Got Mirror

Blob

Date:: Thu Oct 14 20:30:26 2021 UTC
Message:: avoid opening delta base objects in genpack() just to find their size
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 #include <sys/queue.h>
20 #include <sys/tree.h>
21 #include <sys/uio.h>
22 #include <sys/stat.h>
23 
24 #include <stdint.h>
25 #include <imsg.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <sha1.h>
30 #include <limits.h>
31 #include <zlib.h>
32 
33 #include "got_error.h"
34 #include "got_cancel.h"
35 #include "got_object.h"
36 #include "got_path.h"
37 #include "got_reference.h"
38 #include "got_repository_admin.h"
39 #include "got_opentemp.h"
40 
41 #include "got_lib_deltify.h"
42 #include "got_lib_delta.h"
43 #include "got_lib_object.h"
44 #include "got_lib_object_idset.h"
45 #include "got_lib_object_cache.h"
46 #include "got_lib_deflate.h"
47 #include "got_lib_pack.h"
48 #include "got_lib_privsep.h"
49 #include "got_lib_repository.h"
50 
51 #ifndef MAX
52 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
53 #endif
54 
55 struct got_pack_meta {
56 	struct got_object_id id;
57 	char	*path;
58 	int	obj_type;
59 	off_t	size;
60 	time_t	mtime;
61 
62 	/* The best delta we picked */
63 	struct got_pack_meta *head;
64 	struct got_pack_meta *prev;
65 	struct got_delta_instruction *deltas;
66 	int	ndeltas;
67 	int	nchain;
68 
69 	/* Only used for delta window */
70 	struct got_delta_table *dtab;
71 
72 	/* Only used for writing offset deltas */
73 	off_t	off;
74 };
75 
76 struct got_pack_metavec {
77 	struct got_pack_meta **meta;
78 	int nmeta;
79 	int metasz;
80 };
81 
82 static const struct got_error *
83 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
84     const char *path, int obj_type, time_t mtime)
85 {
86 	const struct got_error *err = NULL;
87 	struct got_pack_meta *m;
88 
89 	*new = NULL;
90 
91 	m = calloc(1, sizeof(*m));
92 	if (m == NULL)
93 		return got_error_from_errno("calloc");
94 
95 	memcpy(&m->id, id, sizeof(m->id));
96 
97 	m->path = strdup(path);
98 	if (m->path == NULL) {
99 		err = got_error_from_errno("strdup");
100 		free(m);
101 		return err;
102 	}
103 
104 	m->obj_type = obj_type;
105 	m->mtime = mtime;
106 	*new = m;
107 	return NULL;
108 }
109 
110 static void
111 clear_meta(struct got_pack_meta *meta)
112 {
113 	if (meta == NULL)
114 		return;
115 	free(meta->deltas);
116 	meta->deltas = NULL;
117 	free(meta->path);
118 	meta->path = NULL;
119 }
120 
121 static void
122 free_nmeta(struct got_pack_meta **meta, int nmeta)
123 {
124 	int i;
125 
126 	for (i = 0; i < nmeta; i++)
127 		clear_meta(meta[i]);
128 	free(meta);
129 }
130 
131 static int
132 delta_order_cmp(const void *pa, const void *pb)
133 {
134 	struct got_pack_meta *a, *b;
135 	int cmp;
136 
137 	a = *(struct got_pack_meta **)pa;
138 	b = *(struct got_pack_meta **)pb;
139 
140 	if (a->obj_type != b->obj_type)
141 		return a->obj_type - b->obj_type;
142 	cmp = strcmp(a->path, b->path);
143 	if (cmp != 0)
144 		return cmp;
145 	if (a->mtime != b->mtime)
146 		return a->mtime - b->mtime;
147 	return got_object_id_cmp(&a->id, &b->id);
148 }
149 
150 static int
151 delta_size(struct got_delta_instruction *deltas, int ndeltas)
152 {
153 	int i, size = 32;
154 	for (i = 0; i < ndeltas; i++) {
155 		if (deltas[i].copy)
156 			size += GOT_DELTA_SIZE_SHIFT;
157 		else
158 			size += deltas[i].len + 1;
159 	}
160 	return size;
161 }
162 
163 
164 static const struct got_error *
165 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
166     struct got_repository *repo,
167     got_pack_progress_cb progress_cb, void *progress_arg,
168     got_cancel_cb cancel_cb, void *cancel_arg)
169 {
170 	const struct got_error *err = NULL;
171 	struct got_pack_meta *m = NULL, *base = NULL;
172 	struct got_raw_object *raw = NULL, *base_raw = NULL;
173 	struct got_delta_instruction *deltas;
174 	int i, j, size, ndeltas, best;
175 	const int max_base_candidates = 10;
176 
177 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
178 	for (i = 0; i < nmeta; i++) {
179 		if (cancel_cb) {
180 			err = (*cancel_cb)(cancel_arg);
181 			if (err)
182 				break;
183 		}
184 		if (progress_cb) {
185 			err = progress_cb(progress_arg, 0L, nours, nmeta, i, 0);
186 			if (err)
187 				goto done;
188 		}
189 		m = meta[i];
190 		m->deltas = NULL;
191 		m->ndeltas = 0;
192 
193 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
194 		    m->obj_type == GOT_OBJ_TYPE_TAG)
195 			continue;
196 
197 		err = got_object_raw_open(&raw, repo, &m->id, 8192);
198 		if (err)
199 			goto done;
200 		m->size = raw->size;
201 
202 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
203 		    raw->size + raw->hdrlen);
204 		if (err)
205 			goto done;
206 
207 		if (i > max_base_candidates) {
208 			struct got_pack_meta *n = NULL;
209 			n = meta[i - (max_base_candidates + 1)];
210 			got_deltify_free(n->dtab);
211 			n->dtab = NULL;
212 		}
213 
214 		best = raw->size;
215 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
216 			if (cancel_cb) {
217 				err = (*cancel_cb)(cancel_arg);
218 				if (err)
219 					goto done;
220 			}
221 			base = meta[j];
222 			/* long chains make unpacking slow, avoid such bases */
223 			if (base->nchain >= 128 ||
224 			    base->obj_type != m->obj_type)
225 				continue;
226 
227 			err = got_object_raw_open(&base_raw, repo, &base->id,
228 			    8192);
229 			if (err)
230 				goto done;
231 			err = got_deltify(&deltas, &ndeltas,
232 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
233 			    base->dtab, base_raw->f, base_raw->hdrlen,
234 			    base_raw->size + base_raw->hdrlen);
235 			got_object_raw_close(base_raw);
236 			base_raw = NULL;
237 			if (err)
238 				goto done;
239 
240 			size = delta_size(deltas, ndeltas);
241 			if (size + 32 < best){
242 				/*
243 				 * if we already picked a best delta,
244 				 * replace it.
245 				 */
246 				free(m->deltas);
247 				best = size;
248 				m->deltas = deltas;
249 				m->ndeltas = ndeltas;
250 				m->nchain = base->nchain + 1;
251 				m->prev = base;
252 				m->head = base->head;
253 				if (m->head == NULL)
254 					m->head = base;
255 			} else {
256 				free(deltas);
257 				deltas = NULL;
258 				ndeltas = 0;
259 			}
260 		}
261 
262 		got_object_raw_close(raw);
263 		raw = NULL;
264 	}
265 done:
266 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
267 		got_deltify_free(meta[i]->dtab);
268 		meta[i]->dtab = NULL;
269 	}
270 	if (raw)
271 		got_object_raw_close(raw);
272 	if (base_raw)
273 		got_object_raw_close(base_raw);
274 	return err;
275 }
276 
277 static const struct got_error *
278 search_packidx(int *found, struct got_object_id *id,
279     struct got_repository *repo)
280 {
281 	const struct got_error *err = NULL;
282 	struct got_packidx *packidx = NULL;
283 	int idx;
284 
285 	*found = 0;
286 
287 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
288 	if (err == NULL)
289 		*found = 1; /* object is already packed */
290 	else if (err->code == GOT_ERR_NO_OBJ)
291 		err = NULL;
292 	return err;
293 }
294 
295 static const int obj_types[] = {
296 	GOT_OBJ_TYPE_ANY,
297 	GOT_OBJ_TYPE_COMMIT,
298 	GOT_OBJ_TYPE_TREE,
299 	GOT_OBJ_TYPE_BLOB,
300 	GOT_OBJ_TYPE_TAG,
301 	GOT_OBJ_TYPE_OFFSET_DELTA,
302 	GOT_OBJ_TYPE_REF_DELTA
303 };
304 
305 static const struct got_error *
306 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
307     struct got_object_id *id, const char *path, int obj_type,
308     time_t mtime, int loose_obj_only, struct got_repository *repo)
309 {
310 	const struct got_error *err;
311 	struct got_pack_meta *m;
312 
313 	if (loose_obj_only) {
314 		int is_packed;
315 		err = search_packidx(&is_packed, id, repo);
316 		if (err)
317 			return err;
318 		if (is_packed)
319 			return NULL;
320 	}
321 
322 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
323 	if (err)
324 		return err;
325 
326 	if (v == NULL)
327 		return NULL;
328 
329 	err = alloc_meta(&m, id, path, obj_type, mtime);
330 	if (err)
331 		goto done;
332 
333 	if (v->nmeta == v->metasz){
334 		size_t newsize = 2 * v->metasz;
335 		struct got_pack_meta **new;
336 		new = reallocarray(v->meta, newsize, sizeof(*new));
337 		if (new == NULL) {
338 			err = got_error_from_errno("reallocarray");
339 			goto done;
340 		}
341 		v->meta = new;
342 		v->metasz = newsize; 
343 	}
344 done:
345 	if (err) {
346 		clear_meta(m);
347 		free(m);
348 	} else
349 		v->meta[v->nmeta++] = m;
350 
351 	return err;
352 }
353 
354 static const struct got_error *
355 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
356     struct got_object_idset *idset, struct got_object_id *tree_id,
357     const char *dpath, time_t mtime, struct got_repository *repo,
358     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
359 {
360 	const struct got_error *err;
361 	struct got_tree_object *tree;
362 	char *p = NULL;
363 	int i;
364 
365 	err = got_object_open_as_tree(&tree, repo, tree_id);
366 	if (err)
367 		return err;
368 
369 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
370 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
371 		struct got_object_id *id = got_tree_entry_get_id(e);
372 		mode_t mode = got_tree_entry_get_mode(e);
373 
374 		if (cancel_cb) {
375 			err = (*cancel_cb)(cancel_arg);
376 			if (err)
377 				break;
378 		}
379 
380 		if (got_object_tree_entry_is_submodule(e) ||
381 		    got_object_idset_contains(idset, id))
382 			continue;
383 		
384 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
385 		    got_tree_entry_get_name(e)) == -1) {
386 			err = got_error_from_errno("asprintf");
387 			break;
388 		}
389 
390 		if (S_ISDIR(mode)) {
391 			struct got_object_qid *qid;
392 			err = got_object_qid_alloc(&qid, id);
393 			if (err)
394 				break;
395 			STAILQ_INSERT_TAIL(ids, qid, entry);
396 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
397 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
398 			    mtime, loose_obj_only, repo);
399 			if (err)
400 				break;
401 		}
402 		free(p);
403 		p = NULL;
404 	}
405 
406 	got_object_tree_close(tree);
407 	free(p);
408 	return err;
409 }
410 
411 static const struct got_error *
412 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
413     struct got_object_id *tree_id, const char *dpath, time_t mtime,
414     int loose_obj_only, struct got_repository *repo,
415     got_cancel_cb cancel_cb, void *cancel_arg)
416 {
417 	const struct got_error *err = NULL;
418 	struct got_object_id_queue tree_ids;
419 	struct got_object_qid *qid;
420 
421 	if (got_object_idset_contains(idset, tree_id))
422 		return NULL;
423 
424 	err = got_object_qid_alloc(&qid, tree_id);
425 	if (err)
426 		return err;
427 
428 	STAILQ_INIT(&tree_ids);
429 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
430 
431 	while (!STAILQ_EMPTY(&tree_ids)) {
432 		if (cancel_cb) {
433 			err = (*cancel_cb)(cancel_arg);
434 			if (err)
435 				break;
436 		}
437 
438 		qid = STAILQ_FIRST(&tree_ids);
439 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
440 
441 		if (got_object_idset_contains(idset, qid->id)) {
442 			got_object_qid_free(qid);
443 			continue;
444 		}
445 
446 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
447 		    mtime, loose_obj_only, repo);
448 		if (err) {
449 			got_object_qid_free(qid);
450 			break;
451 		}
452 
453 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
454 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
455 		got_object_qid_free(qid);
456 		if (err)
457 			break;
458 	}
459 
460 	got_object_id_queue_free(&tree_ids);
461 	return err;
462 }
463 
464 static const struct got_error *
465 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
466     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
467     got_cancel_cb cancel_cb, void *cancel_arg)
468 {
469 	const struct got_error *err;
470 	struct got_commit_object *commit;
471 
472 	if (got_object_idset_contains(idset, id))
473 		return NULL;
474 
475 	if (loose_obj_only) {
476 		int is_packed;
477 		err = search_packidx(&is_packed, id, repo);
478 		if (err)
479 			return err;
480 		if (is_packed)
481 			return NULL;
482 	}
483 
484 	err = got_object_open_as_commit(&commit, repo, id);
485 	if (err)
486 		return err;
487 
488 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
489 	    got_object_commit_get_committer_time(commit),
490 	    loose_obj_only, repo);
491 	if (err)
492 		goto done;
493 
494 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
495 	    "", got_object_commit_get_committer_time(commit),
496 	    loose_obj_only, repo, cancel_cb, cancel_arg);
497 done:
498 	got_object_commit_close(commit);
499 	return err;
500 }
501 
502 static const struct got_error *
503 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
504     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
505     got_cancel_cb cancel_cb, void *cancel_arg)
506 {
507 	const struct got_error *err;
508 	struct got_tag_object *tag = NULL;
509 
510 	if (got_object_idset_contains(idset, id))
511 		return NULL;
512 
513 	if (loose_obj_only) {
514 		int is_packed;
515 		err = search_packidx(&is_packed, id, repo);
516 		if (err)
517 			return err;
518 		if (is_packed)
519 			return NULL;
520 	}
521 
522 	err = got_object_open_as_tag(&tag, repo, id);
523 	if (err)
524 		return err;
525 
526 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
527 	    got_object_tag_get_tagger_time(tag),
528 	    loose_obj_only, repo);
529 	if (err)
530 		goto done;
531 
532 	switch (got_object_tag_get_object_type(tag)) {
533 	case GOT_OBJ_TYPE_COMMIT:
534 		err = load_commit(v, idset,
535 		    got_object_tag_get_object_id(tag), repo,
536 		    loose_obj_only, cancel_cb, cancel_arg);
537 		break;
538 	case GOT_OBJ_TYPE_TREE:
539 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
540 		    "", got_object_tag_get_tagger_time(tag),
541 		    loose_obj_only, repo, cancel_cb, cancel_arg);
542 		break;
543 	default:
544 		break;
545 	}
546 
547 done:
548 	got_object_tag_close(tag);
549 	return err;
550 }
551 
552 enum findtwixt_color {
553 	COLOR_KEEP = 0,
554 	COLOR_DROP,
555 	COLOR_BLANK,
556 };
557 static const int findtwixt_colors[] = {
558 	COLOR_KEEP,
559 	COLOR_DROP,
560 	COLOR_BLANK
561 };
562 
563 static const struct got_error *
564 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
565     int color, struct got_repository *repo)
566 {
567 	const struct got_error *err;
568 	struct got_object_qid *qid;
569 
570 	err = got_object_qid_alloc(&qid, id);
571 	if (err)
572 		return err;
573 
574 	STAILQ_INSERT_TAIL(ids, qid, entry);
575 	qid->data = (void *)&findtwixt_colors[color];
576 	return NULL;
577 }
578 
579 static const struct got_error *
580 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
581     struct got_object_id *id, struct got_repository *repo,
582     got_cancel_cb cancel_cb, void *cancel_arg)
583 {
584 	const struct got_error *err = NULL;
585 	struct got_commit_object *commit;
586 	const struct got_object_id_queue *parents;
587 	struct got_object_id_queue ids;
588 	struct got_object_qid *qid;
589 
590 	STAILQ_INIT(&ids);
591 
592 	err = got_object_qid_alloc(&qid, id);
593 	if (err)
594 		return err;
595 	STAILQ_INSERT_HEAD(&ids, qid, entry);
596 
597 	while (!STAILQ_EMPTY(&ids)) {
598 		if (cancel_cb) {
599 			err = (*cancel_cb)(cancel_arg);
600 			if (err)
601 				break;
602 		}
603 
604 		qid = STAILQ_FIRST(&ids);
605 		STAILQ_REMOVE_HEAD(&ids, entry);
606 
607 		if (got_object_idset_contains(drop, qid->id)) {
608 			got_object_qid_free(qid);
609 			continue;
610 		}
611 
612 		err = got_object_idset_add(drop, qid->id,
613 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
614 		if (err) {
615 			got_object_qid_free(qid);
616 			break;
617 		}
618 
619 		if (!got_object_idset_contains(keep, qid->id)) {
620 			got_object_qid_free(qid);
621 			continue;
622 		}
623 
624 		err = got_object_open_as_commit(&commit, repo, qid->id);
625 		got_object_qid_free(qid);
626 		if (err)
627 			break;
628 
629 		parents = got_object_commit_get_parent_ids(commit);
630 		if (parents) {
631 			err = got_object_id_queue_copy(parents, &ids);
632 			if (err) {
633 				got_object_commit_close(commit);
634 				break;
635 			}
636 		}
637 		got_object_commit_close(commit);
638 	}
639 
640 	got_object_id_queue_free(&ids);
641 	return err;
642 }
643 
644 struct append_id_arg {
645 	struct got_object_id **array;
646 	int idx;
647 };
648 
649 static const struct got_error *
650 append_id(struct got_object_id *id, void *data, void *arg)
651 {
652 	struct append_id_arg *a = arg;
653 
654 	a->array[a->idx] = got_object_id_dup(id);
655 	if (a->array[a->idx] == NULL)
656 		return got_error_from_errno("got_object_id_dup");
657 
658 	a->idx++;
659 	return NULL;
660 }
661 
662 static const struct got_error *
663 findtwixt(struct got_object_id ***res, int *nres,
664     struct got_object_id **head, int nhead,
665     struct got_object_id **tail, int ntail,
666     struct got_repository *repo,
667     got_cancel_cb cancel_cb, void *cancel_arg)
668 {
669 	const struct got_error *err = NULL;
670 	struct got_object_id_queue ids;
671 	struct got_object_idset *keep, *drop;
672 	struct got_object_qid *qid;
673 	int i, ncolor, nkeep, obj_type;
674 
675 	STAILQ_INIT(&ids);
676 	*res = NULL;
677 	*nres = 0;
678 
679 	keep = got_object_idset_alloc();
680 	if (keep == NULL)
681 		return got_error_from_errno("got_object_idset_alloc");
682 
683 	drop = got_object_idset_alloc();
684 	if (drop == NULL) {
685 		err = got_error_from_errno("got_object_idset_alloc");
686 		goto done;
687 	}
688 
689 	for (i = 0; i < nhead; i++) {
690 		struct got_object_id *id = head[i];
691 		if (id == NULL)
692 			continue;
693 		err = got_object_get_type(&obj_type, repo, id);
694 		if (err)
695 			return err;
696 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
697 			continue;
698 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
699 		if (err)
700 			goto done;
701 	}		
702 	for (i = 0; i < ntail; i++) {
703 		struct got_object_id *id = tail[i];
704 		if (id == NULL)
705 			continue;
706 		err = got_object_get_type(&obj_type, repo, id);
707 		if (err)
708 			return err;
709 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
710 			continue;
711 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
712 		if (err)
713 			goto done;
714 	}
715 
716 	while (!STAILQ_EMPTY(&ids)) {
717 		int qcolor;
718 		qid = STAILQ_FIRST(&ids);
719 		qcolor = *((int *)qid->data);
720 
721 		if (got_object_idset_contains(drop, qid->id))
722 			ncolor = COLOR_DROP;
723 		else if (got_object_idset_contains(keep, qid->id))
724 			ncolor = COLOR_KEEP;
725 		else
726 			ncolor = COLOR_BLANK;
727 
728 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
729 		    qcolor == COLOR_KEEP)) {
730 			STAILQ_REMOVE_HEAD(&ids, entry);
731 			got_object_qid_free(qid);
732 			continue;
733 		}
734 
735 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
736 			err = drop_commit(keep, drop, qid->id, repo,
737 			    cancel_cb, cancel_arg);
738 			if (err)
739 				goto done;
740 		} else if (ncolor == COLOR_BLANK) {
741 			struct got_commit_object *commit;
742 			struct got_object_id *id;
743 			const struct got_object_id_queue *parents;
744 			struct got_object_qid *pid;
745 
746 			id = got_object_id_dup(qid->id);
747 			if (id == NULL) {
748 				err = got_error_from_errno("got_object_id_dup");
749 				goto done;
750 			}
751 			if (qcolor == COLOR_KEEP)
752 				err = got_object_idset_add(keep, id,
753 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
754 			else
755 				err = got_object_idset_add(drop, id,
756 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
757 			if (err) {
758 				free(id);
759 				goto done;
760 			}
761 
762 			err = got_object_open_as_commit(&commit, repo, id);
763 			if (err) {
764 				free(id);
765 				goto done;
766 			}
767 			parents = got_object_commit_get_parent_ids(commit);
768 			if (parents) {
769 				STAILQ_FOREACH(pid, parents, entry) {
770 					err = queue_commit_id(&ids, pid->id,
771 					    qcolor, repo);
772 					if (err) {
773 						free(id);
774 						goto done;
775 					}
776 				}
777 			}
778 			got_object_commit_close(commit);
779 			commit = NULL;
780 		} else {
781 			/* should not happen */
782 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
783 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
784 			goto done;
785 		}
786 
787 		STAILQ_REMOVE_HEAD(&ids, entry);
788 		got_object_qid_free(qid);
789 	}
790 
791 	nkeep = got_object_idset_num_elements(keep);
792 	if (nkeep > 0) {
793 		struct append_id_arg arg;
794 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
795 		if (arg.array == NULL) {
796 			err = got_error_from_errno("calloc");
797 			goto done;
798 		}
799 		arg.idx = 0;
800 		err = got_object_idset_for_each(keep, append_id, &arg);
801 		if (err) {
802 			free(arg.array);
803 			goto done;
804 		}
805 		*res = arg.array;
806 		*nres = nkeep;
807 	}
808 done:
809 	got_object_idset_free(keep);
810 	got_object_idset_free(drop);
811 	got_object_id_queue_free(&ids);
812 	return err;
813 }
814 
815 static const struct got_error *
816 read_meta(struct got_pack_meta ***meta, int *nmeta,
817     struct got_object_id **theirs, int ntheirs,
818     struct got_object_id **ours, int nours, struct got_repository *repo,
819     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
820     got_cancel_cb cancel_cb, void *cancel_arg)
821 {
822 	const struct got_error *err = NULL;
823 	struct got_object_id **ids = NULL;
824 	struct got_object_idset *idset;
825 	int i, nobj = 0, obj_type;
826 	struct got_pack_metavec v;
827 
828 	*meta = NULL;
829 	*nmeta = 0;
830 
831 	idset = got_object_idset_alloc();
832 	if (idset == NULL)
833 		return got_error_from_errno("got_object_idset_alloc");
834 
835 	v.nmeta = 0;
836 	v.metasz = 64;
837 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
838 	if (v.meta == NULL) {
839 		err = got_error_from_errno("calloc");
840 		goto done;
841 	}
842 
843 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
844 	    cancel_cb, cancel_arg);
845 	if (err || nobj == 0)
846 		goto done;
847 
848 	for (i = 0; i < ntheirs; i++) {
849 		struct got_object_id *id = theirs[i];
850 		if (id == NULL)
851 			continue;
852 		err = got_object_get_type(&obj_type, repo, id);
853 		if (err)
854 			return err;
855 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
856 			continue;
857 		err = load_commit(NULL, idset, id, repo,
858 		    loose_obj_only, cancel_cb, cancel_arg);
859 		if (err)
860 			goto done;
861 		if (progress_cb) {
862 			err = progress_cb(progress_arg, 0L, nours,
863 			    v.nmeta, 0, 0);
864 			if (err)
865 				goto done;
866 		}
867 	}
868 
869 	for (i = 0; i < ntheirs; i++) {
870 		struct got_object_id *id = theirs[i];
871 		int *cached_type;
872 		if (id == NULL)
873 			continue;
874 		cached_type = got_object_idset_get(idset, id);
875 		if (cached_type == NULL) {
876 			err = got_object_get_type(&obj_type, repo, id);
877 			if (err)
878 				goto done;
879 		} else
880 			obj_type = *cached_type;
881 		if (obj_type != GOT_OBJ_TYPE_TAG)
882 			continue;
883 		err = load_tag(NULL, idset, id, repo,
884 		    loose_obj_only, cancel_cb, cancel_arg);
885 		if (err)
886 			goto done;
887 		if (progress_cb) {
888 			err = progress_cb(progress_arg, 0L, nours,
889 			    v.nmeta, 0, 0);
890 			if (err)
891 				goto done;
892 		}
893 	}
894 
895 	for (i = 0; i < nobj; i++) {
896 		err = load_commit(&v, idset, ids[i], repo,
897 		    loose_obj_only, cancel_cb, cancel_arg);
898 		if (err)
899 			goto done;
900 		if (progress_cb) {
901 			err = progress_cb(progress_arg, 0L, nours,
902 			    v.nmeta, 0, 0);
903 			if (err)
904 				goto done;
905 		}
906 	}
907 
908 	for (i = 0; i < nours; i++) {
909 		struct got_object_id *id = ours[i];
910 		int *cached_type;
911 		if (id == NULL)
912 			continue;
913 		cached_type = got_object_idset_get(idset, id);
914 		if (cached_type == NULL) {
915 			err = got_object_get_type(&obj_type, repo, id);
916 			if (err)
917 				goto done;
918 		} else
919 			obj_type = *cached_type;
920 		if (obj_type != GOT_OBJ_TYPE_TAG)
921 			continue;
922 		err = load_tag(&v, idset, id, repo,
923 		    loose_obj_only, cancel_cb, cancel_arg);
924 		if (err)
925 			goto done;
926 		if (progress_cb) {
927 			err = progress_cb(progress_arg, 0L, nours,
928 			    v.nmeta, 0, 0);
929 			if (err)
930 				goto done;
931 		}
932 	}
933 
934 done:
935 	for (i = 0; i < nobj; i++) {
936 		free(ids[i]);
937 	}
938 	free(ids);
939 	got_object_idset_free(idset);
940 	if (err == NULL) {
941 		*meta = v.meta;
942 		*nmeta = v.nmeta;
943 	} else
944 		free(v.meta);
945 
946 	return err;
947 }
948 
949 const struct got_error *
950 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
951 {
952 	size_t n;
953 
954 	SHA1Update(ctx, buf, len);
955 	n = fwrite(buf, 1, len, f);
956 	if (n != len)
957 		return got_ferror(f, GOT_ERR_IO);
958 	return NULL;
959 }
960 
961 static void
962 putbe32(char *b, uint32_t n)
963 {
964 	b[0] = n >> 24;
965 	b[1] = n >> 16;
966 	b[2] = n >> 8;
967 	b[3] = n >> 0;
968 }
969 
970 static int
971 write_order_cmp(const void *pa, const void *pb)
972 {
973 	struct got_pack_meta *a, *b, *ahd, *bhd;
974 
975 	a = *(struct got_pack_meta **)pa;
976 	b = *(struct got_pack_meta **)pb;
977 	ahd = (a->head == NULL) ? a : a->head;
978 	bhd = (b->head == NULL) ? b : b->head;
979 	if (ahd->mtime != bhd->mtime)
980 		return bhd->mtime - ahd->mtime;
981 	if (ahd != bhd)
982 		return (uintptr_t)bhd - (uintptr_t)ahd;
983 	if (a->nchain != b->nchain)
984 		return a->nchain - b->nchain;
985 	return a->mtime - b->mtime;
986 }
987 
988 static const struct got_error *
989 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
990 {
991 	size_t i;
992 
993 	*hdrlen = 0;
994 
995 	hdr[0] = obj_type << 4;
996 	hdr[0] |= len & 0xf;
997 	len >>= 4;
998 	for (i = 1; len != 0; i++){
999 		if (i >= bufsize)
1000 			return got_error(GOT_ERR_NO_SPACE);
1001 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1002 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1003 		len >>= GOT_DELTA_SIZE_SHIFT;
1004 	}
1005 
1006 	*hdrlen = i;
1007 	return NULL;
1008 }
1009 
1010 static const struct got_error *
1011 encodedelta(struct got_pack_meta *m, struct got_raw_object *o,
1012     off_t base_size, FILE *f)
1013 {
1014 	unsigned char buf[16], *bp;
1015 	int i, j;
1016 	off_t n;
1017 	size_t w;
1018 	struct got_delta_instruction *d;
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 	w = fwrite(buf, 1, i, f);
1029 	if (w != i)
1030 		return got_ferror(f, GOT_ERR_IO);
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 	w = fwrite(buf, 1, i, f);
1041 	if (w != i)
1042 		return got_ferror(f, GOT_ERR_IO);
1043 
1044 	for (j = 0; j < m->ndeltas; j++) {
1045 		d = &m->deltas[j];
1046 		if (d->copy) {
1047 			n = d->offset;
1048 			bp = &buf[1];
1049 			buf[0] = GOT_DELTA_BASE_COPY;
1050 			for (i = 0; i < 4; i++) {
1051 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
1052 				buf[0] |= 1 << i;
1053 				*bp++ = n & 0xff;
1054 				n >>= 8;
1055 				if (n == 0)
1056 					break;
1057 			}
1058 
1059 			n = d->len;
1060 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
1061 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
1062 				for (i = 0; i < 3 && n > 0; i++) {
1063 					buf[0] |= 1 << (i + 4);
1064 					*bp++ = n & 0xff;
1065 					n >>= 8;
1066 				}
1067 			}
1068 			w = fwrite(buf, 1, bp - buf, f);
1069 			if (w != bp - buf)
1070 				return got_ferror(f, GOT_ERR_IO);
1071 		} else {
1072 			char content[128];
1073 			size_t r;
1074 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
1075 				return got_error_from_errno("fseeko");
1076 			n = 0;
1077 			while (n != d->len) {
1078 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
1079 				w = fwrite(buf, 1, 1, f);
1080 				if (w != 1)
1081 					return got_ferror(f, GOT_ERR_IO);
1082 				r = fread(content, 1, buf[0], o->f);
1083 				if (r != buf[0])
1084 					return got_ferror(o->f, GOT_ERR_IO);
1085 				w = fwrite(content, 1, buf[0], f);
1086 				if (w != buf[0])
1087 					return got_ferror(f, GOT_ERR_IO);
1088 				n += buf[0];
1089 			}
1090 		}
1091 	}
1092 
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;
1123 	off_t nd;
1124 	SHA1_CTX ctx;
1125 	struct got_pack_meta *m;
1126 	struct got_raw_object *raw = NULL;
1127 	FILE *delta_file = NULL;
1128 	char 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 			if (delta_file == NULL) {
1181 				delta_file = got_opentemp();
1182 				if (delta_file == NULL) {
1183 					err = got_error_from_errno(
1184 					    "got_opentemp");
1185 					goto done;
1186 				}
1187 			}
1188 			if (ftruncate(fileno(delta_file), 0L) == -1) {
1189 				err = got_error_from_errno("ftruncate");
1190 				goto done;
1191 			}
1192 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1193 				err = got_error_from_errno("fseeko");
1194 				goto done;
1195 			}
1196 			err = encodedelta(m, raw, m->prev->size, delta_file);
1197 			if (err)
1198 				goto done;
1199 			nd = ftello(delta_file);
1200 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1201 				err = got_error_from_errno("fseeko");
1202 				goto done;
1203 			}
1204 			if (use_offset_deltas && m->prev->off != 0) {
1205 				err = packhdr(&nh, buf, sizeof(buf),
1206 				    GOT_OBJ_TYPE_OFFSET_DELTA, nd);
1207 				if (err)
1208 					goto done;
1209 				nh += packoff(buf + nh,
1210 				    m->off - m->prev->off);
1211 				err = hwrite(packfile, buf, nh, &ctx);
1212 				if (err)
1213 					goto done;
1214 				packfile_size += nh;
1215 			} else {
1216 				err = packhdr(&nh, buf, sizeof(buf),
1217 				    GOT_OBJ_TYPE_REF_DELTA, nd);
1218 				err = hwrite(packfile, buf, nh, &ctx);
1219 				if (err)
1220 					goto done;
1221 				packfile_size += nh;
1222 				err = hwrite(packfile, m->prev->id.sha1,
1223 				    sizeof(m->prev->id.sha1), &ctx);
1224 				packfile_size += sizeof(m->prev->id.sha1);
1225 				if (err)
1226 					goto done;
1227 			}
1228 			err = got_deflate_to_file(&outlen, delta_file,
1229 			    packfile, &csum);
1230 			if (err)
1231 				goto done;
1232 			packfile_size += outlen;
1233 		}
1234 		got_object_raw_close(raw);
1235 		raw = NULL;
1236 	}
1237 	SHA1Final(pack_sha1, &ctx);
1238 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1239 	if (n != SHA1_DIGEST_LENGTH)
1240 		err = got_ferror(packfile, GOT_ERR_IO);
1241 	packfile_size += SHA1_DIGEST_LENGTH;
1242 	packfile_size += sizeof(struct got_packfile_hdr);
1243 	err = progress_cb(progress_arg, packfile_size, nours,
1244 	    nmeta, nmeta, nmeta);
1245 	if (err)
1246 		goto done;
1247 done:
1248 	if (delta_file && fclose(delta_file) == EOF && err == NULL)
1249 		err = got_error_from_errno("fclose");
1250 	if (raw)
1251 		got_object_raw_close(raw);
1252 	return err;
1253 }
1254 
1255 const struct got_error *
1256 got_pack_create(uint8_t *packsha1, FILE *packfile,
1257     struct got_object_id **theirs, int ntheirs,
1258     struct got_object_id **ours, int nours,
1259     struct got_repository *repo, int loose_obj_only, int allow_empty,
1260     got_pack_progress_cb progress_cb, void *progress_arg,
1261     got_cancel_cb cancel_cb, void *cancel_arg)
1262 {
1263 	const struct got_error *err;
1264 	struct got_pack_meta **meta;
1265 	int nmeta;
1266 
1267 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1268 	    loose_obj_only, progress_cb, progress_arg, cancel_cb, cancel_arg);
1269 	if (err)
1270 		return err;
1271 
1272 	if (nmeta == 0 && !allow_empty) {
1273 		err = got_error(GOT_ERR_CANNOT_PACK);
1274 		goto done;
1275 	}
1276 	if (nmeta > 0) {
1277 		err = pick_deltas(meta, nmeta, nours, repo,
1278 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
1279 		if (err)
1280 			goto done;
1281 	}
1282 
1283 	err = genpack(packsha1, packfile, meta, nmeta, nours, 1, repo,
1284 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
1285 	if (err)
1286 		goto done;
1287 done:
1288 	free_nmeta(meta, nmeta);
1289 	return err;
1290 }