op public repos

Blob

Date:: Wed Jun 23 20:48:35 2021 UTC
Message:: switch from SIMPLEQ to equivalent STAILQ macros The singly-linked tail queue macros were added to OpenBSD 6.9 and are more widely available on other systems. ok stsp
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/uio.h>
21 #include <sys/stat.h>
22 
23 #include <stdint.h>
24 #include <imsg.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <sha1.h>
29 #include <limits.h>
30 #include <zlib.h>
31 
32 #include "got_error.h"
33 #include "got_cancel.h"
34 #include "got_object.h"
35 #include "got_reference.h"
36 #include "got_repository_admin.h"
37 
38 #include "got_lib_deltify.h"
39 #include "got_lib_delta.h"
40 #include "got_lib_object.h"
41 #include "got_lib_object_idset.h"
42 #include "got_lib_object_cache.h"
43 #include "got_lib_deflate.h"
44 #include "got_lib_pack.h"
45 #include "got_lib_privsep.h"
46 #include "got_lib_repository.h"
47 
48 #ifndef MAX
49 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
50 #endif
51 
52 struct got_pack_meta {
53 	struct got_object_id id;
54 	char	*path;
55 	int	obj_type;
56 	time_t	mtime;
57 
58 	/* The best delta we picked */
59 	struct got_pack_meta *head;
60 	struct got_pack_meta *prev;
61 	struct got_delta_instruction *deltas;
62 	int	ndeltas;
63 	int	nchain;
64 
65 	/* Only used for delta window */
66 	struct got_delta_table *dtab;
67 
68 	/* Only used for writing offset deltas */
69 	off_t	off;
70 };
71 
72 struct got_pack_metavec {
73 	struct got_pack_meta **meta;
74 	int nmeta;
75 	int metasz;
76 };
77 
78 static const struct got_error *
79 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
80     const char *path, int obj_type, time_t mtime)
81 {
82 	const struct got_error *err = NULL;
83 	struct got_pack_meta *m;
84 
85 	*new = NULL;
86 
87 	m = calloc(1, sizeof(*m));
88 	if (m == NULL)
89 		return got_error_from_errno("malloc");
90 
91 	memcpy(&m->id, id, sizeof(m->id));
92 
93 	m->path = strdup(path);
94 	if (m->path == NULL) {
95 		err = got_error_from_errno("strdup");
96 		free(m);
97 		return err;
98 	}
99 
100 	m->obj_type = obj_type;
101 	m->mtime = mtime;
102 	*new = m;
103 	return NULL;
104 }
105 
106 static void
107 clear_meta(struct got_pack_meta *meta)
108 {
109 	if (meta == NULL)
110 		return;
111 	free(meta->deltas);
112 	meta->deltas = NULL;
113 	free(meta->path);
114 	meta->path = NULL;
115 }
116 
117 static void
118 free_nmeta(struct got_pack_meta **meta, int nmeta)
119 {
120 	int i;
121 
122 	for (i = 0; i < nmeta; i++)
123 		clear_meta(meta[i]);
124 	free(meta);
125 }
126 
127 static int
128 delta_order_cmp(const void *pa, const void *pb)
129 {
130 	struct got_pack_meta *a, *b;
131 	int cmp;
132 
133 	a = *(struct got_pack_meta **)pa;
134 	b = *(struct got_pack_meta **)pb;
135 
136 	if (a->obj_type != b->obj_type)
137 		return a->obj_type - b->obj_type;
138 	cmp = strcmp(a->path, b->path);
139 	if (cmp != 0)
140 		return cmp;
141 	if (a->mtime != b->mtime)
142 		return a->mtime - b->mtime;
143 	return got_object_id_cmp(&a->id, &b->id);
144 }
145 
146 static int
147 delta_size(struct got_delta_instruction *deltas, int ndeltas)
148 {
149 	int i, size = 32;
150 	for (i = 0; i < ndeltas; i++) {
151 		if (deltas[i].copy)
152 			size += GOT_DELTA_SIZE_SHIFT;
153 		else
154 			size += deltas[i].len + 1;
155 	}
156 	return size;
157 }
158 
159 
160 static const struct got_error *
161 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
162     struct got_repository *repo,
163     got_pack_progress_cb progress_cb, void *progress_arg,
164     got_cancel_cb cancel_cb, void *cancel_arg)
165 {
166 	const struct got_error *err = NULL;
167 	struct got_pack_meta *m = NULL, *base = NULL;
168 	struct got_raw_object *raw = NULL, *base_raw = NULL;
169 	struct got_delta_instruction *deltas;
170 	int i, j, size, ndeltas, best;
171 	const int max_base_candidates = 10;
172 
173 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
174 	for (i = 0; i < nmeta; i++) {
175 		if (cancel_cb) {
176 			err = (*cancel_cb)(cancel_arg);
177 			if (err)
178 				break;
179 		}
180 		if (progress_cb) {
181 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
182 			if (err)
183 				goto done;
184 		}
185 		m = meta[i];
186 		m->deltas = NULL;
187 		m->ndeltas = 0;
188 
189 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
190 		    m->obj_type == GOT_OBJ_TYPE_TAG)
191 			continue;
192 
193 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
194 		if (err)
195 			goto done;
196 
197 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
198 		    raw->size + raw->hdrlen);
199 		if (err)
200 			goto done;
201 
202 		if (i > max_base_candidates) {
203 			struct got_pack_meta *n = NULL;
204 			n = meta[i - (max_base_candidates + 1)];
205 			got_deltify_free(n->dtab);
206 			n->dtab = NULL;
207 		}
208 
209 		best = raw->size;
210 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
211 			if (cancel_cb) {
212 				err = (*cancel_cb)(cancel_arg);
213 				if (err)
214 					goto done;
215 			}
216 			base = meta[j];
217 			/* long chains make unpacking slow, avoid such bases */
218 			if (base->nchain >= 128 ||
219 			    base->obj_type != m->obj_type)
220 				continue;
221 
222 			err = got_object_raw_open(&base_raw, repo, &base->id,
223 			    8192);
224 			if (err)
225 				goto done;
226 			err = got_deltify(&deltas, &ndeltas,
227 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
228 			    base->dtab, base_raw->f, base_raw->hdrlen,
229 			    base_raw->size + base_raw->hdrlen);
230 			got_object_raw_close(base_raw);
231 			base_raw = NULL;
232 			if (err)
233 				goto done;
234 
235 			size = delta_size(deltas, ndeltas);
236 			if (size + 32 < best){
237 				/*
238 				 * if we already picked a best delta,
239 				 * replace it.
240 				 */
241 				free(m->deltas);
242 				best = size;
243 				m->deltas = deltas;
244 				m->ndeltas = ndeltas;
245 				m->nchain = base->nchain + 1;
246 				m->prev = base;
247 				m->head = base->head;
248 				if (m->head == NULL)
249 					m->head = base;
250 			} else {
251 				free(deltas);
252 				deltas = NULL;
253 				ndeltas = 0;
254 			}
255 		}
256 
257 		got_object_raw_close(raw);
258 		raw = NULL;
259 	}
260 done:
261 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
262 		got_deltify_free(meta[i]->dtab);
263 		meta[i]->dtab = NULL;
264 	}
265 	if (raw)
266 		got_object_raw_close(raw);
267 	if (base_raw)
268 		got_object_raw_close(base_raw);
269 	return err;
270 }
271 
272 static const struct got_error *
273 search_packidx(int *found, struct got_object_id *id,
274     struct got_repository *repo)
275 {
276 	const struct got_error *err = NULL;
277 	struct got_packidx *packidx = NULL;
278 	int idx;
279 
280 	*found = 0;
281 
282 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
283 	if (err == NULL)
284 		*found = 1; /* object is already packed */
285 	else if (err->code == GOT_ERR_NO_OBJ)
286 		err = NULL;
287 	return err;
288 }
289 
290 static const struct got_error *
291 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
292     struct got_object_id *id, const char *path, int obj_type,
293     time_t mtime, int loose_obj_only, struct got_repository *repo)
294 {
295 	const struct got_error *err;
296 	struct got_pack_meta *m;
297 
298 	if (loose_obj_only) {
299 		int is_packed;
300 		err = search_packidx(&is_packed, id, repo);
301 		if (err)
302 			return err;
303 		if (is_packed)
304 			return NULL;
305 	}
306 
307 	err = got_object_idset_add(idset, id, NULL);
308 	if (err)
309 		return err;
310 
311 	if (v == NULL)
312 		return NULL;
313 
314 	err = alloc_meta(&m, id, path, obj_type, mtime);
315 	if (err)
316 		goto done;
317 
318 	if (v->nmeta == v->metasz){
319 		size_t newsize = 2 * v->metasz;
320 		struct got_pack_meta **new;
321 		new = reallocarray(v->meta, newsize, sizeof(*new));
322 		if (new == NULL) {
323 			err = got_error_from_errno("reallocarray");
324 			goto done;
325 		}
326 		v->meta = new;
327 		v->metasz = newsize; 
328 	}
329 done:
330 	if (err) {
331 		clear_meta(m);
332 		free(m);
333 	} else
334 		v->meta[v->nmeta++] = m;
335 
336 	return err;
337 }
338 
339 static const struct got_error *
340 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
341     struct got_object_idset *idset, struct got_object_id *tree_id,
342     const char *dpath, time_t mtime, struct got_repository *repo,
343     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
344 {
345 	const struct got_error *err;
346 	struct got_tree_object *tree;
347 	char *p = NULL;
348 	int i;
349 
350 	err = got_object_open_as_tree(&tree, repo, tree_id);
351 	if (err)
352 		return err;
353 
354 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
355 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
356 		struct got_object_id *id = got_tree_entry_get_id(e);
357 		mode_t mode = got_tree_entry_get_mode(e);
358 
359 		if (cancel_cb) {
360 			err = (*cancel_cb)(cancel_arg);
361 			if (err)
362 				break;
363 		}
364 
365 		if (got_object_tree_entry_is_symlink(e) ||
366 		    got_object_tree_entry_is_submodule(e) ||
367 		    got_object_idset_contains(idset, id))
368 			continue;
369 		
370 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
371 		    got_tree_entry_get_name(e)) == -1) {
372 			err = got_error_from_errno("asprintf");
373 			break;
374 		}
375 
376 		if (S_ISDIR(mode)) {
377 			struct got_object_qid *qid;
378 			err = got_object_qid_alloc(&qid, id);
379 			if (err)
380 				break;
381 			STAILQ_INSERT_TAIL(ids, qid, entry);
382 		} else if (S_ISREG(mode)) {
383 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
384 			    mtime, loose_obj_only, repo);
385 			if (err)
386 				break;
387 		}
388 		free(p);
389 		p = NULL;
390 	}
391 
392 	got_object_tree_close(tree);
393 	free(p);
394 	return err;
395 }
396 
397 static const struct got_error *
398 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
399     struct got_object_id *tree_id, const char *dpath, time_t mtime,
400     int loose_obj_only, struct got_repository *repo,
401     got_cancel_cb cancel_cb, void *cancel_arg)
402 {
403 	const struct got_error *err = NULL;
404 	struct got_object_id_queue tree_ids;
405 	struct got_object_qid *qid;
406 
407 	if (got_object_idset_contains(idset, tree_id))
408 		return NULL;
409 
410 	err = got_object_qid_alloc(&qid, tree_id);
411 	if (err)
412 		return err;
413 
414 	STAILQ_INIT(&tree_ids);
415 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
416 
417 	while (!STAILQ_EMPTY(&tree_ids)) {
418 		if (cancel_cb) {
419 			err = (*cancel_cb)(cancel_arg);
420 			if (err)
421 				break;
422 		}
423 
424 		qid = STAILQ_FIRST(&tree_ids);
425 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
426 
427 		if (got_object_idset_contains(idset, qid->id)) {
428 			got_object_qid_free(qid);
429 			continue;
430 		}
431 
432 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
433 		    mtime, loose_obj_only, repo);
434 		if (err) {
435 			got_object_qid_free(qid);
436 			break;
437 		}
438 
439 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
440 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
441 		got_object_qid_free(qid);
442 		if (err)
443 			break;
444 	}
445 
446 	got_object_id_queue_free(&tree_ids);
447 	return err;
448 }
449 
450 static const struct got_error *
451 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
452     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
453     got_cancel_cb cancel_cb, void *cancel_arg)
454 {
455 	const struct got_error *err;
456 	struct got_commit_object *commit;
457 
458 	if (got_object_idset_contains(idset, id))
459 		return NULL;
460 
461 	if (loose_obj_only) {
462 		int is_packed;
463 		err = search_packidx(&is_packed, id, repo);
464 		if (err)
465 			return err;
466 		if (is_packed)
467 			return NULL;
468 	}
469 
470 	err = got_object_open_as_commit(&commit, repo, id);
471 	if (err)
472 		return err;
473 
474 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
475 	    got_object_commit_get_committer_time(commit),
476 	    loose_obj_only, repo);
477 	if (err)
478 		goto done;
479 
480 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
481 	    "", got_object_commit_get_committer_time(commit),
482 	    loose_obj_only, repo, cancel_cb, cancel_arg);
483 done:
484 	got_object_commit_close(commit);
485 	return err;
486 }
487 
488 static const struct got_error *
489 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
490     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
491     got_cancel_cb cancel_cb, void *cancel_arg)
492 {
493 	const struct got_error *err;
494 	struct got_tag_object *tag = NULL;
495 
496 	if (got_object_idset_contains(idset, id))
497 		return NULL;
498 
499 	if (loose_obj_only) {
500 		int is_packed;
501 		err = search_packidx(&is_packed, id, repo);
502 		if (err)
503 			return err;
504 		if (is_packed)
505 			return NULL;
506 	}
507 
508 	err = got_object_open_as_tag(&tag, repo, id);
509 	if (err)
510 		return err;
511 
512 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
513 	    got_object_tag_get_tagger_time(tag),
514 	    loose_obj_only, repo);
515 	if (err)
516 		goto done;
517 
518 	switch (got_object_tag_get_object_type(tag)) {
519 	case GOT_OBJ_TYPE_COMMIT:
520 		err = load_commit(NULL, idset,
521 		    got_object_tag_get_object_id(tag), repo,
522 		    loose_obj_only, cancel_cb, cancel_arg);
523 		break;
524 	case GOT_OBJ_TYPE_TREE:
525 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
526 		    "", got_object_tag_get_tagger_time(tag),
527 		    loose_obj_only, repo, cancel_cb, cancel_arg);
528 		break;
529 	default:
530 		break;
531 	}
532 
533 done:
534 	got_object_tag_close(tag);
535 	return err;
536 }
537 
538 enum findtwixt_color {
539 	COLOR_KEEP = 0,
540 	COLOR_DROP,
541 	COLOR_BLANK,
542 };
543 static const int findtwixt_colors[] = {
544 	COLOR_KEEP,
545 	COLOR_DROP,
546 	COLOR_BLANK
547 };
548 
549 static const struct got_error *
550 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
551     int color, struct got_repository *repo)
552 {
553 	const struct got_error *err;
554 	struct got_object_qid *qid;
555 
556 	err = got_object_qid_alloc(&qid, id);
557 	if (err)
558 		return err;
559 
560 	STAILQ_INSERT_TAIL(ids, qid, entry);
561 	qid->data = (void *)&findtwixt_colors[color];
562 	return NULL;
563 }
564 
565 static const struct got_error *
566 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
567     struct got_object_id *id, struct got_repository *repo,
568     got_cancel_cb cancel_cb, void *cancel_arg)
569 {
570 	const struct got_error *err = NULL;
571 	struct got_commit_object *commit;
572 	const struct got_object_id_queue *parents;
573 	struct got_object_id_queue ids;
574 	struct got_object_qid *qid;
575 
576 	STAILQ_INIT(&ids);
577 
578 	err = got_object_qid_alloc(&qid, id);
579 	if (err)
580 		return err;
581 	STAILQ_INSERT_HEAD(&ids, qid, entry);
582 
583 	while (!STAILQ_EMPTY(&ids)) {
584 		if (cancel_cb) {
585 			err = (*cancel_cb)(cancel_arg);
586 			if (err)
587 				break;
588 		}
589 
590 		qid = STAILQ_FIRST(&ids);
591 		STAILQ_REMOVE_HEAD(&ids, entry);
592 
593 		if (got_object_idset_contains(drop, qid->id)) {
594 			got_object_qid_free(qid);
595 			continue;
596 		}
597 
598 		err = got_object_idset_add(drop, qid->id, NULL);
599 		if (err) {
600 			got_object_qid_free(qid);
601 			break;
602 		}
603 
604 		if (!got_object_idset_contains(keep, qid->id)) {
605 			got_object_qid_free(qid);
606 			continue;
607 		}
608 
609 		err = got_object_open_as_commit(&commit, repo, qid->id);
610 		got_object_qid_free(qid);
611 		if (err)
612 			break;
613 
614 		parents = got_object_commit_get_parent_ids(commit);
615 		if (parents) {
616 			err = got_object_id_queue_copy(parents, &ids);
617 			if (err) {
618 				got_object_commit_close(commit);
619 				break;
620 			}
621 		}
622 		got_object_commit_close(commit);
623 	}
624 
625 	got_object_id_queue_free(&ids);
626 	return err;
627 }
628 
629 struct append_id_arg {
630 	struct got_object_id **array;
631 	int idx;
632 };
633 
634 static const struct got_error *
635 append_id(struct got_object_id *id, void *data, void *arg)
636 {
637 	struct append_id_arg *a = arg;
638 
639 	a->array[a->idx] = got_object_id_dup(id);
640 	if (a->array[a->idx] == NULL)
641 		return got_error_from_errno("got_object_id_dup");
642 
643 	a->idx++;
644 	return NULL;
645 }
646 
647 static const struct got_error *
648 findtwixt(struct got_object_id ***res, int *nres,
649     struct got_object_id **head, int nhead,
650     struct got_object_id **tail, int ntail,
651     struct got_repository *repo,
652     got_cancel_cb cancel_cb, void *cancel_arg)
653 {
654 	const struct got_error *err = NULL;
655 	struct got_object_id_queue ids;
656 	struct got_object_idset *keep, *drop;
657 	struct got_object_qid *qid;
658 	int i, ncolor, nkeep, obj_type;
659 
660 	STAILQ_INIT(&ids);
661 	*res = NULL;
662 	*nres = 0;
663 
664 	keep = got_object_idset_alloc();
665 	if (keep == NULL)
666 		return got_error_from_errno("got_object_idset_alloc");
667 
668 	drop = got_object_idset_alloc();
669 	if (drop == NULL) {
670 		err = got_error_from_errno("got_object_idset_alloc");
671 		goto done;
672 	}
673 
674 	for (i = 0; i < nhead; i++) {
675 		struct got_object_id *id = head[i];
676 		if (id == NULL)
677 			continue;
678 		err = got_object_get_type(&obj_type, repo, id);
679 		if (err)
680 			return err;
681 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
682 			continue;
683 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
684 		if (err)
685 			goto done;
686 	}		
687 	for (i = 0; i < ntail; i++) {
688 		struct got_object_id *id = tail[i];
689 		if (id == NULL)
690 			continue;
691 		err = got_object_get_type(&obj_type, repo, id);
692 		if (err)
693 			return err;
694 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
695 			continue;
696 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
697 		if (err)
698 			goto done;
699 	}
700 
701 	while (!STAILQ_EMPTY(&ids)) {
702 		int qcolor;
703 		qid = STAILQ_FIRST(&ids);
704 		qcolor = *((int *)qid->data);
705 
706 		if (got_object_idset_contains(drop, qid->id))
707 			ncolor = COLOR_DROP;
708 		else if (got_object_idset_contains(keep, qid->id))
709 			ncolor = COLOR_KEEP;
710 		else
711 			ncolor = COLOR_BLANK;
712 
713 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
714 		    qcolor == COLOR_KEEP)) {
715 			STAILQ_REMOVE_HEAD(&ids, entry);
716 			got_object_qid_free(qid);
717 			continue;
718 		}
719 
720 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
721 			err = drop_commit(keep, drop, qid->id, repo,
722 			    cancel_cb, cancel_arg);
723 			if (err)
724 				goto done;
725 		} else if (ncolor == COLOR_BLANK) {
726 			struct got_commit_object *commit;
727 			struct got_object_id *id;
728 			const struct got_object_id_queue *parents;
729 			struct got_object_qid *pid;
730 
731 			id = got_object_id_dup(qid->id);
732 			if (id == NULL) {
733 				err = got_error_from_errno("got_object_id_dup");
734 				goto done;
735 			}
736 			if (qcolor == COLOR_KEEP)
737 				err = got_object_idset_add(keep, id, NULL);
738 			else
739 				err = got_object_idset_add(drop, id, NULL);
740 			if (err) {
741 				free(id);
742 				goto done;
743 			}
744 
745 			err = got_object_open_as_commit(&commit, repo, id);
746 			if (err) {
747 				free(id);
748 				goto done;
749 			}
750 			parents = got_object_commit_get_parent_ids(commit);
751 			if (parents) {
752 				STAILQ_FOREACH(pid, parents, entry) {
753 					err = queue_commit_id(&ids, pid->id,
754 					    qcolor, repo);
755 					if (err) {
756 						free(id);
757 						goto done;
758 					}
759 				}
760 			}
761 			got_object_commit_close(commit);
762 			commit = NULL;
763 		} else {
764 			/* should not happen */
765 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
766 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
767 			goto done;
768 		}
769 
770 		STAILQ_REMOVE_HEAD(&ids, entry);
771 		got_object_qid_free(qid);
772 	}
773 
774 	nkeep = got_object_idset_num_elements(keep);
775 	if (nkeep > 0) {
776 		struct append_id_arg arg;
777 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
778 		if (arg.array == NULL) {
779 			err = got_error_from_errno("calloc");
780 			goto done;
781 		}
782 		arg.idx = 0;
783 		err = got_object_idset_for_each(keep, append_id, &arg);
784 		if (err) {
785 			free(arg.array);
786 			goto done;
787 		}
788 		*res = arg.array;
789 		*nres = nkeep;
790 	}
791 done:
792 	got_object_idset_free(keep);
793 	got_object_idset_free(drop);
794 	got_object_id_queue_free(&ids);
795 	return err;
796 }
797 
798 static const struct got_error *
799 read_meta(struct got_pack_meta ***meta, int *nmeta,
800     struct got_object_id **theirs, int ntheirs,
801     struct got_object_id **ours, int nours, struct got_repository *repo,
802     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
803     got_cancel_cb cancel_cb, void *cancel_arg)
804 {
805 	const struct got_error *err = NULL;
806 	struct got_object_id **ids = NULL;
807 	struct got_object_idset *idset;
808 	int i, nobj = 0, obj_type;
809 	struct got_pack_metavec v;
810 
811 	*meta = NULL;
812 	*nmeta = 0;
813 
814 	idset = got_object_idset_alloc();
815 	if (idset == NULL)
816 		return got_error_from_errno("got_object_idset_alloc");
817 
818 	v.nmeta = 0;
819 	v.metasz = 64;
820 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
821 	if (v.meta == NULL) {
822 		err = got_error_from_errno("reallocarray");
823 		goto done;
824 	}
825 
826 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
827 	    cancel_cb, cancel_arg);
828 	if (err || nobj == 0)
829 		goto done;
830 
831 	for (i = 0; i < ntheirs; i++) {
832 		struct got_object_id *id = theirs[i];
833 		if (id == NULL)
834 			continue;
835 		err = got_object_get_type(&obj_type, repo, id);
836 		if (err)
837 			return err;
838 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
839 			continue;
840 		err = load_commit(NULL, idset, id, repo,
841 		    loose_obj_only, cancel_cb, cancel_arg);
842 		if (err)
843 			goto done;
844 		if (progress_cb) {
845 			err = progress_cb(progress_arg, 0L, nours,
846 			    v.nmeta, 0, 0);
847 			if (err)
848 				goto done;
849 		}
850 	}
851 
852 	for (i = 0; i < nobj; i++) {
853 		err = load_commit(&v, idset, ids[i], repo,
854 		    loose_obj_only, cancel_cb, cancel_arg);
855 		if (err)
856 			goto done;
857 		if (progress_cb) {
858 			err = progress_cb(progress_arg, 0L, nours,
859 			    v.nmeta, 0, 0);
860 			if (err)
861 				goto done;
862 		}
863 	}
864 
865 	for (i = 0; i < ntheirs; i++) {
866 		struct got_object_id *id = ours[i];
867 		if (id == NULL)
868 			continue;
869 		err = got_object_get_type(&obj_type, repo, id);
870 		if (err)
871 			return err;
872 		if (obj_type != GOT_OBJ_TYPE_TAG)
873 			continue;
874 		err = load_tag(NULL, idset, id, repo,
875 		    loose_obj_only, cancel_cb, cancel_arg);
876 		if (err)
877 			goto done;
878 		if (progress_cb) {
879 			err = progress_cb(progress_arg, 0L, nours,
880 			    v.nmeta, 0, 0);
881 			if (err)
882 				goto done;
883 		}
884 	}
885 
886 	for (i = 0; i < nours; i++) {
887 		struct got_object_id *id = ours[i];
888 		if (id == NULL)
889 			continue;
890 		err = got_object_get_type(&obj_type, repo, id);
891 		if (err)
892 			return err;
893 		if (obj_type != GOT_OBJ_TYPE_TAG)
894 			continue;
895 		err = load_tag(&v, idset, id, repo,
896 		    loose_obj_only, cancel_cb, cancel_arg);
897 		if (err)
898 			goto done;
899 		if (progress_cb) {
900 			err = progress_cb(progress_arg, 0L, nours,
901 			    v.nmeta, 0, 0);
902 			if (err)
903 				goto done;
904 		}
905 	}
906 
907 done:
908 	for (i = 0; i < nobj; i++) {
909 		free(ids[i]);
910 	}
911 	free(ids);
912 	got_object_idset_free(idset);
913 	if (err == NULL) {
914 		*meta = v.meta;
915 		*nmeta = v.nmeta;
916 	} else
917 		free(v.meta);
918 
919 	return err;
920 }
921 
922 const struct got_error *
923 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
924 {
925 	size_t n;
926 
927 	SHA1Update(ctx, buf, len);
928 	n = fwrite(buf, 1, len, f);
929 	if (n != len)
930 		return got_ferror(f, GOT_ERR_IO);
931 	return NULL;
932 }
933 
934 static void
935 putbe32(char *b, uint32_t n)
936 {
937 	b[0] = n >> 24;
938 	b[1] = n >> 16;
939 	b[2] = n >> 8;
940 	b[3] = n >> 0;
941 }
942 
943 static int
944 write_order_cmp(const void *pa, const void *pb)
945 {
946 	struct got_pack_meta *a, *b, *ahd, *bhd;
947 
948 	a = *(struct got_pack_meta **)pa;
949 	b = *(struct got_pack_meta **)pb;
950 	ahd = (a->head == NULL) ? a : a->head;
951 	bhd = (b->head == NULL) ? b : b->head;
952 	if (ahd->mtime != bhd->mtime)
953 		return bhd->mtime - ahd->mtime;
954 	if (ahd != bhd)
955 		return (uintptr_t)bhd - (uintptr_t)ahd;
956 	if (a->nchain != b->nchain)
957 		return a->nchain - b->nchain;
958 	return a->mtime - b->mtime;
959 }
960 
961 static const struct got_error *
962 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
963 {
964 	size_t i;
965 
966 	*hdrlen = 0;
967 
968 	hdr[0] = obj_type << 4;
969 	hdr[0] |= len & 0xf;
970 	len >>= 4;
971 	for (i = 1; len != 0; i++){
972 		if (i >= bufsize)
973 			return got_error(GOT_ERR_NO_SPACE);
974 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
975 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
976 		len >>= GOT_DELTA_SIZE_SHIFT;
977 	}
978 
979 	*hdrlen = i;
980 	return NULL;
981 }
982 
983 static const struct got_error *
984 append(char **p, int *len, int *sz, void *seg, int nseg)
985 {
986 	char *n;
987 
988 	if (*len + nseg >= *sz) {
989 		while (*len + nseg >= *sz)
990 			*sz += *sz / 2;
991 		n = realloc(*p, *sz);
992 		if (n == NULL)
993 			return got_error_from_errno("realloc");
994 		*p = n;
995 	}
996 	memcpy(*p + *len, seg, nseg);
997 	*len += nseg;
998 	return NULL;
999 }
1000 
1001 
1002 static const struct got_error *
1003 encodedelta(int *nd, struct got_pack_meta *m, struct got_raw_object *o,
1004     off_t base_size, char **pp)
1005 {
1006 	const struct got_error *err = NULL;
1007 	char *p;
1008 	unsigned char buf[16], *bp;
1009 	int len, sz, i, j;
1010 	off_t n;
1011 	struct got_delta_instruction *d;
1012 
1013 	*pp = NULL;
1014 	*nd = 0;
1015 
1016 	sz = 128;
1017 	len = 0;
1018 	p = malloc(sz);
1019 	if (p == NULL)
1020 		return got_error_from_errno("malloc");
1021 
1022 	/* base object size */
1023 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
1024 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
1025 	for (i = 1; n > 0; i++) {
1026 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1027 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1028 		n >>= GOT_DELTA_SIZE_SHIFT;
1029 	}
1030 	err = append(&p, &len, &sz, buf, i);
1031 	if (err)
1032 		return err;
1033 
1034 	/* target object size */
1035 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
1036 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
1037 	for (i = 1; n > 0; i++) {
1038 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
1039 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
1040 		n >>= GOT_DELTA_SIZE_SHIFT;
1041 	}
1042 	err = append(&p, &len, &sz, buf, i);
1043 	if (err)
1044 		return err;
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 			err = append(&p, &len, &sz, buf, bp - buf);
1070 			if (err)
1071 				return err;
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 				err = append(&p, &len, &sz, buf, 1);
1081 				if (err)
1082 					return err;
1083 				r = fread(content, 1, buf[0], o->f);
1084 				if (r != buf[0])
1085 					return got_ferror(o->f, GOT_ERR_IO);
1086 				err = append(&p, &len, &sz, content, buf[0]);
1087 				if (err)
1088 					return err;
1089 				n += buf[0];
1090 			}
1091 		}
1092 	}
1093 	*pp = p;
1094 	*nd = len;
1095 	return NULL;
1096 }
1097 
1098 static int
1099 packoff(char *hdr, off_t off)
1100 {
1101 	int i, j;
1102 	char rbuf[8];
1103 
1104 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1105 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1106 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1107 		    GOT_DELTA_SIZE_MORE;
1108 	}
1109 
1110 	j = 0;
1111 	while (i > 0)
1112 		hdr[j++] = rbuf[--i];
1113 	return j;
1114 }
1115 
1116 static const struct got_error *
1117 genpack(uint8_t *pack_sha1, FILE *packfile,
1118     struct got_pack_meta **meta, int nmeta, int nours,
1119     int use_offset_deltas, struct got_repository *repo,
1120     got_pack_progress_cb progress_cb, void *progress_arg,
1121     got_cancel_cb cancel_cb, void *cancel_arg)
1122 {
1123 	const struct got_error *err = NULL;
1124 	int i, nh, nd;
1125 	SHA1_CTX ctx;
1126 	struct got_pack_meta *m;
1127 	struct got_raw_object *raw;
1128 	char *p = NULL, buf[32];
1129 	size_t outlen, n;
1130 	struct got_deflate_checksum csum;
1131 	off_t packfile_size = 0;
1132 
1133 	SHA1Init(&ctx);
1134 	csum.output_sha1 = &ctx;
1135 	csum.output_crc = NULL;
1136 
1137 	err = hwrite(packfile, "PACK", 4, &ctx);
1138 	if (err)
1139 		return err;
1140 	putbe32(buf, GOT_PACKFILE_VERSION);
1141 	err = hwrite(packfile, buf, 4, &ctx);
1142 	if (err)
1143 		goto done;
1144 	putbe32(buf, nmeta);
1145 	err = hwrite(packfile, buf, 4, &ctx);
1146 	if (err)
1147 		goto done;
1148 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1149 	for (i = 0; i < nmeta; i++) {
1150 		if (progress_cb) {
1151 			err = progress_cb(progress_arg, packfile_size, nours,
1152 			    nmeta, nmeta, i);
1153 			if (err)
1154 				goto done;
1155 		}
1156 		m = meta[i];
1157 		m->off = ftello(packfile);
1158 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
1159 		if (err)
1160 			goto done;
1161 		if (m->deltas == NULL) {
1162 			err = packhdr(&nh, buf, sizeof(buf),
1163 			    m->obj_type, raw->size);
1164 			if (err)
1165 				goto done;
1166 			err = hwrite(packfile, buf, nh, &ctx);
1167 			if (err)
1168 				goto done;
1169 			packfile_size += nh;
1170 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1171 				err = got_error_from_errno("fseeko");
1172 				goto done;
1173 			}
1174 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1175 			    &csum);
1176 			if (err)
1177 				goto done;
1178 			packfile_size += outlen;
1179 		} else {
1180 			FILE *delta_file;
1181 			struct got_raw_object *base_raw;
1182 			err = got_object_raw_open(&base_raw, repo,
1183 			    &m->prev->id, 8192);
1184 			if (err)
1185 				goto done;
1186 			err = encodedelta(&nd, m, raw, base_raw->size, &p);
1187 			if (err)
1188 				goto done;
1189 			got_object_raw_close(base_raw);
1190 			if (use_offset_deltas && m->prev->off != 0) {
1191 				err = packhdr(&nh, buf, sizeof(buf),
1192 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1193 				if (err)
1194 					goto done;
1195 				nh += packoff(buf + nh,
1196 				    m->off - m->prev->off);
1197 				err = hwrite(packfile, buf, nh, &ctx);
1198 				if (err)
1199 					goto done;
1200 				packfile_size += nh;
1201 			} else {
1202 				err = packhdr(&nh, buf, sizeof(buf),
1203 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1204 				err = hwrite(packfile, buf, nh, &ctx);
1205 				if (err)
1206 					goto done;
1207 				packfile_size += nh;
1208 				err = hwrite(packfile, m->prev->id.sha1,
1209 				    sizeof(m->prev->id.sha1), &ctx);
1210 				packfile_size += sizeof(m->prev->id.sha1);
1211 				if (err)
1212 					goto done;
1213 			}
1214 			/* XXX need got_deflate_from_mem() */
1215 			delta_file = fmemopen(p, nd, "r");
1216 			if (delta_file == NULL) {
1217 				err = got_error_from_errno("fmemopen");
1218 				goto done;
1219 			}
1220 			err = got_deflate_to_file(&outlen, delta_file,
1221 			    packfile, &csum);
1222 			fclose(delta_file);
1223 			if (err)
1224 				goto done;
1225 			packfile_size += outlen;
1226 			free(p);
1227 			p = NULL;
1228 		}
1229 		got_object_raw_close(raw);
1230 		raw = NULL;
1231 	}
1232 	SHA1Final(pack_sha1, &ctx);
1233 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1234 	if (n != SHA1_DIGEST_LENGTH)
1235 		err = got_ferror(packfile, GOT_ERR_IO);
1236 	packfile_size += SHA1_DIGEST_LENGTH;
1237 	packfile_size += 16; /* pack file header */
1238 	err = progress_cb(progress_arg, packfile_size, nours,
1239 	    nmeta, nmeta, nmeta);
1240 	if (err)
1241 		goto done;
1242 done:
1243 	free(p);
1244 	return err;
1245 }
1246 
1247 const struct got_error *
1248 got_pack_create(uint8_t *packsha1, FILE *packfile,
1249     struct got_object_id **theirs, int ntheirs,
1250     struct got_object_id **ours, int nours,
1251     struct got_repository *repo, int loose_obj_only,
1252     got_pack_progress_cb progress_cb, void *progress_arg,
1253     got_cancel_cb cancel_cb, void *cancel_arg)
1254 {
1255 	const struct got_error *err;
1256 	struct got_pack_meta **meta;
1257 	int nmeta;
1258 
1259 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1260 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1261 	if (err)
1262 		return err;
1263 
1264 	if (nmeta == 0) {
1265 		err = got_error(GOT_ERR_CANNOT_PACK);
1266 		goto done;
1267 	}
1268 
1269 	err = pick_deltas(meta, nmeta, nours, repo,
1270 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1271 	if (err)
1272 		goto done;
1273 
1274 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1275 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1276 	if (err)
1277 		goto done;
1278 done:
1279 	free_nmeta(meta, nmeta);
1280 	return err;
1281 }