188 			err = hwrite(packfile, buf, nh, &ctx);
1189 			if (err)
1190 				goto done;
1191 			packfile_size += nh;
1192 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1193 				err = got_error_from_errno("fseeko");
1194 				goto done;
1195 			}
1196 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1197 			    &csum);
1198 			if (err)
1199 				goto done;
1200 			packfile_size += outlen;
1201 		} else {
1202 			off_t remain;
1203 			if (delta_file == NULL) {
1204 				delta_file = got_opentemp();
1205 				if (delta_file == NULL) {
1206 					err = got_error_from_errno(
1207 					    "got_opentemp");
1208 					goto done;
1209 				}
1210 			}
1211 			if (ftruncate(fileno(delta_file), 0L) == -1) {
1212 				err = got_error_from_errno("ftruncate");
1213 				goto done;
1214 			}
1215 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1216 				err = got_error_from_errno("fseeko");
1217 				goto done;
1218 			}
1219 			if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1220 			    == -1) {
1221 				err = got_error_from_errno("fseeko");
1222 				goto done;
1223 			}
1224 			remain = m->delta_len;
1225 			while (remain > 0) {
1226 				char delta_buf[8192];
1227 				size_t r, w, n;
1228 				n = MIN(remain, sizeof(delta_buf));
1229 				r = fread(delta_buf, 1, n, delta_cache);
1230 				if (r != n) {
1231 					err = got_ferror(delta_cache,
1232 					    GOT_ERR_IO);
1233 					goto done;
1234 				}
1235 				w = fwrite(delta_buf, 1, n, delta_file);
1236 				if (w != n) {
1237 					err = got_ferror(delta_file,
1238 					    GOT_ERR_IO);
1239 					goto done;
1240 				}
1241 				remain -= n;
1242 			}
1243 			if (use_offset_deltas && m->prev->off != 0) {
1244 				err = packhdr(&nh, buf, sizeof(buf),
1245 				    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1246 				if (err)
1247 					goto done;
1248 				nh += packoff(buf + nh,
1249 				    m->off - m->prev->off);
1250 				err = hwrite(packfile, buf, nh, &ctx);
1251 				if (err)
1252 					goto done;
1253 				packfile_size += nh;
1254 			} else {
1255 				err = packhdr(&nh, buf, sizeof(buf),
1256 				    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1257 				err = hwrite(packfile, buf, nh, &ctx);
1258 				if (err)
1259 					goto done;
1260 				packfile_size += nh;
1261 				err = hwrite(packfile, m->prev->id.sha1,
1262 				    sizeof(m->prev->id.sha1), &ctx);
1263 				packfile_size += sizeof(m->prev->id.sha1);
1264 				if (err)
1265 					goto done;
1266 			}
1267 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1268 				err = got_error_from_errno("fseeko");
1269 				goto done;
1270 			}
1271 			err = got_deflate_to_file(&outlen, delta_file,
1272 			    packfile, &csum);
1273 			if (err)
1274 				goto done;
1275 			packfile_size += outlen;
1276 		}
1277 		got_object_raw_close(raw);
1278 		raw = NULL;
1279 	}
1280 	SHA1Final(pack_sha1, &ctx);
1281 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1282 	if (n != SHA1_DIGEST_LENGTH)
1283 		err = got_ferror(packfile, GOT_ERR_IO);
1284 	packfile_size += SHA1_DIGEST_LENGTH;
1285 	packfile_size += sizeof(struct got_packfile_hdr);
1286 	err = progress_cb(progress_arg, packfile_size, nours,
1287 	    nmeta, nmeta, nmeta);
1288 	if (err)
1289 		goto done;
1290 done:
1291 	if (delta_file && fclose(delta_file) == EOF && err == NULL)
1292 		err = got_error_from_errno("fclose");
1293 	if (raw)
1294 		got_object_raw_close(raw);
1295 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1296 		err = got_error_from_errno("close");
1297 	return err;
1298 }
1299 
1300 const struct got_error *
1301 got_pack_create(uint8_t *packsha1, FILE *packfile,
1302     struct got_object_id **theirs, int ntheirs,
1303     struct got_object_id **ours, int nours,
1304     struct got_repository *repo, int loose_obj_only, int allow_empty,
1305     got_pack_progress_cb progress_cb, void *progress_arg,
1306     got_cancel_cb cancel_cb, void *cancel_arg)
1307 {
1308 	const struct got_error *err;
1309 	struct got_pack_meta **meta;
1310 	int nmeta;
1311 	FILE *delta_cache = NULL;
1312 
1313 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1314 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1315 	if (err)
1316 		return err;
1317 
1318 	if (nmeta == 0 && !allow_empty) {
1319 		err = got_error(GOT_ERR_CANNOT_PACK);
1320 		goto done;
1321 	}
1322 
1323 	delta_cache = got_opentemp();
1324 	if (delta_cache == NULL) {
1325 		err = got_error_from_errno("got_opentemp");
1326 		goto done;
1327 	}
1328 
1329 	if (nmeta > 0) {
1330 		err = pick_deltas(meta, nmeta, nours, delta_cache, repo,
1331 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1332 		if (err)
1333 			goto done;
1334 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1335 			err = got_error_from_errno("fseeko");
1336 			goto done;
1337 		}
1338 	}
1339 
1340 	err = genpack(packsha1, packfile, delta_cache, meta, nmeta, nours, 1,
1341 	    repo, progress_cb, progress_arg, cancel_cb, cancel_arg);
1342 	if (err)
1343 		goto done;
1344 done:
1345 	free_nmeta(meta, nmeta);
1346 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1347 		err = got_error_from_errno("fclose");
1348 	return err;
1349 }