Gotweb

Blob

Date:: Fri Oct 15 19:22:03 2021 UTC
Message:: cache raw objects in order to speed up gotadmin pack
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 MAX
50 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
51 #endif
52 
53 struct got_pack_meta {
54 	struct got_object_id id;
55 	char	*path;
56 	int	obj_type;
57 	off_t	size;
58 	time_t	mtime;
59 
60 	/* The best delta we picked */
61 	struct got_pack_meta *head;
62 	struct got_pack_meta *prev;
63 	struct got_delta_instruction *deltas;
64 	int	ndeltas;
65 	int	nchain;
66 
67 	/* Only used for delta window */
68 	struct got_delta_table *dtab;
69 
70 	/* Only used for writing offset deltas */
71 	off_t	off;
72 };
73 
74 struct got_pack_metavec {
75 	struct got_pack_meta **meta;
76 	int nmeta;
77 	int metasz;
78 };
79 
80 static const struct got_error *
81 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
82     const char *path, int obj_type, time_t mtime)
83 {
84 	const struct got_error *err = NULL;
85 	struct got_pack_meta *m;
86 
87 	*new = NULL;
88 
89 	m = calloc(1, sizeof(*m));
90 	if (m == NULL)
91 		return got_error_from_errno("calloc");
92 
93 	memcpy(&m->id, id, sizeof(m->id));
94 
95 	m->path = strdup(path);
96 	if (m->path == NULL) {
97 		err = got_error_from_errno("strdup");
98 		free(m);
99 		return err;
100 	}
101 
102 	m->obj_type = obj_type;
103 	m->mtime = mtime;
104 	*new = m;
105 	return NULL;
106 }
107 
108 static void
109 clear_meta(struct got_pack_meta *meta)
110 {
111 	if (meta == NULL)
112 		return;
113 	free(meta->deltas);
114 	meta->deltas = NULL;
115 	free(meta->path);
116 	meta->path = NULL;
117 }
118 
119 static void
120 free_nmeta(struct got_pack_meta **meta, int nmeta)
121 {
122 	int i;
123 
124 	for (i = 0; i < nmeta; i++)
125 		clear_meta(meta[i]);
126 	free(meta);
127 }
128 
129 static int
130 delta_order_cmp(const void *pa, const void *pb)
131 {
132 	struct got_pack_meta *a, *b;
133 	int cmp;
134 
135 	a = *(struct got_pack_meta **)pa;
136 	b = *(struct got_pack_meta **)pb;
137 
138 	if (a->obj_type != b->obj_type)
139 		return a->obj_type - b->obj_type;
140 	cmp = strcmp(a->path, b->path);
141 	if (cmp != 0)
142 		return cmp;
143 	if (a->mtime != b->mtime)
144 		return a->mtime - b->mtime;
145 	return got_object_id_cmp(&a->id, &b->id);
146 }
147 
148 static int
149 delta_size(struct got_delta_instruction *deltas, int ndeltas)
150 {
151 	int i, size = 32;
152 	for (i = 0; i < ndeltas; i++) {
153 		if (deltas[i].copy)
154 			size += GOT_DELTA_SIZE_SHIFT;
155 		else
156 			size += deltas[i].len + 1;
157 	}
158 	return size;
159 }
160 
161 
162 static const struct got_error *
163 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
164     struct got_repository *repo,
165     got_pack_progress_cb progress_cb, void *progress_arg,
166     got_cancel_cb cancel_cb, void *cancel_arg)
167 {
168 	const struct got_error *err = NULL;
169 	struct got_pack_meta *m = NULL, *base = NULL;
170 	struct got_raw_object *raw = NULL, *base_raw = NULL;
171 	struct got_delta_instruction *deltas;
172 	int i, j, size, ndeltas, best;
173 	const int max_base_candidates = 10;
174 	int outfd = -1;
175 
176 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
177 	for (i = 0; i < nmeta; i++) {
178 		if (cancel_cb) {
179 			err = (*cancel_cb)(cancel_arg);
180 			if (err)
181 				break;
182 		}
183 		if (progress_cb) {
184 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
185 			if (err)
186 				goto done;
187 		}
188 		m = meta[i];
189 		m->deltas = NULL;
190 		m->ndeltas = 0;
191 
192 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
193 		    m->obj_type == GOT_OBJ_TYPE_TAG)
194 			continue;
195 
196 		err = got_object_raw_open(&raw, &outfd, repo, &m->id, 8192);
197 		if (err)
198 			goto done;
199 		m->size = raw->size;
200 
201 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
202 		    raw->size + raw->hdrlen);
203 		if (err)
204 			goto done;
205 
206 		if (i > max_base_candidates) {
207 			struct got_pack_meta *n = NULL;
208 			n = meta[i - (max_base_candidates + 1)];
209 			got_deltify_free(n->dtab);
210 			n->dtab = NULL;
211 		}
212 
213 		best = raw->size;
214 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
215 			if (cancel_cb) {
216 				err = (*cancel_cb)(cancel_arg);
217 				if (err)
218 					goto done;
219 			}
220 			base = meta[j];
221 			/* long chains make unpacking slow, avoid such bases */
222 			if (base->nchain >= 128 ||
223 			    base->obj_type != m->obj_type)
224 				continue;
225 
226 			err = got_object_raw_open(&base_raw, &outfd, repo,
227 			    &base->id, 8192);
228 			if (err)
229 				goto done;
230 			err = got_deltify(&deltas, &ndeltas,
231 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
232 			    base->dtab, base_raw->f, base_raw->hdrlen,
233 			    base_raw->size + base_raw->hdrlen);
234 			got_object_raw_close(base_raw);
235 			base_raw = NULL;
236 			if (err)
237 				goto done;
238 
239 			size = delta_size(deltas, ndeltas);
240 			if (size + 32 < best){
241 				/*
242 				 * if we already picked a best delta,
243 				 * replace it.
244 				 */
245 				free(m->deltas);
246 				best = size;
247 				m->deltas = deltas;
248 				m->ndeltas = ndeltas;
249 				m->nchain = base->nchain + 1;
250 				m->prev = base;
251 				m->head = base->head;
252 				if (m->head == NULL)
253 					m->head = base;
254 			} else {
255 				free(deltas);
256 				deltas = NULL;
257 				ndeltas = 0;
258 			}
259 		}
260 
261 		got_object_raw_close(raw);
262 		raw = NULL;
263 	}
264 done:
265 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
266 		got_deltify_free(meta[i]->dtab);
267 		meta[i]->dtab = NULL;
268 	}
269 	if (raw)
270 		got_object_raw_close(raw);
271 	if (base_raw)
272 		got_object_raw_close(base_raw);
273 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
274 		err = got_error_from_errno("close");
275 	return err;
276 }
277 
278 static const struct got_error *
279 search_packidx(int *found, struct got_object_id *id,
280     struct got_repository *repo)
281 {
282 	const struct got_error *err = NULL;
283 	struct got_packidx *packidx = NULL;
284 	int idx;
285 
286 	*found = 0;
287 
288 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
289 	if (err == NULL)
290 		*found = 1; /* object is already packed */
291 	else if (err->code == GOT_ERR_NO_OBJ)
292 		err = NULL;
293 	return err;
294 }
295 
296 static const int obj_types[] = {
297 	GOT_OBJ_TYPE_ANY,
298 	GOT_OBJ_TYPE_COMMIT,
299 	GOT_OBJ_TYPE_TREE,
300 	GOT_OBJ_TYPE_BLOB,
301 	GOT_OBJ_TYPE_TAG,
302 	GOT_OBJ_TYPE_OFFSET_DELTA,
303 	GOT_OBJ_TYPE_REF_DELTA
304 };
305 
306 static const struct got_error *
307 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
308     struct got_object_id *id, const char *path, int obj_type,
309     time_t mtime, int loose_obj_only, struct got_repository *repo)
310 {
311 	const struct got_error *err;
312 	struct got_pack_meta *m;
313 
314 	if (loose_obj_only) {
315 		int is_packed;
316 		err = search_packidx(&is_packed, id, repo);
317 		if (err)
318 			return err;
319 		if (is_packed)
320 			return NULL;
321 	}
322 
323 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
324 	if (err)
325 		return err;
326 
327 	if (v == NULL)
328 		return NULL;
329 
330 	err = alloc_meta(&m, id, path, obj_type, mtime);
331 	if (err)
332 		goto done;
333 
334 	if (v->nmeta == v->metasz){
335 		size_t newsize = 2 * v->metasz;
336 		struct got_pack_meta **new;
337 		new = reallocarray(v->meta, newsize, sizeof(*new));
338 		if (new == NULL) {
339 			err = got_error_from_errno("reallocarray");
340 			goto done;
341 		}
342 		v->meta = new;
343 		v->metasz = newsize; 
344 	}
345 done:
346 	if (err) {
347 		clear_meta(m);
348 		free(m);
349 	} else
350 		v->meta[v->nmeta++] = m;
351 
352 	return err;
353 }
354 
355 static const struct got_error *
356 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
357     struct got_object_idset *idset, struct got_object_id *tree_id,
358     const char *dpath, time_t mtime, struct got_repository *repo,
359     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
360 {
361 	const struct got_error *err;
362 	struct got_tree_object *tree;
363 	char *p = NULL;
364 	int i;
365 
366 	err = got_object_open_as_tree(&tree, repo, tree_id);
367 	if (err)
368 		return err;
369 
370 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
371 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
372 		struct got_object_id *id = got_tree_entry_get_id(e);
373 		mode_t mode = got_tree_entry_get_mode(e);
374 
375 		if (cancel_cb) {
376 			err = (*cancel_cb)(cancel_arg);
377 			if (err)
378 				break;
379 		}
380 
381 		if (got_object_tree_entry_is_submodule(e) ||
382 		    got_object_idset_contains(idset, id))
383 			continue;
384 		
385 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
386 		    got_tree_entry_get_name(e)) == -1) {
387 			err = got_error_from_errno("asprintf");
388 			break;
389 		}
390 
391 		if (S_ISDIR(mode)) {
392 			struct got_object_qid *qid;
393 			err = got_object_qid_alloc(&qid, id);
394 			if (err)
395 				break;
396 			STAILQ_INSERT_TAIL(ids, qid, entry);
397 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
398 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
399 			    mtime, loose_obj_only, repo);
400 			if (err)
401 				break;
402 		}
403 		free(p);
404 		p = NULL;
405 	}
406 
407 	got_object_tree_close(tree);
408 	free(p);
409 	return err;
410 }
411 
412 static const struct got_error *
413 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
414     struct got_object_id *tree_id, const char *dpath, time_t mtime,
415     int loose_obj_only, struct got_repository *repo,
416     got_cancel_cb cancel_cb, void *cancel_arg)
417 {
418 	const struct got_error *err = NULL;
419 	struct got_object_id_queue tree_ids;
420 	struct got_object_qid *qid;
421 
422 	if (got_object_idset_contains(idset, tree_id))
423 		return NULL;
424 
425 	err = got_object_qid_alloc(&qid, tree_id);
426 	if (err)
427 		return err;
428 
429 	STAILQ_INIT(&tree_ids);
430 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
431 
432 	while (!STAILQ_EMPTY(&tree_ids)) {
433 		if (cancel_cb) {
434 			err = (*cancel_cb)(cancel_arg);
435 			if (err)
436 				break;
437 		}
438 
439 		qid = STAILQ_FIRST(&tree_ids);
440 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
441 
442 		if (got_object_idset_contains(idset, qid->id)) {
443 			got_object_qid_free(qid);
444 			continue;
445 		}
446 
447 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
448 		    mtime, loose_obj_only, repo);
449 		if (err) {
450 			got_object_qid_free(qid);
451 			break;
452 		}
453 
454 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
455 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
456 		got_object_qid_free(qid);
457 		if (err)
458 			break;
459 	}
460 
461 	got_object_id_queue_free(&tree_ids);
462 	return err;
463 }
464 
465 static const struct got_error *
466 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
467     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
468     got_cancel_cb cancel_cb, void *cancel_arg)
469 {
470 	const struct got_error *err;
471 	struct got_commit_object *commit;
472 
473 	if (got_object_idset_contains(idset, id))
474 		return NULL;
475 
476 	if (loose_obj_only) {
477 		int is_packed;
478 		err = search_packidx(&is_packed, id, repo);
479 		if (err)
480 			return err;
481 		if (is_packed)
482 			return NULL;
483 	}
484 
485 	err = got_object_open_as_commit(&commit, repo, id);
486 	if (err)
487 		return err;
488 
489 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
490 	    got_object_commit_get_committer_time(commit),
491 	    loose_obj_only, repo);
492 	if (err)
493 		goto done;
494 
495 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
496 	    "", got_object_commit_get_committer_time(commit),
497 	    loose_obj_only, repo, cancel_cb, cancel_arg);
498 done:
499 	got_object_commit_close(commit);
500 	return err;
501 }
502 
503 static const struct got_error *
504 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
505     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
506     got_cancel_cb cancel_cb, void *cancel_arg)
507 {
508 	const struct got_error *err;
509 	struct got_tag_object *tag = NULL;
510 
511 	if (got_object_idset_contains(idset, id))
512 		return NULL;
513 
514 	if (loose_obj_only) {
515 		int is_packed;
516 		err = search_packidx(&is_packed, id, repo);
517 		if (err)
518 			return err;
519 		if (is_packed)
520 			return NULL;
521 	}
522 
523 	err = got_object_open_as_tag(&tag, repo, id);
524 	if (err)
525 		return err;
526 
527 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
528 	    got_object_tag_get_tagger_time(tag),
529 	    loose_obj_only, repo);
530 	if (err)
531 		goto done;
532 
533 	switch (got_object_tag_get_object_type(tag)) {
534 	case GOT_OBJ_TYPE_COMMIT:
535 		err = load_commit(v, idset,
536 		    got_object_tag_get_object_id(tag), repo,
537 		    loose_obj_only, cancel_cb, cancel_arg);
538 		break;
539 	case GOT_OBJ_TYPE_TREE:
540 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
541 		    "", got_object_tag_get_tagger_time(tag),
542 		    loose_obj_only, repo, cancel_cb, cancel_arg);
543 		break;
544 	default:
545 		break;
546 	}
547 
548 done:
549 	got_object_tag_close(tag);
550 	return err;
551 }
552 
553 enum findtwixt_color {
554 	COLOR_KEEP = 0,
555 	COLOR_DROP,
556 	COLOR_BLANK,
557 };
558 static const int findtwixt_colors[] = {
559 	COLOR_KEEP,
560 	COLOR_DROP,
561 	COLOR_BLANK
562 };
563 
564 static const struct got_error *
565 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
566     int color, struct got_repository *repo)
567 {
568 	const struct got_error *err;
569 	struct got_object_qid *qid;
570 
571 	err = got_object_qid_alloc(&qid, id);
572 	if (err)
573 		return err;
574 
575 	STAILQ_INSERT_TAIL(ids, qid, entry);
576 	qid->data = (void *)&findtwixt_colors[color];
577 	return NULL;
578 }
579 
580 static const struct got_error *
581 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
582     struct got_object_id *id, struct got_repository *repo,
583     got_cancel_cb cancel_cb, void *cancel_arg)
584 {
585 	const struct got_error *err = NULL;
586 	struct got_commit_object *commit;
587 	const struct got_object_id_queue *parents;
588 	struct got_object_id_queue ids;
589 	struct got_object_qid *qid;
590 
591 	STAILQ_INIT(&ids);
592 
593 	err = got_object_qid_alloc(&qid, id);
594 	if (err)
595 		return err;
596 	STAILQ_INSERT_HEAD(&ids, qid, entry);
597 
598 	while (!STAILQ_EMPTY(&ids)) {
599 		if (cancel_cb) {
600 			err = (*cancel_cb)(cancel_arg);
601 			if (err)
602 				break;
603 		}
604 
605 		qid = STAILQ_FIRST(&ids);
606 		STAILQ_REMOVE_HEAD(&ids, entry);
607 
608 		if (got_object_idset_contains(drop, qid->id)) {
609 			got_object_qid_free(qid);
610 			continue;
611 		}
612 
613 		err = got_object_idset_add(drop, qid->id,
614 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
615 		if (err) {
616 			got_object_qid_free(qid);
617 			break;
618 		}
619 
620 		if (!got_object_idset_contains(keep, qid->id)) {
621 			got_object_qid_free(qid);
622 			continue;
623 		}
624 
625 		err = got_object_open_as_commit(&commit, repo, qid->id);
626 		got_object_qid_free(qid);
627 		if (err)
628 			break;
629 
630 		parents = got_object_commit_get_parent_ids(commit);
631 		if (parents) {
632 			err = got_object_id_queue_copy(parents, &ids);
633 			if (err) {
634 				got_object_commit_close(commit);
635 				break;
636 			}
637 		}
638 		got_object_commit_close(commit);
639 	}
640 
641 	got_object_id_queue_free(&ids);
642 	return err;
643 }
644 
645 struct append_id_arg {
646 	struct got_object_id **array;
647 	int idx;
648 };
649 
650 static const struct got_error *
651 append_id(struct got_object_id *id, void *data, void *arg)
652 {
653 	struct append_id_arg *a = arg;
654 
655 	a->array[a->idx] = got_object_id_dup(id);
656 	if (a->array[a->idx] == NULL)
657 		return got_error_from_errno("got_object_id_dup");
658 
659 	a->idx++;
660 	return NULL;
661 }
662 
663 static const struct got_error *
664 findtwixt(struct got_object_id ***res, int *nres,
665     struct got_object_id **head, int nhead,
666     struct got_object_id **tail, int ntail,
667     struct got_repository *repo,
668     got_cancel_cb cancel_cb, void *cancel_arg)
669 {
670 	const struct got_error *err = NULL;
671 	struct got_object_id_queue ids;
672 	struct got_object_idset *keep, *drop;
673 	struct got_object_qid *qid;
674 	int i, ncolor, nkeep, obj_type;
675 
676 	STAILQ_INIT(&ids);
677 	*res = NULL;
678 	*nres = 0;
679 
680 	keep = got_object_idset_alloc();
681 	if (keep == NULL)
682 		return got_error_from_errno("got_object_idset_alloc");
683 
684 	drop = got_object_idset_alloc();
685 	if (drop == NULL) {
686 		err = got_error_from_errno("got_object_idset_alloc");
687 		goto done;
688 	}
689 
690 	for (i = 0; i < nhead; i++) {
691 		struct got_object_id *id = head[i];
692 		if (id == NULL)
693 			continue;
694 		err = got_object_get_type(&obj_type, repo, id);
695 		if (err)
696 			return err;
697 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
698 			continue;
699 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
700 		if (err)
701 			goto done;
702 	}		
703 	for (i = 0; i < ntail; i++) {
704 		struct got_object_id *id = tail[i];
705 		if (id == NULL)
706 			continue;
707 		err = got_object_get_type(&obj_type, repo, id);
708 		if (err)
709 			return err;
710 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
711 			continue;
712 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
713 		if (err)
714 			goto done;
715 	}
716 
717 	while (!STAILQ_EMPTY(&ids)) {
718 		int qcolor;
719 		qid = STAILQ_FIRST(&ids);
720 		qcolor = *((int *)qid->data);
721 
722 		if (got_object_idset_contains(drop, qid->id))
723 			ncolor = COLOR_DROP;
724 		else if (got_object_idset_contains(keep, qid->id))
725 			ncolor = COLOR_KEEP;
726 		else
727 			ncolor = COLOR_BLANK;
728 
729 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
730 		    qcolor == COLOR_KEEP)) {
731 			STAILQ_REMOVE_HEAD(&ids, entry);
732 			got_object_qid_free(qid);
733 			continue;
734 		}
735 
736 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
737 			err = drop_commit(keep, drop, qid->id, repo,
738 			    cancel_cb, cancel_arg);
739 			if (err)
740 				goto done;
741 		} else if (ncolor == COLOR_BLANK) {
742 			struct got_commit_object *commit;
743 			struct got_object_id *id;
744 			const struct got_object_id_queue *parents;
745 			struct got_object_qid *pid;
746 
747 			id = got_object_id_dup(qid->id);
748 			if (id == NULL) {
749 				err = got_error_from_errno("got_object_id_dup");
750 				goto done;
751 			}
752 			if (qcolor == COLOR_KEEP)
753 				err = got_object_idset_add(keep, id,
754 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
755 			else
756 				err = got_object_idset_add(drop, id,
757 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
758 			if (err) {
759 				free(id);
760 				goto done;
761 			}
762 
763 			err = got_object_open_as_commit(&commit, repo, id);
764 			if (err) {
765 				free(id);
766 				goto done;
767 			}
768 			parents = got_object_commit_get_parent_ids(commit);
769 			if (parents) {
770 				STAILQ_FOREACH(pid, parents, entry) {
771 					err = queue_commit_id(&ids, pid->id,
772 					    qcolor, repo);
773 					if (err) {
774 						free(id);
775 						goto done;
776 					}
777 				}
778 			}
779 			got_object_commit_close(commit);
780 			commit = NULL;
781 		} else {
782 			/* should not happen */
783 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
784 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
785 			goto done;
786 		}
787 
788 		STAILQ_REMOVE_HEAD(&ids, entry);
789 		got_object_qid_free(qid);
790 	}
791 
792 	nkeep = got_object_idset_num_elements(keep);
793 	if (nkeep > 0) {
794 		struct append_id_arg arg;
795 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
796 		if (arg.array == NULL) {
797 			err = got_error_from_errno("calloc");
798 			goto done;
799 		}
800 		arg.idx = 0;
801 		err = got_object_idset_for_each(keep, append_id, &arg);
802 		if (err) {
803 			free(arg.array);
804 			goto done;
805 		}
806 		*res = arg.array;
807 		*nres = nkeep;
808 	}
809 done:
810 	got_object_idset_free(keep);
811 	got_object_idset_free(drop);
812 	got_object_id_queue_free(&ids);
813 	return err;
814 }
815 
816 static const struct got_error *
817 read_meta(struct got_pack_meta ***meta, int *nmeta,
818     struct got_object_id **theirs, int ntheirs,
819     struct got_object_id **ours, int nours, struct got_repository *repo,
820     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
821     got_cancel_cb cancel_cb, void *cancel_arg)
822 {
823 	const struct got_error *err = NULL;
824 	struct got_object_id **ids = NULL;
825 	struct got_object_idset *idset;
826 	int i, nobj = 0, obj_type;
827 	struct got_pack_metavec v;
828 
829 	*meta = NULL;
830 	*nmeta = 0;
831 
832 	idset = got_object_idset_alloc();
833 	if (idset == NULL)
834 		return got_error_from_errno("got_object_idset_alloc");
835 
836 	v.nmeta = 0;
837 	v.metasz = 64;
838 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
839 	if (v.meta == NULL) {
840 		err = got_error_from_errno("calloc");
841 		goto done;
842 	}
843 
844 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
845 	    cancel_cb, cancel_arg);
846 	if (err || nobj == 0)
847 		goto done;
848 
849 	for (i = 0; i < ntheirs; i++) {
850 		struct got_object_id *id = theirs[i];
851 		if (id == NULL)
852 			continue;
853 		err = got_object_get_type(&obj_type, repo, id);
854 		if (err)
855 			return err;
856 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
857 			continue;
858 		err = load_commit(NULL, idset, id, repo,
859 		    loose_obj_only, cancel_cb, cancel_arg);
860 		if (err)
861 			goto done;
862 		if (progress_cb) {
863 			err = progress_cb(progress_arg, 0L, nours,
864 			    v.nmeta, 0, 0);
865 			if (err)
866 				goto done;
867 		}
868 	}
869 
870 	for (i = 0; i < ntheirs; i++) {
871 		struct got_object_id *id = theirs[i];
872 		int *cached_type;
873 		if (id == NULL)
874 			continue;
875 		cached_type = got_object_idset_get(idset, id);
876 		if (cached_type == NULL) {
877 			err = got_object_get_type(&obj_type, repo, id);
878 			if (err)
879 				goto done;
880 		} else
881 			obj_type = *cached_type;
882 		if (obj_type != GOT_OBJ_TYPE_TAG)
883 			continue;
884 		err = load_tag(NULL, idset, id, repo,
885 		    loose_obj_only, cancel_cb, cancel_arg);
886 		if (err)
887 			goto done;
888 		if (progress_cb) {
889 			err = progress_cb(progress_arg, 0L, nours,
890 			    v.nmeta, 0, 0);
891 			if (err)
892 				goto done;
893 		}
894 	}
895 
896 	for (i = 0; i < nobj; i++) {
897 		err = load_commit(&v, idset, ids[i], repo,
898 		    loose_obj_only, cancel_cb, cancel_arg);
899 		if (err)
900 			goto done;
901 		if (progress_cb) {
902 			err = progress_cb(progress_arg, 0L, nours,
903 			    v.nmeta, 0, 0);
904 			if (err)
905 				goto done;
906 		}
907 	}
908 
909 	for (i = 0; i < nours; i++) {
910 		struct got_object_id *id = ours[i];
911 		int *cached_type;
912 		if (id == NULL)
913 			continue;
914 		cached_type = got_object_idset_get(idset, id);
915 		if (cached_type == NULL) {
916 			err = got_object_get_type(&obj_type, repo, id);
917 			if (err)
918 				goto done;
919 		} else
920 			obj_type = *cached_type;
921 		if (obj_type != GOT_OBJ_TYPE_TAG)
922 			continue;
923 		err = load_tag(&v, idset, id, repo,
924 		    loose_obj_only, cancel_cb, cancel_arg);
925 		if (err)
926 			goto done;
927 		if (progress_cb) {
928 			err = progress_cb(progress_arg, 0L, nours,
929 			    v.nmeta, 0, 0);
930 			if (err)
931 				goto done;
932 		}
933 	}
934 
935 done:
936 	for (i = 0; i < nobj; i++) {
937 		free(ids[i]);
938 	}
939 	free(ids);
940 	got_object_idset_free(idset);
941 	if (err == NULL) {
942 		*meta = v.meta;
943 		*nmeta = v.nmeta;
944 	} else
945 		free(v.meta);
946 
947 	return err;
948 }
949 
950 const struct got_error *
951 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
952 {
953 	size_t n;
954 
955 	SHA1Update(ctx, buf, len);
956 	n = fwrite(buf, 1, len, f);
957 	if (n != len)
958 		return got_ferror(f, GOT_ERR_IO);
959 	return NULL;
960 }
961 
962 static void
963 putbe32(char *b, uint32_t n)
964 {
965 	b[0] = n >> 24;
966 	b[1] = n >> 16;
967 	b[2] = n >> 8;
968 	b[3] = n >> 0;
969 }
970 
971 static int
972 write_order_cmp(const void *pa, const void *pb)
973 {
974 	struct got_pack_meta *a, *b, *ahd, *bhd;
975 
976 	a = *(struct got_pack_meta **)pa;
977 	b = *(struct got_pack_meta **)pb;
978 	ahd = (a->head == NULL) ? a : a->head;
979 	bhd = (b->head == NULL) ? b : b->head;
980 	if (ahd->mtime != bhd->mtime)
981 		return bhd->mtime - ahd->mtime;
982 	if (ahd != bhd)
983 		return (uintptr_t)bhd - (uintptr_t)ahd;
984 	if (a->nchain != b->nchain)
985 		return a->nchain - b->nchain;
986 	return a->mtime - b->mtime;
987 }
988 
989 static const struct got_error *
990 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
991 {
992 	size_t i;
993 
994 	*hdrlen = 0;
995 
996 	hdr[0] = obj_type << 4;
997 	hdr[0] |= len & 0xf;
998 	len >>= 4;
999 	for (i = 1; len != 0; i++){
1000 		if (i >= bufsize)
1001 			return got_error(GOT_ERR_NO_SPACE);
1002 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1003 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1004 		len >>= GOT_DELTA_SIZE_SHIFT;
1005 	}
1006 
1007 	*hdrlen = i;
1008 	return NULL;
1009 }
1010 
1011 static const struct got_error *
1012 encodedelta(struct got_pack_meta *m, struct got_raw_object *o,
1013     off_t base_size, FILE *f)
1014 {
1015 	unsigned char buf[16], *bp;
1016 	int i, j;
1017 	off_t n;
1018 	size_t w;
1019 	struct got_delta_instruction *d;
1020 
1021 	/* base object size */
1022 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1023 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1024 	for (i = 1; n > 0; i++) {
1025 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1026 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1027 		n >>= GOT_DELTA_SIZE_SHIFT;
1028 	}
1029 	w = fwrite(buf, 1, i, f);
1030 	if (w != i)
1031 		return got_ferror(f, GOT_ERR_IO);
1032 
1033 	/* target object size */
1034 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1035 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1036 	for (i = 1; n > 0; i++) {
1037 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1038 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1039 		n >>= GOT_DELTA_SIZE_SHIFT;
1040 	}
1041 	w = fwrite(buf, 1, i, f);
1042 	if (w != i)
1043 		return got_ferror(f, GOT_ERR_IO);
1044 
1045 	for (j = 0; j < m->ndeltas; j++) {
1046 		d = &m->deltas[j];
1047 		if (d->copy) {
1048 			n = d->offset;
1049 			bp = &buf[1];
1050 			buf[0] = GOT_DELTA_BASE_COPY;
1051 			for (i = 0; i < 4; i++) {
1052 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1053 				buf[0] |= 1 << i;
1054 				*bp++ = n & 0xff;
1055 				n >>= 8;
1056 				if (n == 0)
1057 					break;
1058 			}
1059 
1060 			n = d->len;
1061 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1062 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1063 				for (i = 0; i < 3 && n > 0; i++) {
1064 					buf[0] |= 1 << (i + 4);
1065 					*bp++ = n & 0xff;
1066 					n >>= 8;
1067 				}
1068 			}
1069 			w = fwrite(buf, 1, bp - buf, f);
1070 			if (w != bp - buf)
1071 				return got_ferror(f, GOT_ERR_IO);
1072 		} else {
1073 			char content[128];
1074 			size_t r;
1075 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1076 				return got_error_from_errno("fseeko");
1077 			n = 0;
1078 			while (n != d->len) {
1079 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1080 				w = fwrite(buf, 1, 1, f);
1081 				if (w != 1)
1082 					return got_ferror(f, GOT_ERR_IO);
1083 				r = fread(content, 1, buf[0], o->f);
1084 				if (r != buf[0])
1085 					return got_ferror(o->f, GOT_ERR_IO);
1086 				w = fwrite(content, 1, buf[0], f);
1087 				if (w != buf[0])
1088 					return got_ferror(f, GOT_ERR_IO);
1089 				n += buf[0];
1090 			}
1091 		}
1092 	}
1093 
1094 	return NULL;
1095 }
1096 
1097 static int
1098 packoff(char *hdr, off_t off)
1099 {
1100 	int i, j;
1101 	char rbuf[8];
1102 
1103 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1104 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1105 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1106 		    GOT_DELTA_SIZE_MORE;
1107 	}
1108 
1109 	j = 0;
1110 	while (i > 0)
1111 		hdr[j++] = rbuf[--i];
1112 	return j;
1113 }
1114 
1115 static const struct got_error *
1116 genpack(uint8_t *pack_sha1, FILE *packfile,
1117     struct got_pack_meta **meta, int nmeta, int nours,
1118     int use_offset_deltas, struct got_repository *repo,
1119     got_pack_progress_cb progress_cb, void *progress_arg,
1120     got_cancel_cb cancel_cb, void *cancel_arg)
1121 {
1122 	const struct got_error *err = NULL;
1123 	int i, nh;
1124 	off_t nd;
1125 	SHA1_CTX ctx;
1126 	struct got_pack_meta *m;
1127 	struct got_raw_object *raw = NULL;
1128 	FILE *delta_file = NULL;
1129 	char buf[32];
1130 	size_t outlen, n;
1131 	struct got_deflate_checksum csum;
1132 	off_t packfile_size = 0;
1133 	int outfd = -1;
1134 
1135 	SHA1Init(&ctx);
1136 	csum.output_sha1 = &ctx;
1137 	csum.output_crc = NULL;
1138 
1139 	err = hwrite(packfile, "PACK", 4, &ctx);
1140 	if (err)
1141 		return err;
1142 	putbe32(buf, GOT_PACKFILE_VERSION);
1143 	err = hwrite(packfile, buf, 4, &ctx);
1144 	if (err)
1145 		goto done;
1146 	putbe32(buf, nmeta);
1147 	err = hwrite(packfile, buf, 4, &ctx);
1148 	if (err)
1149 		goto done;
1150 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1151 	for (i = 0; i < nmeta; i++) {
1152 		if (progress_cb) {
1153 			err = progress_cb(progress_arg, packfile_size, nours,
1154 			    nmeta, nmeta, i);
1155 			if (err)
1156 				goto done;
1157 		}
1158 		m = meta[i];
1159 		m->off = ftello(packfile);
1160 		err = got_object_raw_open(&raw, &outfd, repo, &m->id, 8192);
1161 		if (err)
1162 			goto done;
1163 		if (m->deltas == NULL) {
1164 			err = packhdr(&nh, buf, sizeof(buf),
1165 			    m->obj_type, raw->size);
1166 			if (err)
1167 				goto done;
1168 			err = hwrite(packfile, buf, nh, &ctx);
1169 			if (err)
1170 				goto done;
1171 			packfile_size += nh;
1172 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1173 				err = got_error_from_errno("fseeko");
1174 				goto done;
1175 			}
1176 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1177 			    &csum);
1178 			if (err)
1179 				goto done;
1180 			packfile_size += outlen;
1181 		} else {
1182 			if (delta_file == NULL) {
1183 				delta_file = got_opentemp();
1184 				if (delta_file == NULL) {
1185 					err = got_error_from_errno(
1186 					    "got_opentemp");
1187 					goto done;
1188 				}
1189 			}
1190 			if (ftruncate(fileno(delta_file), 0L) == -1) {
1191 				err = got_error_from_errno("ftruncate");
1192 				goto done;
1193 			}
1194 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1195 				err = got_error_from_errno("fseeko");
1196 				goto done;
1197 			}
1198 			err = encodedelta(m, raw, m->prev->size, delta_file);
1199 			if (err)
1200 				goto done;
1201 			nd = ftello(delta_file);
1202 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1203 				err = got_error_from_errno("fseeko");
1204 				goto done;
1205 			}
1206 			if (use_offset_deltas && m->prev->off != 0) {
1207 				err = packhdr(&nh, buf, sizeof(buf),
1208 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1209 				if (err)
1210 					goto done;
1211 				nh += packoff(buf + nh,
1212 				    m->off - m->prev->off);
1213 				err = hwrite(packfile, buf, nh, &ctx);
1214 				if (err)
1215 					goto done;
1216 				packfile_size += nh;
1217 			} else {
1218 				err = packhdr(&nh, buf, sizeof(buf),
1219 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1220 				err = hwrite(packfile, buf, nh, &ctx);
1221 				if (err)
1222 					goto done;
1223 				packfile_size += nh;
1224 				err = hwrite(packfile, m->prev->id.sha1,
1225 				    sizeof(m->prev->id.sha1), &ctx);
1226 				packfile_size += sizeof(m->prev->id.sha1);
1227 				if (err)
1228 					goto done;
1229 			}
1230 			err = got_deflate_to_file(&outlen, delta_file,
1231 			    packfile, &csum);
1232 			if (err)
1233 				goto done;
1234 			packfile_size += outlen;
1235 		}
1236 		got_object_raw_close(raw);
1237 		raw = NULL;
1238 	}
1239 	SHA1Final(pack_sha1, &ctx);
1240 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1241 	if (n != SHA1_DIGEST_LENGTH)
1242 		err = got_ferror(packfile, GOT_ERR_IO);
1243 	packfile_size += SHA1_DIGEST_LENGTH;
1244 	packfile_size += sizeof(struct got_packfile_hdr);
1245 	err = progress_cb(progress_arg, packfile_size, nours,
1246 	    nmeta, nmeta, nmeta);
1247 	if (err)
1248 		goto done;
1249 done:
1250 	if (delta_file && fclose(delta_file) == EOF && err == NULL)
1251 		err = got_error_from_errno("fclose");
1252 	if (raw)
1253 		got_object_raw_close(raw);
1254 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1255 		err = got_error_from_errno("close");
1256 	return err;
1257 }
1258 
1259 const struct got_error *
1260 got_pack_create(uint8_t *packsha1, FILE *packfile,
1261     struct got_object_id **theirs, int ntheirs,
1262     struct got_object_id **ours, int nours,
1263     struct got_repository *repo, int loose_obj_only, int allow_empty,
1264     got_pack_progress_cb progress_cb, void *progress_arg,
1265     got_cancel_cb cancel_cb, void *cancel_arg)
1266 {
1267 	const struct got_error *err;
1268 	struct got_pack_meta **meta;
1269 	int nmeta;
1270 
1271 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1272 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1273 	if (err)
1274 		return err;
1275 
1276 	if (nmeta == 0 && !allow_empty) {
1277 		err = got_error(GOT_ERR_CANNOT_PACK);
1278 		goto done;
1279 	}
1280 	if (nmeta > 0) {
1281 		err = pick_deltas(meta, nmeta, nours, repo,
1282 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1283 		if (err)
1284 			goto done;
1285 	}
1286 
1287 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1288 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1289 	if (err)
1290 		goto done;
1291 done:
1292 	free_nmeta(meta, nmeta);
1293 	return err;
1294 }