op public repos

Blob

Date:: Wed Jan 5 19:57:10 2022 UTC
Message:: use time-based rate-limiting for gotadmin progress output Suggested by naddy some time ago. ok tracey
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 #include <sys/time.h>
24 
25 #include <stdint.h>
26 #include <imsg.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <sha1.h>
31 #include <time.h>
32 #include <limits.h>
33 #include <zlib.h>
34 
35 #include "got_error.h"
36 #include "got_cancel.h"
37 #include "got_object.h"
38 #include "got_path.h"
39 #include "got_reference.h"
40 #include "got_repository_admin.h"
41 #include "got_opentemp.h"
42 
43 #include "got_lib_deltify.h"
44 #include "got_lib_delta.h"
45 #include "got_lib_object.h"
46 #include "got_lib_object_idset.h"
47 #include "got_lib_object_cache.h"
48 #include "got_lib_deflate.h"
49 #include "got_lib_pack.h"
50 #include "got_lib_privsep.h"
51 #include "got_lib_repository.h"
52 #include "got_lib_ratelimit.h"
53 
54 #ifndef MIN
55 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
56 #endif
57 
58 #ifndef MAX
59 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
60 #endif
61 
62 struct got_pack_meta {
63 	struct got_object_id id;
64 	char	*path;
65 	int	obj_type;
66 	off_t	size;
67 	time_t	mtime;
68 
69 	/* The best delta we picked */
70 	struct got_pack_meta *head;
71 	struct got_pack_meta *prev;
72 	off_t	delta_offset;	/* offset in delta cache file */
73 	off_t	delta_len;	/* length in delta cache file */
74 	int	nchain;
75 
76 	/* Only used for delta window */
77 	struct got_delta_table *dtab;
78 
79 	/* Only used for writing offset deltas */
80 	off_t	off;
81 };
82 
83 struct got_pack_metavec {
84 	struct got_pack_meta **meta;
85 	int nmeta;
86 	int metasz;
87 };
88 
89 static const struct got_error *
90 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
91     const char *path, int obj_type, time_t mtime)
92 {
93 	const struct got_error *err = NULL;
94 	struct got_pack_meta *m;
95 
96 	*new = NULL;
97 
98 	m = calloc(1, sizeof(*m));
99 	if (m == NULL)
100 		return got_error_from_errno("calloc");
101 
102 	memcpy(&m->id, id, sizeof(m->id));
103 
104 	m->path = strdup(path);
105 	if (m->path == NULL) {
106 		err = got_error_from_errno("strdup");
107 		free(m);
108 		return err;
109 	}
110 
111 	m->obj_type = obj_type;
112 	m->mtime = mtime;
113 	*new = m;
114 	return NULL;
115 }
116 
117 static void
118 clear_meta(struct got_pack_meta *meta)
119 {
120 	if (meta == NULL)
121 		return;
122 	free(meta->path);
123 	meta->path = NULL;
124 }
125 
126 static void
127 free_nmeta(struct got_pack_meta **meta, int nmeta)
128 {
129 	int i;
130 
131 	for (i = 0; i < nmeta; i++)
132 		clear_meta(meta[i]);
133 	free(meta);
134 }
135 
136 static int
137 delta_order_cmp(const void *pa, const void *pb)
138 {
139 	struct got_pack_meta *a, *b;
140 	int cmp;
141 
142 	a = *(struct got_pack_meta **)pa;
143 	b = *(struct got_pack_meta **)pb;
144 
145 	if (a->obj_type != b->obj_type)
146 		return a->obj_type - b->obj_type;
147 	cmp = strcmp(a->path, b->path);
148 	if (cmp != 0)
149 		return cmp;
150 	if (a->mtime != b->mtime)
151 		return a->mtime - b->mtime;
152 	return got_object_id_cmp(&a->id, &b->id);
153 }
154 
155 static int
156 delta_size(struct got_delta_instruction *deltas, int ndeltas)
157 {
158 	int i, size = 32;
159 	for (i = 0; i < ndeltas; i++) {
160 		if (deltas[i].copy)
161 			size += GOT_DELTA_SIZE_SHIFT;
162 		else
163 			size += deltas[i].len + 1;
164 	}
165 	return size;
166 }
167 
168 static const struct got_error *
169 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
170     struct got_delta_instruction *deltas, int ndeltas,
171     off_t base_size, FILE *f)
172 {
173 	unsigned char buf[16], *bp;
174 	int i, j;
175 	off_t n;
176 	size_t w;
177 	struct got_delta_instruction *d;
178 
179 	/* base object size */
180 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
181 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
182 	for (i = 1; n > 0; i++) {
183 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
184 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
185 		n >>= GOT_DELTA_SIZE_SHIFT;
186 	}
187 	w = fwrite(buf, 1, i, f);
188 	if (w != i)
189 		return got_ferror(f, GOT_ERR_IO);
190 
191 	/* target object size */
192 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
193 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
194 	for (i = 1; n > 0; i++) {
195 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
196 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
197 		n >>= GOT_DELTA_SIZE_SHIFT;
198 	}
199 	w = fwrite(buf, 1, i, f);
200 	if (w != i)
201 		return got_ferror(f, GOT_ERR_IO);
202 
203 	for (j = 0; j < ndeltas; j++) {
204 		d = &deltas[j];
205 		if (d->copy) {
206 			n = d->offset;
207 			bp = &buf[1];
208 			buf[0] = GOT_DELTA_BASE_COPY;
209 			for (i = 0; i < 4; i++) {
210 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
211 				buf[0] |= 1 << i;
212 				*bp++ = n & 0xff;
213 				n >>= 8;
214 				if (n == 0)
215 					break;
216 			}
217 
218 			n = d->len;
219 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
220 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
221 				for (i = 0; i < 3 && n > 0; i++) {
222 					buf[0] |= 1 << (i + 4);
223 					*bp++ = n & 0xff;
224 					n >>= 8;
225 				}
226 			}
227 			w = fwrite(buf, 1, bp - buf, f);
228 			if (w != bp - buf)
229 				return got_ferror(f, GOT_ERR_IO);
230 		} else {
231 			char content[128];
232 			size_t r;
233 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
234 				return got_error_from_errno("fseeko");
235 			n = 0;
236 			while (n != d->len) {
237 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
238 				w = fwrite(buf, 1, 1, f);
239 				if (w != 1)
240 					return got_ferror(f, GOT_ERR_IO);
241 				r = fread(content, 1, buf[0], o->f);
242 				if (r != buf[0])
243 					return got_ferror(o->f, GOT_ERR_IO);
244 				w = fwrite(content, 1, buf[0], f);
245 				if (w != buf[0])
246 					return got_ferror(f, GOT_ERR_IO);
247 				n += buf[0];
248 			}
249 		}
250 	}
251 
252 	return NULL;
253 }
254 
255 static const struct got_error *
256 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
257     struct got_ratelimit *rl, off_t packfile_size, int ncommits,
258     int nobj_total, int obj_deltify, int nobj_written)
259 {
260 	const struct got_error *err;
261 	int elapsed;
262 
263 	if (progress_cb == NULL)
264 		return NULL;
265 
266 	err = got_ratelimit_check(&elapsed, rl);
267 	if (err || !elapsed)
268 		return err;
269 
270 	return progress_cb(progress_arg, packfile_size, ncommits,
271 	    nobj_total, obj_deltify, nobj_written);
272 }
273 
274 static const struct got_error *
275 pick_deltas(struct got_pack_meta **meta, int nmeta, int nours,
276     FILE *delta_cache, struct got_repository *repo,
277     got_pack_progress_cb progress_cb, void *progress_arg,
278     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
279 {
280 	const struct got_error *err = NULL;
281 	struct got_pack_meta *m = NULL, *base = NULL;
282 	struct got_raw_object *raw = NULL, *base_raw = NULL;
283 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
284 	int i, j, size, best_size, ndeltas, best_ndeltas;
285 	const int max_base_candidates = 3;
286 	int outfd = -1;
287 
288 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
289 	for (i = 0; i < nmeta; i++) {
290 		if (cancel_cb) {
291 			err = (*cancel_cb)(cancel_arg);
292 			if (err)
293 				break;
294 		}
295 		err = report_progress(progress_cb, progress_arg, rl,
296 		    0L, nours, nmeta, i, 0);
297 		if (err)
298 			goto done;
299 		m = meta[i];
300 
301 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
302 		    m->obj_type == GOT_OBJ_TYPE_TAG)
303 			continue;
304 
305 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
306 		if (err)
307 			goto done;
308 		m->size = raw->size;
309 
310 		err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
311 		    raw->size + raw->hdrlen);
312 		if (err)
313 			goto done;
314 
315 		if (i > max_base_candidates) {
316 			struct got_pack_meta *n = NULL;
317 			n = meta[i - (max_base_candidates + 1)];
318 			got_deltify_free(n->dtab);
319 			n->dtab = NULL;
320 		}
321 
322 		best_size = raw->size;
323 		best_ndeltas = 0;
324 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
325 			if (cancel_cb) {
326 				err = (*cancel_cb)(cancel_arg);
327 				if (err)
328 					goto done;
329 			}
330 			base = meta[j];
331 			/* long chains make unpacking slow, avoid such bases */
332 			if (base->nchain >= 128 ||
333 			    base->obj_type != m->obj_type)
334 				continue;
335 
336 			err = got_object_raw_open(&base_raw, &outfd, repo,
337 			    &base->id);
338 			if (err)
339 				goto done;
340 			err = got_deltify(&deltas, &ndeltas,
341 			    raw->f, raw->hdrlen, raw->size + raw->hdrlen,
342 			    base->dtab, base_raw->f, base_raw->hdrlen,
343 			    base_raw->size + base_raw->hdrlen);
344 			got_object_raw_close(base_raw);
345 			base_raw = NULL;
346 			if (err)
347 				goto done;
348 
349 			size = delta_size(deltas, ndeltas);
350 			if (size + 32 < best_size){
351 				/*
352 				 * if we already picked a best delta,
353 				 * replace it.
354 				 */
355 				best_size = size;
356 				free(best_deltas);
357 				best_deltas = deltas;
358 				best_ndeltas = ndeltas;
359 				deltas = NULL;
360 				m->nchain = base->nchain + 1;
361 				m->prev = base;
362 				m->head = base->head;
363 				if (m->head == NULL)
364 					m->head = base;
365 			} else {
366 				free(deltas);
367 				deltas = NULL;
368 				ndeltas = 0;
369 			}
370 		}
371 
372 		if (best_ndeltas > 0) {
373 			m->delta_offset = ftello(delta_cache);
374 			err = encode_delta(m, raw, best_deltas,
375 			    best_ndeltas, m->prev->size, delta_cache);
376 			free(best_deltas);
377 			best_deltas = NULL;
378 			best_ndeltas = 0;
379 			if (err)
380 				goto done;
381 			m->delta_len = ftello(delta_cache) - m->delta_offset;
382 		}
383 
384 		got_object_raw_close(raw);
385 		raw = NULL;
386 	}
387 done:
388 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
389 		got_deltify_free(meta[i]->dtab);
390 		meta[i]->dtab = NULL;
391 	}
392 	if (raw)
393 		got_object_raw_close(raw);
394 	if (base_raw)
395 		got_object_raw_close(base_raw);
396 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
397 		err = got_error_from_errno("close");
398 	free(deltas);
399 	free(best_deltas);
400 	return err;
401 }
402 
403 static const struct got_error *
404 search_packidx(int *found, struct got_object_id *id,
405     struct got_repository *repo)
406 {
407 	const struct got_error *err = NULL;
408 	struct got_packidx *packidx = NULL;
409 	int idx;
410 
411 	*found = 0;
412 
413 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
414 	if (err == NULL)
415 		*found = 1; /* object is already packed */
416 	else if (err->code == GOT_ERR_NO_OBJ)
417 		err = NULL;
418 	return err;
419 }
420 
421 static const int obj_types[] = {
422 	GOT_OBJ_TYPE_ANY,
423 	GOT_OBJ_TYPE_COMMIT,
424 	GOT_OBJ_TYPE_TREE,
425 	GOT_OBJ_TYPE_BLOB,
426 	GOT_OBJ_TYPE_TAG,
427 	GOT_OBJ_TYPE_OFFSET_DELTA,
428 	GOT_OBJ_TYPE_REF_DELTA
429 };
430 
431 static const struct got_error *
432 add_meta(struct got_pack_metavec *v, struct got_object_idset *idset,
433     struct got_object_id *id, const char *path, int obj_type,
434     time_t mtime, int loose_obj_only, struct got_repository *repo)
435 {
436 	const struct got_error *err;
437 	struct got_pack_meta *m;
438 
439 	if (loose_obj_only) {
440 		int is_packed;
441 		err = search_packidx(&is_packed, id, repo);
442 		if (err)
443 			return err;
444 		if (is_packed)
445 			return NULL;
446 	}
447 
448 	err = got_object_idset_add(idset, id, (void *)&obj_types[obj_type]);
449 	if (err)
450 		return err;
451 
452 	if (v == NULL)
453 		return NULL;
454 
455 	err = alloc_meta(&m, id, path, obj_type, mtime);
456 	if (err)
457 		goto done;
458 
459 	if (v->nmeta == v->metasz){
460 		size_t newsize = 2 * v->metasz;
461 		struct got_pack_meta **new;
462 		new = reallocarray(v->meta, newsize, sizeof(*new));
463 		if (new == NULL) {
464 			err = got_error_from_errno("reallocarray");
465 			goto done;
466 		}
467 		v->meta = new;
468 		v->metasz = newsize; 
469 	}
470 done:
471 	if (err) {
472 		clear_meta(m);
473 		free(m);
474 	} else
475 		v->meta[v->nmeta++] = m;
476 
477 	return err;
478 }
479 
480 static const struct got_error *
481 load_tree_entries(struct got_object_id_queue *ids, struct got_pack_metavec *v,
482     struct got_object_idset *idset, struct got_object_id *tree_id,
483     const char *dpath, time_t mtime, struct got_repository *repo,
484     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
485 {
486 	const struct got_error *err;
487 	struct got_tree_object *tree;
488 	char *p = NULL;
489 	int i;
490 
491 	err = got_object_open_as_tree(&tree, repo, tree_id);
492 	if (err)
493 		return err;
494 
495 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
496 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
497 		struct got_object_id *id = got_tree_entry_get_id(e);
498 		mode_t mode = got_tree_entry_get_mode(e);
499 
500 		if (cancel_cb) {
501 			err = (*cancel_cb)(cancel_arg);
502 			if (err)
503 				break;
504 		}
505 
506 		if (got_object_tree_entry_is_submodule(e) ||
507 		    got_object_idset_contains(idset, id))
508 			continue;
509 		
510 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
511 		    got_tree_entry_get_name(e)) == -1) {
512 			err = got_error_from_errno("asprintf");
513 			break;
514 		}
515 
516 		if (S_ISDIR(mode)) {
517 			struct got_object_qid *qid;
518 			err = got_object_qid_alloc(&qid, id);
519 			if (err)
520 				break;
521 			STAILQ_INSERT_TAIL(ids, qid, entry);
522 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
523 			err = add_meta(v, idset, id, p, GOT_OBJ_TYPE_BLOB,
524 			    mtime, loose_obj_only, repo);
525 			if (err)
526 				break;
527 		}
528 		free(p);
529 		p = NULL;
530 	}
531 
532 	got_object_tree_close(tree);
533 	free(p);
534 	return err;
535 }
536 
537 static const struct got_error *
538 load_tree(struct got_pack_metavec *v, struct got_object_idset *idset,
539     struct got_object_id *tree_id, const char *dpath, time_t mtime,
540     int loose_obj_only, struct got_repository *repo,
541     got_cancel_cb cancel_cb, void *cancel_arg)
542 {
543 	const struct got_error *err = NULL;
544 	struct got_object_id_queue tree_ids;
545 	struct got_object_qid *qid;
546 
547 	if (got_object_idset_contains(idset, tree_id))
548 		return NULL;
549 
550 	err = got_object_qid_alloc(&qid, tree_id);
551 	if (err)
552 		return err;
553 
554 	STAILQ_INIT(&tree_ids);
555 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
556 
557 	while (!STAILQ_EMPTY(&tree_ids)) {
558 		if (cancel_cb) {
559 			err = (*cancel_cb)(cancel_arg);
560 			if (err)
561 				break;
562 		}
563 
564 		qid = STAILQ_FIRST(&tree_ids);
565 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
566 
567 		if (got_object_idset_contains(idset, qid->id)) {
568 			got_object_qid_free(qid);
569 			continue;
570 		}
571 
572 		err = add_meta(v, idset, qid->id, dpath, GOT_OBJ_TYPE_TREE,
573 		    mtime, loose_obj_only, repo);
574 		if (err) {
575 			got_object_qid_free(qid);
576 			break;
577 		}
578 
579 		err = load_tree_entries(&tree_ids, v, idset, qid->id, dpath,
580 		    mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
581 		got_object_qid_free(qid);
582 		if (err)
583 			break;
584 	}
585 
586 	got_object_id_queue_free(&tree_ids);
587 	return err;
588 }
589 
590 static const struct got_error *
591 load_commit(struct got_pack_metavec *v, struct got_object_idset *idset,
592     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
593     got_cancel_cb cancel_cb, void *cancel_arg)
594 {
595 	const struct got_error *err;
596 	struct got_commit_object *commit;
597 
598 	if (got_object_idset_contains(idset, id))
599 		return NULL;
600 
601 	if (loose_obj_only) {
602 		int is_packed;
603 		err = search_packidx(&is_packed, id, repo);
604 		if (err)
605 			return err;
606 		if (is_packed)
607 			return NULL;
608 	}
609 
610 	err = got_object_open_as_commit(&commit, repo, id);
611 	if (err)
612 		return err;
613 
614 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_COMMIT,
615 	    got_object_commit_get_committer_time(commit),
616 	    loose_obj_only, repo);
617 	if (err)
618 		goto done;
619 
620 	err = load_tree(v, idset, got_object_commit_get_tree_id(commit),
621 	    "", got_object_commit_get_committer_time(commit),
622 	    loose_obj_only, repo, cancel_cb, cancel_arg);
623 done:
624 	got_object_commit_close(commit);
625 	return err;
626 }
627 
628 static const struct got_error *
629 load_tag(struct got_pack_metavec *v, struct got_object_idset *idset,
630     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
631     got_cancel_cb cancel_cb, void *cancel_arg)
632 {
633 	const struct got_error *err;
634 	struct got_tag_object *tag = NULL;
635 
636 	if (got_object_idset_contains(idset, id))
637 		return NULL;
638 
639 	if (loose_obj_only) {
640 		int is_packed;
641 		err = search_packidx(&is_packed, id, repo);
642 		if (err)
643 			return err;
644 		if (is_packed)
645 			return NULL;
646 	}
647 
648 	err = got_object_open_as_tag(&tag, repo, id);
649 	if (err)
650 		return err;
651 
652 	err = add_meta(v, idset, id, "", GOT_OBJ_TYPE_TAG,
653 	    got_object_tag_get_tagger_time(tag),
654 	    loose_obj_only, repo);
655 	if (err)
656 		goto done;
657 
658 	switch (got_object_tag_get_object_type(tag)) {
659 	case GOT_OBJ_TYPE_COMMIT:
660 		err = load_commit(v, idset,
661 		    got_object_tag_get_object_id(tag), repo,
662 		    loose_obj_only, cancel_cb, cancel_arg);
663 		break;
664 	case GOT_OBJ_TYPE_TREE:
665 		err = load_tree(v, idset, got_object_tag_get_object_id(tag),
666 		    "", got_object_tag_get_tagger_time(tag),
667 		    loose_obj_only, repo, cancel_cb, cancel_arg);
668 		break;
669 	default:
670 		break;
671 	}
672 
673 done:
674 	got_object_tag_close(tag);
675 	return err;
676 }
677 
678 enum findtwixt_color {
679 	COLOR_KEEP = 0,
680 	COLOR_DROP,
681 	COLOR_BLANK,
682 };
683 static const int findtwixt_colors[] = {
684 	COLOR_KEEP,
685 	COLOR_DROP,
686 	COLOR_BLANK
687 };
688 
689 static const struct got_error *
690 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
691     int color, struct got_repository *repo)
692 {
693 	const struct got_error *err;
694 	struct got_object_qid *qid;
695 
696 	err = got_object_qid_alloc(&qid, id);
697 	if (err)
698 		return err;
699 
700 	STAILQ_INSERT_TAIL(ids, qid, entry);
701 	qid->data = (void *)&findtwixt_colors[color];
702 	return NULL;
703 }
704 
705 static const struct got_error *
706 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
707     struct got_object_id *id, struct got_repository *repo,
708     got_cancel_cb cancel_cb, void *cancel_arg)
709 {
710 	const struct got_error *err = NULL;
711 	struct got_commit_object *commit;
712 	const struct got_object_id_queue *parents;
713 	struct got_object_id_queue ids;
714 	struct got_object_qid *qid;
715 
716 	STAILQ_INIT(&ids);
717 
718 	err = got_object_qid_alloc(&qid, id);
719 	if (err)
720 		return err;
721 	STAILQ_INSERT_HEAD(&ids, qid, entry);
722 
723 	while (!STAILQ_EMPTY(&ids)) {
724 		if (cancel_cb) {
725 			err = (*cancel_cb)(cancel_arg);
726 			if (err)
727 				break;
728 		}
729 
730 		qid = STAILQ_FIRST(&ids);
731 		STAILQ_REMOVE_HEAD(&ids, entry);
732 
733 		if (got_object_idset_contains(drop, qid->id)) {
734 			got_object_qid_free(qid);
735 			continue;
736 		}
737 
738 		err = got_object_idset_add(drop, qid->id,
739 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
740 		if (err) {
741 			got_object_qid_free(qid);
742 			break;
743 		}
744 
745 		if (!got_object_idset_contains(keep, qid->id)) {
746 			got_object_qid_free(qid);
747 			continue;
748 		}
749 
750 		err = got_object_open_as_commit(&commit, repo, qid->id);
751 		got_object_qid_free(qid);
752 		if (err)
753 			break;
754 
755 		parents = got_object_commit_get_parent_ids(commit);
756 		if (parents) {
757 			err = got_object_id_queue_copy(parents, &ids);
758 			if (err) {
759 				got_object_commit_close(commit);
760 				break;
761 			}
762 		}
763 		got_object_commit_close(commit);
764 	}
765 
766 	got_object_id_queue_free(&ids);
767 	return err;
768 }
769 
770 struct append_id_arg {
771 	struct got_object_id **array;
772 	int idx;
773 };
774 
775 static const struct got_error *
776 append_id(struct got_object_id *id, void *data, void *arg)
777 {
778 	struct append_id_arg *a = arg;
779 
780 	a->array[a->idx] = got_object_id_dup(id);
781 	if (a->array[a->idx] == NULL)
782 		return got_error_from_errno("got_object_id_dup");
783 
784 	a->idx++;
785 	return NULL;
786 }
787 
788 static const struct got_error *
789 findtwixt(struct got_object_id ***res, int *nres,
790     struct got_object_id **head, int nhead,
791     struct got_object_id **tail, int ntail,
792     struct got_repository *repo,
793     got_cancel_cb cancel_cb, void *cancel_arg)
794 {
795 	const struct got_error *err = NULL;
796 	struct got_object_id_queue ids;
797 	struct got_object_idset *keep, *drop;
798 	struct got_object_qid *qid;
799 	int i, ncolor, nkeep, obj_type;
800 
801 	STAILQ_INIT(&ids);
802 	*res = NULL;
803 	*nres = 0;
804 
805 	keep = got_object_idset_alloc();
806 	if (keep == NULL)
807 		return got_error_from_errno("got_object_idset_alloc");
808 
809 	drop = got_object_idset_alloc();
810 	if (drop == NULL) {
811 		err = got_error_from_errno("got_object_idset_alloc");
812 		goto done;
813 	}
814 
815 	for (i = 0; i < nhead; i++) {
816 		struct got_object_id *id = head[i];
817 		if (id == NULL)
818 			continue;
819 		err = got_object_get_type(&obj_type, repo, id);
820 		if (err)
821 			return err;
822 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
823 			continue;
824 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
825 		if (err)
826 			goto done;
827 	}		
828 	for (i = 0; i < ntail; i++) {
829 		struct got_object_id *id = tail[i];
830 		if (id == NULL)
831 			continue;
832 		err = got_object_get_type(&obj_type, repo, id);
833 		if (err)
834 			return err;
835 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
836 			continue;
837 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
838 		if (err)
839 			goto done;
840 	}
841 
842 	while (!STAILQ_EMPTY(&ids)) {
843 		int qcolor;
844 		qid = STAILQ_FIRST(&ids);
845 		qcolor = *((int *)qid->data);
846 
847 		if (got_object_idset_contains(drop, qid->id))
848 			ncolor = COLOR_DROP;
849 		else if (got_object_idset_contains(keep, qid->id))
850 			ncolor = COLOR_KEEP;
851 		else
852 			ncolor = COLOR_BLANK;
853 
854 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
855 		    qcolor == COLOR_KEEP)) {
856 			STAILQ_REMOVE_HEAD(&ids, entry);
857 			got_object_qid_free(qid);
858 			continue;
859 		}
860 
861 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
862 			err = drop_commit(keep, drop, qid->id, repo,
863 			    cancel_cb, cancel_arg);
864 			if (err)
865 				goto done;
866 		} else if (ncolor == COLOR_BLANK) {
867 			struct got_commit_object *commit;
868 			struct got_object_id *id;
869 			const struct got_object_id_queue *parents;
870 			struct got_object_qid *pid;
871 
872 			id = got_object_id_dup(qid->id);
873 			if (id == NULL) {
874 				err = got_error_from_errno("got_object_id_dup");
875 				goto done;
876 			}
877 			if (qcolor == COLOR_KEEP)
878 				err = got_object_idset_add(keep, id,
879 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
880 			else
881 				err = got_object_idset_add(drop, id,
882 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
883 			if (err) {
884 				free(id);
885 				goto done;
886 			}
887 
888 			err = got_object_open_as_commit(&commit, repo, id);
889 			if (err) {
890 				free(id);
891 				goto done;
892 			}
893 			parents = got_object_commit_get_parent_ids(commit);
894 			if (parents) {
895 				STAILQ_FOREACH(pid, parents, entry) {
896 					err = queue_commit_id(&ids, pid->id,
897 					    qcolor, repo);
898 					if (err) {
899 						free(id);
900 						goto done;
901 					}
902 				}
903 			}
904 			got_object_commit_close(commit);
905 			commit = NULL;
906 		} else {
907 			/* should not happen */
908 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
909 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
910 			goto done;
911 		}
912 
913 		STAILQ_REMOVE_HEAD(&ids, entry);
914 		got_object_qid_free(qid);
915 	}
916 
917 	nkeep = got_object_idset_num_elements(keep);
918 	if (nkeep > 0) {
919 		struct append_id_arg arg;
920 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
921 		if (arg.array == NULL) {
922 			err = got_error_from_errno("calloc");
923 			goto done;
924 		}
925 		arg.idx = 0;
926 		err = got_object_idset_for_each(keep, append_id, &arg);
927 		if (err) {
928 			free(arg.array);
929 			goto done;
930 		}
931 		*res = arg.array;
932 		*nres = nkeep;
933 	}
934 done:
935 	got_object_idset_free(keep);
936 	got_object_idset_free(drop);
937 	got_object_id_queue_free(&ids);
938 	return err;
939 }
940 
941 static const struct got_error *
942 read_meta(struct got_pack_meta ***meta, int *nmeta,
943     struct got_object_id **theirs, int ntheirs,
944     struct got_object_id **ours, int nours, struct got_repository *repo,
945     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
946     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
947 {
948 	const struct got_error *err = NULL;
949 	struct got_object_id **ids = NULL;
950 	struct got_object_idset *idset;
951 	int i, nobj = 0, obj_type;
952 	struct got_pack_metavec v;
953 
954 	*meta = NULL;
955 	*nmeta = 0;
956 
957 	idset = got_object_idset_alloc();
958 	if (idset == NULL)
959 		return got_error_from_errno("got_object_idset_alloc");
960 
961 	v.nmeta = 0;
962 	v.metasz = 64;
963 	v.meta = calloc(v.metasz, sizeof(struct got_pack_meta *));
964 	if (v.meta == NULL) {
965 		err = got_error_from_errno("calloc");
966 		goto done;
967 	}
968 
969 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
970 	    cancel_cb, cancel_arg);
971 	if (err || nobj == 0)
972 		goto done;
973 
974 	for (i = 0; i < ntheirs; i++) {
975 		struct got_object_id *id = theirs[i];
976 		if (id == NULL)
977 			continue;
978 		err = got_object_get_type(&obj_type, repo, id);
979 		if (err)
980 			return err;
981 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
982 			continue;
983 		err = load_commit(NULL, idset, id, repo,
984 		    loose_obj_only, cancel_cb, cancel_arg);
985 		if (err)
986 			goto done;
987 		err = report_progress(progress_cb, progress_arg, rl,
988 		    0L, nours, v.nmeta, 0, 0);
989 		if (err)
990 			goto done;
991 	}
992 
993 	for (i = 0; i < ntheirs; i++) {
994 		struct got_object_id *id = theirs[i];
995 		int *cached_type;
996 		if (id == NULL)
997 			continue;
998 		cached_type = got_object_idset_get(idset, id);
999 		if (cached_type == NULL) {
1000 			err = got_object_get_type(&obj_type, repo, id);
1001 			if (err)
1002 				goto done;
1003 		} else
1004 			obj_type = *cached_type;
1005 		if (obj_type != GOT_OBJ_TYPE_TAG)
1006 			continue;
1007 		err = load_tag(NULL, idset, id, repo,
1008 		    loose_obj_only, cancel_cb, cancel_arg);
1009 		if (err)
1010 			goto done;
1011 		err = report_progress(progress_cb, progress_arg, rl,
1012 		    0L, nours, v.nmeta, 0, 0);
1013 		if (err)
1014 			goto done;
1015 	}
1016 
1017 	for (i = 0; i < nobj; i++) {
1018 		err = load_commit(&v, idset, ids[i], repo,
1019 		    loose_obj_only, cancel_cb, cancel_arg);
1020 		if (err)
1021 			goto done;
1022 		if (err)
1023 			goto done;
1024 		err = report_progress(progress_cb, progress_arg, rl,
1025 		    0L, nours, v.nmeta, 0, 0);
1026 		if (err)
1027 			goto done;
1028 	}
1029 
1030 	for (i = 0; i < nours; i++) {
1031 		struct got_object_id *id = ours[i];
1032 		int *cached_type;
1033 		if (id == NULL)
1034 			continue;
1035 		cached_type = got_object_idset_get(idset, id);
1036 		if (cached_type == NULL) {
1037 			err = got_object_get_type(&obj_type, repo, id);
1038 			if (err)
1039 				goto done;
1040 		} else
1041 			obj_type = *cached_type;
1042 		if (obj_type != GOT_OBJ_TYPE_TAG)
1043 			continue;
1044 		err = load_tag(&v, idset, id, repo,
1045 		    loose_obj_only, cancel_cb, cancel_arg);
1046 		if (err)
1047 			goto done;
1048 		err = report_progress(progress_cb, progress_arg, rl,
1049 		    0L, nours, v.nmeta, 0, 0);
1050 		if (err)
1051 			goto done;
1052 	}
1053 
1054 	if (progress_cb) {
1055 		err = progress_cb(progress_arg, 0L, nours, v.nmeta, 0, 0);
1056 		if (err)
1057 			goto done;
1058 	}
1059 done:
1060 	for (i = 0; i < nobj; i++) {
1061 		free(ids[i]);
1062 	}
1063 	free(ids);
1064 	got_object_idset_free(idset);
1065 	if (err == NULL) {
1066 		*meta = v.meta;
1067 		*nmeta = v.nmeta;
1068 	} else
1069 		free(v.meta);
1070 
1071 	return err;
1072 }
1073 
1074 const struct got_error *
1075 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1076 {
1077 	size_t n;
1078 
1079 	SHA1Update(ctx, buf, len);
1080 	n = fwrite(buf, 1, len, f);
1081 	if (n != len)
1082 		return got_ferror(f, GOT_ERR_IO);
1083 	return NULL;
1084 }
1085 
1086 static void
1087 putbe32(char *b, uint32_t n)
1088 {
1089 	b[0] = n >> 24;
1090 	b[1] = n >> 16;
1091 	b[2] = n >> 8;
1092 	b[3] = n >> 0;
1093 }
1094 
1095 static int
1096 write_order_cmp(const void *pa, const void *pb)
1097 {
1098 	struct got_pack_meta *a, *b, *ahd, *bhd;
1099 
1100 	a = *(struct got_pack_meta **)pa;
1101 	b = *(struct got_pack_meta **)pb;
1102 	ahd = (a->head == NULL) ? a : a->head;
1103 	bhd = (b->head == NULL) ? b : b->head;
1104 	if (ahd->mtime != bhd->mtime)
1105 		return bhd->mtime - ahd->mtime;
1106 	if (ahd != bhd)
1107 		return (uintptr_t)bhd - (uintptr_t)ahd;
1108 	if (a->nchain != b->nchain)
1109 		return a->nchain - b->nchain;
1110 	return a->mtime - b->mtime;
1111 }
1112 
1113 static const struct got_error *
1114 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1115 {
1116 	size_t i;
1117 
1118 	*hdrlen = 0;
1119 
1120 	hdr[0] = obj_type << 4;
1121 	hdr[0] |= len & 0xf;
1122 	len >>= 4;
1123 	for (i = 1; len != 0; i++){
1124 		if (i >= bufsize)
1125 			return got_error(GOT_ERR_NO_SPACE);
1126 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1127 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1128 		len >>= GOT_DELTA_SIZE_SHIFT;
1129 	}
1130 
1131 	*hdrlen = i;
1132 	return NULL;
1133 }
1134 
1135 static int
1136 packoff(char *hdr, off_t off)
1137 {
1138 	int i, j;
1139 	char rbuf[8];
1140 
1141 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1142 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1143 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1144 		    GOT_DELTA_SIZE_MORE;
1145 	}
1146 
1147 	j = 0;
1148 	while (i > 0)
1149 		hdr[j++] = rbuf[--i];
1150 	return j;
1151 }
1152 
1153 static const struct got_error *
1154 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1155     struct got_pack_meta **meta, int nmeta, int nours,
1156     int use_offset_deltas, struct got_repository *repo,
1157     got_pack_progress_cb progress_cb, void *progress_arg,
1158     struct got_ratelimit *rl,
1159     got_cancel_cb cancel_cb, void *cancel_arg)
1160 {
1161 	const struct got_error *err = NULL;
1162 	int i, nh;
1163 	SHA1_CTX ctx;
1164 	struct got_pack_meta *m;
1165 	struct got_raw_object *raw = NULL;
1166 	FILE *delta_file = NULL;
1167 	char buf[32];
1168 	size_t outlen, n;
1169 	struct got_deflate_checksum csum;
1170 	off_t packfile_size = 0;
1171 	int outfd = -1;
1172 
1173 	SHA1Init(&ctx);
1174 	csum.output_sha1 = &ctx;
1175 	csum.output_crc = NULL;
1176 
1177 	err = hwrite(packfile, "PACK", 4, &ctx);
1178 	if (err)
1179 		return err;
1180 	putbe32(buf, GOT_PACKFILE_VERSION);
1181 	err = hwrite(packfile, buf, 4, &ctx);
1182 	if (err)
1183 		goto done;
1184 	putbe32(buf, nmeta);
1185 	err = hwrite(packfile, buf, 4, &ctx);
1186 	if (err)
1187 		goto done;
1188 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), write_order_cmp);
1189 	for (i = 0; i < nmeta; i++) {
1190 		err = report_progress(progress_cb, progress_arg, rl,
1191 		    packfile_size, nours, nmeta, nmeta, i);
1192 		if (err)
1193 			goto done;
1194 		m = meta[i];
1195 		m->off = ftello(packfile);
1196 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
1197 		if (err)
1198 			goto done;
1199 		if (m->delta_len == 0) {
1200 			err = packhdr(&nh, buf, sizeof(buf),
1201 			    m->obj_type, raw->size);
1202 			if (err)
1203 				goto done;
1204 			err = hwrite(packfile, buf, nh, &ctx);
1205 			if (err)
1206 				goto done;
1207 			packfile_size += nh;
1208 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET) == -1) {
1209 				err = got_error_from_errno("fseeko");
1210 				goto done;
1211 			}
1212 			err = got_deflate_to_file(&outlen, raw->f, packfile,
1213 			    &csum);
1214 			if (err)
1215 				goto done;
1216 			packfile_size += outlen;
1217 		} else {
1218 			off_t remain;
1219 			if (delta_file == NULL) {
1220 				delta_file = got_opentemp();
1221 				if (delta_file == NULL) {
1222 					err = got_error_from_errno(
1223 					    "got_opentemp");
1224 					goto done;
1225 				}
1226 			}
1227 			if (ftruncate(fileno(delta_file), 0L) == -1) {
1228 				err = got_error_from_errno("ftruncate");
1229 				goto done;
1230 			}
1231 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1232 				err = got_error_from_errno("fseeko");
1233 				goto done;
1234 			}
1235 			if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1236 			    == -1) {
1237 				err = got_error_from_errno("fseeko");
1238 				goto done;
1239 			}
1240 			remain = m->delta_len;
1241 			while (remain > 0) {
1242 				char delta_buf[8192];
1243 				size_t r, w, n;
1244 				n = MIN(remain, sizeof(delta_buf));
1245 				r = fread(delta_buf, 1, n, delta_cache);
1246 				if (r != n) {
1247 					err = got_ferror(delta_cache,
1248 					    GOT_ERR_IO);
1249 					goto done;
1250 				}
1251 				w = fwrite(delta_buf, 1, n, delta_file);
1252 				if (w != n) {
1253 					err = got_ferror(delta_file,
1254 					    GOT_ERR_IO);
1255 					goto done;
1256 				}
1257 				remain -= n;
1258 			}
1259 			if (use_offset_deltas && m->prev->off != 0) {
1260 				err = packhdr(&nh, buf, sizeof(buf),
1261 				    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1262 				if (err)
1263 					goto done;
1264 				nh += packoff(buf + nh,
1265 				    m->off - m->prev->off);
1266 				err = hwrite(packfile, buf, nh, &ctx);
1267 				if (err)
1268 					goto done;
1269 				packfile_size += nh;
1270 			} else {
1271 				err = packhdr(&nh, buf, sizeof(buf),
1272 				    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1273 				err = hwrite(packfile, buf, nh, &ctx);
1274 				if (err)
1275 					goto done;
1276 				packfile_size += nh;
1277 				err = hwrite(packfile, m->prev->id.sha1,
1278 				    sizeof(m->prev->id.sha1), &ctx);
1279 				packfile_size += sizeof(m->prev->id.sha1);
1280 				if (err)
1281 					goto done;
1282 			}
1283 			if (fseeko(delta_file, 0L, SEEK_SET) == -1) {
1284 				err = got_error_from_errno("fseeko");
1285 				goto done;
1286 			}
1287 			err = got_deflate_to_file(&outlen, delta_file,
1288 			    packfile, &csum);
1289 			if (err)
1290 				goto done;
1291 			packfile_size += outlen;
1292 		}
1293 		got_object_raw_close(raw);
1294 		raw = NULL;
1295 	}
1296 	SHA1Final(pack_sha1, &ctx);
1297 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1298 	if (n != SHA1_DIGEST_LENGTH)
1299 		err = got_ferror(packfile, GOT_ERR_IO);
1300 	packfile_size += SHA1_DIGEST_LENGTH;
1301 	packfile_size += sizeof(struct got_packfile_hdr);
1302 	if (progress_cb) {
1303 		err = progress_cb(progress_arg, packfile_size, nours,
1304 		    nmeta, nmeta, nmeta);
1305 		if (err)
1306 			goto done;
1307 	}
1308 done:
1309 	if (delta_file && fclose(delta_file) == EOF && err == NULL)
1310 		err = got_error_from_errno("fclose");
1311 	if (raw)
1312 		got_object_raw_close(raw);
1313 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1314 		err = got_error_from_errno("close");
1315 	return err;
1316 }
1317 
1318 const struct got_error *
1319 got_pack_create(uint8_t *packsha1, FILE *packfile,
1320     struct got_object_id **theirs, int ntheirs,
1321     struct got_object_id **ours, int nours,
1322     struct got_repository *repo, int loose_obj_only, int allow_empty,
1323     got_pack_progress_cb progress_cb, void *progress_arg,
1324     got_cancel_cb cancel_cb, void *cancel_arg)
1325 {
1326 	const struct got_error *err;
1327 	struct got_pack_meta **meta;
1328 	int nmeta;
1329 	FILE *delta_cache = NULL;
1330 	struct got_ratelimit rl;
1331 
1332 	got_ratelimit_init(&rl, 0, 500);
1333 
1334 	err = read_meta(&meta, &nmeta, theirs, ntheirs, ours, nours, repo,
1335 	    loose_obj_only, progress_cb, progress_arg, &rl,
1336 	    cancel_cb, cancel_arg);
1337 	if (err)
1338 		return err;
1339 
1340 	if (nmeta == 0 && !allow_empty) {
1341 		err = got_error(GOT_ERR_CANNOT_PACK);
1342 		goto done;
1343 	}
1344 
1345 	delta_cache = got_opentemp();
1346 	if (delta_cache == NULL) {
1347 		err = got_error_from_errno("got_opentemp");
1348 		goto done;
1349 	}
1350 
1351 	if (nmeta > 0) {
1352 		err = pick_deltas(meta, nmeta, nours, delta_cache, repo,
1353 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1354 		if (err)
1355 			goto done;
1356 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1357 			err = got_error_from_errno("fseeko");
1358 			goto done;
1359 		}
1360 	}
1361 
1362 	err = genpack(packsha1, packfile, delta_cache, meta, nmeta, nours, 1,
1363 	    repo, progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1364 	if (err)
1365 		goto done;
1366 done:
1367 	free_nmeta(meta, nmeta);
1368 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1369 		err = got_error_from_errno("fclose");
1370 	return err;
1371 }