op public repos

Blob

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