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 <endian.h>
26 #include <stdint.h>
27 #include <imsg.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <sha1.h>
32 #include <time.h>
33 #include <unistd.h>
34 #include <limits.h>
35 #include <zlib.h>
36 
37 #include "got_error.h"
38 #include "got_cancel.h"
39 #include "got_object.h"
40 #include "got_path.h"
41 #include "got_reference.h"
42 #include "got_repository_admin.h"
43 #include "got_opentemp.h"
44 
45 #include "got_lib_deltify.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_idset.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_deflate.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_repository.h"
54 #include "got_lib_ratelimit.h"
55 #include "got_lib_inflate.h"
56 
57 #ifndef MIN
58 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
59 #endif
60 
61 #ifndef MAX
62 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
63 #endif
64 
65 #ifndef nitems
66 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
67 #endif
68 
69 struct got_pack_meta {
70 	struct got_object_id id;
71 	char	*path;
72 	int	obj_type;
73 	off_t	size;
74 	time_t	mtime;
75 
76 	/* The best delta we picked */
77 	struct got_pack_meta *head;
78 	struct got_pack_meta *prev;
79 	unsigned char *delta_buf; /* if encoded in memory (compressed) */
80 	off_t	delta_offset;	/* offset in delta cache file (compressed) */
81 	off_t	delta_len;	/* encoded delta length */
82 	off_t	delta_compressed_len; /* encoded+compressed delta length */
83 	int	nchain;
84 
85 	int	have_reused_delta;
86 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
87 	struct got_object_id *base_obj_id;
88 
89 	/* Only used for delta window */
90 	struct got_delta_table *dtab;
91 
92 	/* Only used for writing offset deltas */
93 	off_t	off;
94 };
95 
96 struct got_pack_metavec {
97 	struct got_pack_meta **meta;
98 	int nmeta;
99 	int metasz;
100 };
101 
102 static const struct got_error *
103 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
104     const char *path, int obj_type, time_t mtime)
105 {
106 	const struct got_error *err = NULL;
107 	struct got_pack_meta *m;
108 
109 	*new = NULL;
110 
111 	m = calloc(1, sizeof(*m));
112 	if (m == NULL)
113 		return got_error_from_errno("calloc");
114 
115 	memcpy(&m->id, id, sizeof(m->id));
116 
117 	m->path = strdup(path);
118 	if (m->path == NULL) {
119 		err = got_error_from_errno("strdup");
120 		free(m);
121 		return err;
122 	}
123 
124 	m->obj_type = obj_type;
125 	m->mtime = mtime;
126 	*new = m;
127 	return NULL;
128 }
129 
130 static void
131 clear_meta(struct got_pack_meta *meta)
132 {
133 	if (meta == NULL)
134 		return;
135 	free(meta->path);
136 	meta->path = NULL;
137 	free(meta->delta_buf);
138 	meta->delta_buf = NULL;
139 	free(meta->base_obj_id);
140 	meta->base_obj_id = NULL;
141 }
142 
143 static void
144 free_nmeta(struct got_pack_meta **meta, int nmeta)
145 {
146 	int i;
147 
148 	for (i = 0; i < nmeta; i++)
149 		clear_meta(meta[i]);
150 	free(meta);
151 }
152 
153 static int
154 delta_order_cmp(const void *pa, const void *pb)
155 {
156 	struct got_pack_meta *a, *b;
157 	int cmp;
158 
159 	a = *(struct got_pack_meta **)pa;
160 	b = *(struct got_pack_meta **)pb;
161 
162 	if (a->obj_type != b->obj_type)
163 		return a->obj_type - b->obj_type;
164 	cmp = strcmp(a->path, b->path);
165 	if (cmp != 0)
166 		return cmp;
167 	if (a->mtime < b->mtime)
168 		return -1;
169 	if (a->mtime > b->mtime)
170 		return 1;
171 	return got_object_id_cmp(&a->id, &b->id);
172 }
173 
174 static off_t
175 delta_size(struct got_delta_instruction *deltas, int ndeltas)
176 {
177 	int i;
178 	off_t size = 32;
179 	for (i = 0; i < ndeltas; i++) {
180 		if (deltas[i].copy)
181 			size += GOT_DELTA_SIZE_SHIFT;
182 		else
183 			size += deltas[i].len + 1;
184 	}
185 	return size;
186 }
187 
188 static const struct got_error *
189 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
190 {
191 	char *n;
192 
193 	if (*len + nseg >= *sz) {
194 		while (*len + nseg >= *sz)
195 			*sz += *sz / 2;
196 		n = realloc(*p, *sz);
197 		if (n == NULL)
198 			return got_error_from_errno("realloc");
199 		*p = n;
200 	}
201 	memcpy(*p + *len, seg, nseg);
202 	*len += nseg;
203 	return NULL;
204 }
205 
206 static const struct got_error *
207 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
208     struct got_delta_instruction *deltas, int ndeltas,
209     off_t delta_size, off_t base_size)
210 {
211 	const struct got_error *err;
212 	unsigned char buf[16], *bp;
213 	int i, j;
214 	size_t len = 0, compressed_len;
215 	off_t bufsize = delta_size;
216 	off_t n;
217 	struct got_delta_instruction *d;
218 	uint8_t *delta_buf;
219 
220 	delta_buf = malloc(bufsize);
221 	if (delta_buf == NULL)
222 		return got_error_from_errno("malloc");
223 
224 	/* base object size */
225 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
226 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
227 	for (i = 1; n > 0; i++) {
228 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
229 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
230 		n >>= GOT_DELTA_SIZE_SHIFT;
231 	}
232 	err = append(&delta_buf, &len, &bufsize, buf, i);
233 	if (err)
234 		goto done;
235 
236 	/* target object size */
237 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
238 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
239 	for (i = 1; n > 0; i++) {
240 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
241 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
242 		n >>= GOT_DELTA_SIZE_SHIFT;
243 	}
244 	err = append(&delta_buf, &len, &bufsize, buf, i);
245 	if (err)
246 		goto done;
247 
248 	for (j = 0; j < ndeltas; j++) {
249 		d = &deltas[j];
250 		if (d->copy) {
251 			n = d->offset;
252 			bp = &buf[1];
253 			buf[0] = GOT_DELTA_BASE_COPY;
254 			for (i = 0; i < 4; i++) {
255 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
256 				buf[0] |= 1 << i;
257 				*bp++ = n & 0xff;
258 				n >>= 8;
259 				if (n == 0)
260 					break;
261 			}
262 
263 			n = d->len;
264 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
265 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
266 				for (i = 0; i < 3 && n > 0; i++) {
267 					buf[0] |= 1 << (i + 4);
268 					*bp++ = n & 0xff;
269 					n >>= 8;
270 				}
271 			}
272 			err = append(&delta_buf, &len, &bufsize,
273 			    buf, bp - buf);
274 			if (err)
275 				goto done;
276 		} else if (o->f == NULL) {
277 			n = 0;
278 			while (n != d->len) {
279 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
280 				err = append(&delta_buf, &len, &bufsize,
281 				    buf, 1);
282 				if (err)
283 					goto done;
284 				err = append(&delta_buf, &len, &bufsize,
285 				    o->data + o->hdrlen + d->offset + n,
286 				    buf[0]);
287 				if (err)
288 					goto done;
289 				n += buf[0];
290 			}
291 		} else {
292 			char content[128];
293 			size_t r;
294 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
295 				err = got_error_from_errno("fseeko");
296 				goto done;
297 			}
298 			n = 0;
299 			while (n != d->len) {
300 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
301 				err = append(&delta_buf, &len, &bufsize,
302 				    buf, 1);
303 				if (err)
304 					goto done;
305 				r = fread(content, 1, buf[0], o->f);
306 				if (r != buf[0]) {
307 					err = got_ferror(o->f, GOT_ERR_IO);
308 					goto done;
309 				}
310 				err = append(&delta_buf, &len, &bufsize,
311 				    content, buf[0]);
312 				if (err)
313 					goto done;
314 				n += buf[0];
315 			}
316 		}
317 	}
318 
319 	err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
320 	    NULL, NULL, delta_buf, 0, len);
321 	if (err)
322 		goto done;
323 
324 	m->delta_len = len;
325 	m->delta_compressed_len = compressed_len;
326 done:
327 	free(delta_buf);
328 	return err;
329 }
330 
331 static const struct got_error *
332 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
333     struct got_delta_instruction *deltas, int ndeltas,
334     off_t base_size, FILE *f)
335 {
336 	const struct got_error *err;
337 	unsigned char buf[16], *bp;
338 	int i, j;
339 	off_t n;
340 	struct got_deflate_buf zb;
341 	struct got_delta_instruction *d;
342 	off_t delta_len = 0, compressed_len = 0;
343 
344 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
345 	if (err)
346 		return err;
347 
348 	/* base object size */
349 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
350 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
351 	for (i = 1; n > 0; i++) {
352 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
353 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
354 		n >>= GOT_DELTA_SIZE_SHIFT;
355 	}
356 
357 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
358 	    buf, 0, i, f, NULL);
359 	if (err)
360 		goto done;
361 	delta_len += i;
362 
363 	/* target object size */
364 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
365 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
366 	for (i = 1; n > 0; i++) {
367 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
368 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
369 		n >>= GOT_DELTA_SIZE_SHIFT;
370 	}
371 
372 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
373 	    buf, 0, i, f, NULL);
374 	if (err)
375 		goto done;
376 	delta_len += i;
377 
378 	for (j = 0; j < ndeltas; j++) {
379 		d = &deltas[j];
380 		if (d->copy) {
381 			n = d->offset;
382 			bp = &buf[1];
383 			buf[0] = GOT_DELTA_BASE_COPY;
384 			for (i = 0; i < 4; i++) {
385 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
386 				buf[0] |= 1 << i;
387 				*bp++ = n & 0xff;
388 				n >>= 8;
389 				if (n == 0)
390 					break;
391 			}
392 			n = d->len;
393 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
394 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
395 				for (i = 0; i < 3 && n > 0; i++) {
396 					buf[0] |= 1 << (i + 4);
397 					*bp++ = n & 0xff;
398 					n >>= 8;
399 				}
400 			}
401 			err = got_deflate_append_to_file_mmap(&zb,
402 			    &compressed_len, buf, 0, bp - buf, f, NULL);
403 			if (err)
404 				goto done;
405 			delta_len += (bp - buf);
406 		} else if (o->f == NULL) {
407 			n = 0;
408 			while (n != d->len) {
409 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
410 				err = got_deflate_append_to_file_mmap(&zb,
411 				    &compressed_len, buf, 0, 1, f, NULL);
412 				if (err)
413 					goto done;
414 				delta_len++;
415 				err = got_deflate_append_to_file_mmap(&zb,
416 				    &compressed_len,
417 				    o->data + o->hdrlen + d->offset + n, 0,
418 				    buf[0], f, NULL);
419 				if (err)
420 					goto done;
421 				delta_len += buf[0];
422 				n += buf[0];
423 			}
424 		} else {
425 			char content[128];
426 			size_t r;
427 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
428 				err = got_error_from_errno("fseeko");
429 				goto done;
430 			}
431 			n = 0;
432 			while (n != d->len) {
433 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
434 				err = got_deflate_append_to_file_mmap(&zb,
435 				    &compressed_len, buf, 0, 1, f, NULL);
436 				if (err)
437 					goto done;
438 				delta_len++;
439 				r = fread(content, 1, buf[0], o->f);
440 				if (r != buf[0]) {
441 					err = got_ferror(o->f, GOT_ERR_IO);
442 					goto done;
443 				}
444 				err = got_deflate_append_to_file_mmap(&zb,
445 				    &compressed_len, content, 0, buf[0], f,
446 				    NULL);
447 				if (err)
448 					goto done;
449 				delta_len += buf[0];
450 				n += buf[0];
451 			}
452 		}
453 	}
454 
455 	err = got_deflate_flush(&zb, f, NULL, &compressed_len);
456 	if (err)
457 		goto done;
458 
459 	/* sanity check */
460 	if (compressed_len != ftello(f) - m->delta_offset) {
461 		err = got_error(GOT_ERR_COMPRESSION);
462 		goto done;
463 	}
464 
465 	m->delta_len = delta_len;
466 	m->delta_compressed_len = compressed_len;
467 done:
468 	got_deflate_end(&zb);
469 	return err;
470 }
471 
472 static const struct got_error *
473 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
474     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
475     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
476     int nobj_written)
477 {
478 	const struct got_error *err;
479 	int elapsed;
480 
481 	if (progress_cb == NULL)
482 		return NULL;
483 
484 	err = got_ratelimit_check(&elapsed, rl);
485 	if (err || !elapsed)
486 		return err;
487 
488 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
489 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
490 }
491 
492 static const struct got_error *
493 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
494 {
495 	if (v->nmeta == v->metasz){
496 		size_t newsize = 2 * v->metasz;
497 		struct got_pack_meta **new;
498 		new = reallocarray(v->meta, newsize, sizeof(*new));
499 		if (new == NULL)
500 			return got_error_from_errno("reallocarray");
501 		v->meta = new;
502 		v->metasz = newsize; 
503 	}
504 
505 	v->meta[v->nmeta++] = m;
506 	return NULL;
507 }
508 
509 static const struct got_error *
510 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
511     struct got_object_idset *idset, struct got_pack *pack,
512     struct got_packidx *packidx, int delta_cache_fd,
513     struct got_repository *repo)
514 {
515 	const struct got_error *err = NULL;
516 	struct got_pack_meta *base = NULL;
517 	struct got_object_id *base_obj_id = NULL;
518 	off_t delta_len = 0, delta_compressed_len = 0;
519 	off_t delta_offset = 0, delta_cache_offset = 0;
520 	uint64_t base_size, result_size;
521 
522 	if (m->have_reused_delta)
523 		return NULL;
524 
525 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
526 	    &delta_compressed_len, &delta_offset, &delta_cache_offset,
527 	    &base_obj_id, delta_cache_fd, packidx, idx, &m->id, repo);
528 	if (err)
529 		return err;
530 
531 	if (delta_offset + delta_len < delta_offset)
532 		return got_error(GOT_ERR_BAD_PACKFILE);
533 
534 	base = got_object_idset_get(idset, base_obj_id);
535 	if (base == NULL)
536 		goto done;
537 
538 	m->delta_len = delta_len;
539 	m->delta_compressed_len = delta_compressed_len;
540 	m->delta_offset = delta_cache_offset;
541 	m->prev = base;
542 	m->size = result_size;
543 	m->have_reused_delta = 1;
544 	m->reused_delta_offset = delta_offset;
545 	m->base_obj_id = base_obj_id;
546 	base_obj_id = NULL;
547 	err = add_meta(m, v);
548 done:
549 	free(base_obj_id);
550 	return err;
551 }
552 
553 static const struct got_error *
554 find_pack_for_reuse(struct got_packidx **best_packidx,
555     struct got_repository *repo)
556 {
557 	const struct got_error *err = NULL;
558 	struct got_pathlist_entry *pe;
559 	const char *best_packidx_path = NULL;
560 	int nobj_max = 0;
561 
562 	*best_packidx = NULL;
563 
564 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
565 		const char *path_packidx = pe->path;
566 		struct got_packidx *packidx;
567 		int nobj;
568 
569 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
570 		if (err)
571 			break;
572 
573 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
574 		if (nobj > nobj_max) {
575 			best_packidx_path = path_packidx;
576 			nobj_max = nobj;
577 		}
578 	}
579 
580 	if (best_packidx_path) {
581 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
582 		    repo);
583 	}
584 
585 	return err;
586 }
587 
588 struct search_deltas_arg {
589 	struct got_packidx *packidx;
590 	struct got_pack *pack;
591 	struct got_object_idset *idset;
592 	struct got_pack_metavec *v;
593 	int delta_cache_fd;
594 	struct got_repository *repo;
595 	got_pack_progress_cb progress_cb;
596 	void *progress_arg;
597 	struct got_ratelimit *rl;
598 	got_cancel_cb cancel_cb;
599 	void *cancel_arg;
600 	int ncolored;
601 	int nfound;
602 	int ntrees;
603 	int ncommits;
604 };
605 
606 static const struct got_error *
607 search_delta_for_object(struct got_object_id *id, void *data, void *arg)
608 {
609 	const struct got_error *err;
610 	struct got_pack_meta *m = data;
611 	struct search_deltas_arg *a = arg;
612 	int obj_idx;
613 	struct got_object *obj = NULL;
614 
615 	if (a->cancel_cb) {
616 		err = (*a->cancel_cb)(a->cancel_arg);
617 		if (err)
618 			return err;
619 	}
620 
621 	if (!got_repo_check_packidx_bloom_filter(a->repo,
622 	    a->packidx->path_packidx, id))
623 		return NULL;
624 
625 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
626 	if (obj_idx == -1)
627 		return NULL;
628 
629 	/* TODO:
630 	 * Opening and closing an object just to check its flags
631 	 * is a bit expensive. We could have an imsg which requests
632 	 * plain type/size information for an object without doing
633 	 * work such as traversing the object's entire delta chain
634 	 * to find the base object type, and other such info which
635 	 * we don't really need here.
636 	 */
637 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
638 	    a->packidx, obj_idx, a->repo);
639 	if (err)
640 		return err;
641 
642 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
643 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
644 		    a->delta_cache_fd, a->repo);
645 		if (err)
646 			goto done;
647 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
648 		    a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
649 		    got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
650 	}
651 done:
652 	got_object_close(obj);
653 	return err;
654 }
655 
656 static const struct got_error *
657 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
658     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
659     struct got_repository *repo,
660     got_pack_progress_cb progress_cb, void *progress_arg,
661     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
662 {
663 	const struct got_error *err = NULL;
664 	char *path_packfile = NULL;
665 	struct got_packidx *packidx;
666 	struct got_pack *pack;
667 	struct search_deltas_arg sda;
668 
669 	err = find_pack_for_reuse(&packidx, repo);
670 	if (err)
671 		return err;
672 
673 	if (packidx == NULL)
674 		return NULL;
675 
676 	err = got_packidx_get_packfile_path(&path_packfile,
677 	    packidx->path_packidx);
678 	if (err)
679 		return err;
680 
681 	pack = got_repo_get_cached_pack(repo, path_packfile);
682 	if (pack == NULL) {
683 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
684 		if (err)
685 			goto done;
686 	}
687 
688 	sda.packidx = packidx;
689 	sda.pack = pack;
690 	sda.idset = idset;
691 	sda.v = v;
692 	sda.delta_cache_fd = delta_cache_fd;
693 	sda.repo = repo;
694 	sda.progress_cb = progress_cb;
695 	sda.progress_arg = progress_arg;
696 	sda.rl = rl;
697 	sda.cancel_cb = cancel_cb;
698 	sda.cancel_arg = cancel_arg;
699 	sda.ncolored = ncolored;
700 	sda.nfound = nfound;
701 	sda.ntrees = ntrees;
702 	sda.ncommits = ncommits;
703 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
704 done:
705 	free(path_packfile);
706 	return err;
707 }
708 
709 static const struct got_error *
710 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
711     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
712     struct got_repository *repo,
713     got_pack_progress_cb progress_cb, void *progress_arg,
714     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
715 {
716 	const struct got_error *err = NULL;
717 	struct got_pack_meta *m = NULL, *base = NULL;
718 	struct got_raw_object *raw = NULL, *base_raw = NULL;
719 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
720 	int i, j, ndeltas, best_ndeltas;
721 	off_t size, best_size;
722 	const int max_base_candidates = 3;
723 	size_t delta_memsize = 0;
724 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
725 	int outfd = -1;
726 
727 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
728 	for (i = 0; i < nmeta; i++) {
729 		if (cancel_cb) {
730 			err = (*cancel_cb)(cancel_arg);
731 			if (err)
732 				break;
733 		}
734 		err = report_progress(progress_cb, progress_arg, rl,
735 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
736 		    nreused + i, 0);
737 		if (err)
738 			goto done;
739 		m = meta[i];
740 
741 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
742 		    m->obj_type == GOT_OBJ_TYPE_TAG)
743 			continue;
744 
745 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
746 		if (err)
747 			goto done;
748 		m->size = raw->size;
749 
750 		if (raw->f == NULL) {
751 			err = got_deltify_init_mem(&m->dtab, raw->data,
752 			    raw->hdrlen, raw->size + raw->hdrlen);
753 		} else {
754 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
755 			    raw->size + raw->hdrlen);
756 		}
757 		if (err)
758 			goto done;
759 
760 		if (i > max_base_candidates) {
761 			struct got_pack_meta *n = NULL;
762 			n = meta[i - (max_base_candidates + 1)];
763 			got_deltify_free(n->dtab);
764 			n->dtab = NULL;
765 		}
766 
767 		best_size = raw->size;
768 		best_ndeltas = 0;
769 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
770 			if (cancel_cb) {
771 				err = (*cancel_cb)(cancel_arg);
772 				if (err)
773 					goto done;
774 			}
775 			base = meta[j];
776 			/* long chains make unpacking slow, avoid such bases */
777 			if (base->nchain >= 128 ||
778 			    base->obj_type != m->obj_type)
779 				continue;
780 
781 			err = got_object_raw_open(&base_raw, &outfd, repo,
782 			    &base->id);
783 			if (err)
784 				goto done;
785 
786 			if (raw->f == NULL && base_raw->f == NULL) {
787 				err = got_deltify_mem_mem(&deltas, &ndeltas,
788 				    raw->data, raw->hdrlen,
789 				    raw->size + raw->hdrlen,
790 				    base->dtab, base_raw->data,
791 				    base_raw->hdrlen,
792 				    base_raw->size + base_raw->hdrlen);
793 			} else if (raw->f == NULL) {
794 				err = got_deltify_mem_file(&deltas, &ndeltas,
795 				    raw->data, raw->hdrlen,
796 				    raw->size + raw->hdrlen,
797 				    base->dtab, base_raw->f,
798 				    base_raw->hdrlen,
799 				    base_raw->size + base_raw->hdrlen);
800 			} else if (base_raw->f == NULL) {
801 				err = got_deltify_file_mem(&deltas, &ndeltas,
802 				    raw->f, raw->hdrlen,
803 				    raw->size + raw->hdrlen,
804 				    base->dtab, base_raw->data,
805 				    base_raw->hdrlen,
806 				    base_raw->size + base_raw->hdrlen);
807 			} else {
808 				err = got_deltify(&deltas, &ndeltas,
809 				    raw->f, raw->hdrlen,
810 				    raw->size + raw->hdrlen,
811 				    base->dtab, base_raw->f, base_raw->hdrlen,
812 				    base_raw->size + base_raw->hdrlen);
813 			}
814 			got_object_raw_close(base_raw);
815 			base_raw = NULL;
816 			if (err)
817 				goto done;
818 
819 			size = delta_size(deltas, ndeltas);
820 			if (size + 32 < best_size){
821 				/*
822 				 * if we already picked a best delta,
823 				 * replace it.
824 				 */
825 				best_size = size;
826 				free(best_deltas);
827 				best_deltas = deltas;
828 				best_ndeltas = ndeltas;
829 				deltas = NULL;
830 				m->nchain = base->nchain + 1;
831 				m->prev = base;
832 				m->head = base->head;
833 				if (m->head == NULL)
834 					m->head = base;
835 			} else {
836 				free(deltas);
837 				deltas = NULL;
838 				ndeltas = 0;
839 			}
840 		}
841 
842 		if (best_ndeltas > 0) {
843 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
844 			    delta_memsize + best_size <= max_delta_memsize) {
845 				delta_memsize += best_size;
846 				err = encode_delta_in_mem(m, raw, best_deltas,
847 				    best_ndeltas, best_size, m->prev->size);
848 			} else {
849 				m->delta_offset = ftello(delta_cache);
850 				err = encode_delta(m, raw, best_deltas,
851 				    best_ndeltas, m->prev->size, delta_cache);
852 			}
853 			free(best_deltas);
854 			best_deltas = NULL;
855 			best_ndeltas = 0;
856 			if (err)
857 				goto done;
858 		}
859 
860 		got_object_raw_close(raw);
861 		raw = NULL;
862 	}
863 done:
864 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
865 		got_deltify_free(meta[i]->dtab);
866 		meta[i]->dtab = NULL;
867 	}
868 	if (raw)
869 		got_object_raw_close(raw);
870 	if (base_raw)
871 		got_object_raw_close(base_raw);
872 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
873 		err = got_error_from_errno("close");
874 	free(deltas);
875 	free(best_deltas);
876 	return err;
877 }
878 
879 static const struct got_error *
880 search_packidx(int *found, struct got_object_id *id,
881     struct got_repository *repo)
882 {
883 	const struct got_error *err = NULL;
884 	struct got_packidx *packidx = NULL;
885 	int idx;
886 
887 	*found = 0;
888 
889 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
890 	if (err == NULL)
891 		*found = 1; /* object is already packed */
892 	else if (err->code == GOT_ERR_NO_OBJ)
893 		err = NULL;
894 	return err;
895 }
896 
897 static const struct got_error *
898 add_object(int want_meta, struct got_object_idset *idset,
899     struct got_object_id *id, const char *path, int obj_type,
900     time_t mtime, int loose_obj_only, struct got_repository *repo,
901     int *ncolored, int *nfound, int *ntrees,
902     got_pack_progress_cb progress_cb, void *progress_arg,
903     struct got_ratelimit *rl)
904 {
905 	const struct got_error *err;
906 	struct got_pack_meta *m = NULL;
907 
908 	if (loose_obj_only) {
909 		int is_packed;
910 		err = search_packidx(&is_packed, id, repo);
911 		if (err)
912 			return err;
913 		if (is_packed && want_meta)
914 			return NULL;
915 	}
916 
917 	if (want_meta) {
918 		err = alloc_meta(&m, id, path, obj_type, mtime);
919 		if (err)
920 			return err;
921 
922 		(*nfound)++;
923 		err = report_progress(progress_cb, progress_arg, rl,
924 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
925 		if (err) {
926 			clear_meta(m);
927 			free(m);
928 			return err;
929 		}
930 	}
931 
932 	err = got_object_idset_add(idset, id, m);
933 	if (err) {
934 		clear_meta(m);
935 		free(m);
936 	}
937 	return err;
938 }
939 
940 static const struct got_error *
941 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
942     struct got_object_idset *idset, struct got_object_id *tree_id,
943     const char *dpath, time_t mtime, struct got_repository *repo,
944     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
945     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;
949 	struct got_tree_object *tree;
950 	char *p = NULL;
951 	int i;
952 
953 	err = got_object_open_as_tree(&tree, repo, tree_id);
954 	if (err)
955 		return err;
956 
957 	(*ntrees)++;
958 	err = report_progress(progress_cb, progress_arg, rl,
959 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
960 	if (err)
961 		return err;
962 
963 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
964 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
965 		struct got_object_id *id = got_tree_entry_get_id(e);
966 		mode_t mode = got_tree_entry_get_mode(e);
967 
968 		if (cancel_cb) {
969 			err = (*cancel_cb)(cancel_arg);
970 			if (err)
971 				break;
972 		}
973 
974 		if (got_object_tree_entry_is_submodule(e) ||
975 		    got_object_idset_contains(idset, id))
976 			continue;
977 		
978 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
979 		    got_tree_entry_get_name(e)) == -1) {
980 			err = got_error_from_errno("asprintf");
981 			break;
982 		}
983 
984 		if (S_ISDIR(mode)) {
985 			struct got_object_qid *qid;
986 			err = got_object_qid_alloc(&qid, id);
987 			if (err)
988 				break;
989 			STAILQ_INSERT_TAIL(ids, qid, entry);
990 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
991 			err = add_object(want_meta, idset, id, p,
992 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo,
993 			    ncolored, nfound, ntrees,
994 			    progress_cb, progress_arg, rl);
995 			if (err)
996 				break;
997 		}
998 		free(p);
999 		p = NULL;
1000 	}
1001 
1002 	got_object_tree_close(tree);
1003 	free(p);
1004 	return err;
1005 }
1006 
1007 static const struct got_error *
1008 load_tree(int want_meta, struct got_object_idset *idset,
1009     struct got_object_id *tree_id, const char *dpath, time_t mtime,
1010     struct got_repository *repo, int loose_obj_only,
1011     int *ncolored, int *nfound, int *ntrees,
1012     got_pack_progress_cb progress_cb, void *progress_arg,
1013     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1014 {
1015 	const struct got_error *err = NULL;
1016 	struct got_object_id_queue tree_ids;
1017 	struct got_object_qid *qid;
1018 
1019 	if (got_object_idset_contains(idset, tree_id))
1020 		return NULL;
1021 
1022 	err = got_object_qid_alloc(&qid, tree_id);
1023 	if (err)
1024 		return err;
1025 
1026 	STAILQ_INIT(&tree_ids);
1027 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
1028 
1029 	while (!STAILQ_EMPTY(&tree_ids)) {
1030 		if (cancel_cb) {
1031 			err = (*cancel_cb)(cancel_arg);
1032 			if (err)
1033 				break;
1034 		}
1035 
1036 		qid = STAILQ_FIRST(&tree_ids);
1037 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
1038 
1039 		if (got_object_idset_contains(idset, &qid->id)) {
1040 			got_object_qid_free(qid);
1041 			continue;
1042 		}
1043 
1044 		err = add_object(want_meta, idset, &qid->id, dpath,
1045 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo,
1046 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1047 		if (err) {
1048 			got_object_qid_free(qid);
1049 			break;
1050 		}
1051 
1052 		err = load_tree_entries(&tree_ids, want_meta, idset, &qid->id,
1053 		    dpath, mtime, repo, loose_obj_only, ncolored, nfound,
1054 		    ntrees, progress_cb, progress_arg, rl,
1055 		    cancel_cb, cancel_arg);
1056 		got_object_qid_free(qid);
1057 		if (err)
1058 			break;
1059 	}
1060 
1061 	got_object_id_queue_free(&tree_ids);
1062 	return err;
1063 }
1064 
1065 static const struct got_error *
1066 load_commit(int want_meta, struct got_object_idset *idset,
1067     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1068     int *ncolored, int *nfound, int *ntrees,
1069     got_pack_progress_cb progress_cb, void *progress_arg,
1070     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1071 {
1072 	const struct got_error *err;
1073 	struct got_commit_object *commit;
1074 
1075 	if (got_object_idset_contains(idset, id))
1076 		return NULL;
1077 
1078 	if (loose_obj_only) {
1079 		int is_packed;
1080 		err = search_packidx(&is_packed, id, repo);
1081 		if (err)
1082 			return err;
1083 		if (is_packed && want_meta)
1084 			return NULL;
1085 	}
1086 
1087 	err = got_object_open_as_commit(&commit, repo, id);
1088 	if (err)
1089 		return err;
1090 
1091 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1092 	    got_object_commit_get_committer_time(commit),
1093 	    loose_obj_only, repo,
1094 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1095 	if (err)
1096 		goto done;
1097 
1098 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1099 	    "", got_object_commit_get_committer_time(commit),
1100 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1101 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1102 done:
1103 	got_object_commit_close(commit);
1104 	return err;
1105 }
1106 
1107 static const struct got_error *
1108 load_tag(int want_meta, struct got_object_idset *idset,
1109     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1110     int *ncolored, int *nfound, int *ntrees,
1111     got_pack_progress_cb progress_cb, void *progress_arg,
1112     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1113 {
1114 	const struct got_error *err;
1115 	struct got_tag_object *tag = NULL;
1116 
1117 	if (got_object_idset_contains(idset, id))
1118 		return NULL;
1119 
1120 	if (loose_obj_only) {
1121 		int is_packed;
1122 		err = search_packidx(&is_packed, id, repo);
1123 		if (err)
1124 			return err;
1125 		if (is_packed && want_meta)
1126 			return NULL;
1127 	}
1128 
1129 	err = got_object_open_as_tag(&tag, repo, id);
1130 	if (err)
1131 		return err;
1132 
1133 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1134 	    got_object_tag_get_tagger_time(tag), loose_obj_only, repo,
1135 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1136 	if (err)
1137 		goto done;
1138 
1139 	switch (got_object_tag_get_object_type(tag)) {
1140 	case GOT_OBJ_TYPE_COMMIT:
1141 		err = load_commit(want_meta, idset,
1142 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1143 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1144 		    cancel_cb, cancel_arg);
1145 		break;
1146 	case GOT_OBJ_TYPE_TREE:
1147 		err = load_tree(want_meta, idset,
1148 		    got_object_tag_get_object_id(tag), "",
1149 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1150 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1151 		    cancel_cb, cancel_arg);
1152 		break;
1153 	default:
1154 		break;
1155 	}
1156 
1157 done:
1158 	got_object_tag_close(tag);
1159 	return err;
1160 }
1161 
1162 enum findtwixt_color {
1163 	COLOR_KEEP = 0,
1164 	COLOR_DROP,
1165 	COLOR_BLANK,
1166 	COLOR_SKIP,
1167 };
1168 
1169 static const int findtwixt_colors[] = {
1170 	COLOR_KEEP,
1171 	COLOR_DROP,
1172 	COLOR_BLANK,
1173 	COLOR_SKIP,
1174 };
1175 
1176 static const struct got_error *
1177 paint_commit(struct got_object_qid *qid, int color)
1178 {
1179 	if (color < 0 || color >= nitems(findtwixt_colors))
1180 		return got_error(GOT_ERR_RANGE);
1181 
1182 	qid->data = (void *)&findtwixt_colors[color];
1183 	return NULL;
1184 }
1185 
1186 static const struct got_error *
1187 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1188     int color, struct got_repository *repo)
1189 {
1190 	const struct got_error *err;
1191 	struct got_object_qid *qid;
1192 
1193 	err = got_object_qid_alloc(&qid, id);
1194 	if (err)
1195 		return err;
1196 
1197 	STAILQ_INSERT_TAIL(ids, qid, entry);
1198 	return paint_commit(qid, color);
1199 }
1200 
1201 struct append_id_arg {
1202 	struct got_object_id **array;
1203 	int idx;
1204 	struct got_object_idset *drop;
1205 	struct got_object_idset *skip;
1206 };
1207 
1208 static const struct got_error *
1209 append_id(struct got_object_id *id, void *data, void *arg)
1210 {
1211 	struct append_id_arg *a = arg;
1212 
1213 	if (got_object_idset_contains(a->skip, id) ||
1214 	    got_object_idset_contains(a->drop, id))
1215 		return NULL;
1216 
1217 	a->array[++a->idx] = got_object_id_dup(id);
1218 	if (a->array[a->idx] == NULL)
1219 		return got_error_from_errno("got_object_id_dup");
1220 
1221 	return NULL;
1222 }
1223 
1224 static const struct got_error *
1225 queue_commit_or_tag_id(struct got_object_id *id, int color,
1226     struct got_object_id_queue *ids, struct got_repository *repo)
1227 {
1228 	const struct got_error *err;
1229 	struct got_tag_object *tag = NULL;
1230 	int obj_type;
1231 
1232 	err = got_object_get_type(&obj_type, repo, id);
1233 	if (err)
1234 		return err;
1235 
1236 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1237 		err = got_object_open_as_tag(&tag, repo, id);
1238 		if (err)
1239 			return err;
1240 		obj_type = got_object_tag_get_object_type(tag);
1241 		id = got_object_tag_get_object_id(tag);
1242 	}
1243 
1244 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1245 		err = queue_commit_id(ids, id, color, repo);
1246 		if (err)
1247 			goto done;
1248 	}
1249 done:
1250 	if (tag)
1251 		got_object_tag_close(tag);
1252 	return err;
1253 }
1254 
1255 static const struct got_error *
1256 paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1257     struct got_object_idset *keep, struct got_object_idset *drop,
1258     struct got_object_idset *skip, struct got_repository *repo,
1259     got_pack_progress_cb progress_cb, void *progress_arg,
1260     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1261 {
1262 	const struct got_error *err = NULL;
1263 	struct got_commit_object *commit = NULL;
1264 	const struct got_object_id_queue *parents;
1265 	struct got_object_qid *qid;
1266 	int nqueued = nids, nskip = 0;
1267 
1268 	while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1269 		int color;
1270 
1271 		if (cancel_cb) {
1272 			err = cancel_cb(cancel_arg);
1273 			if (err)
1274 				break;
1275 		}
1276 
1277 		qid = STAILQ_FIRST(ids);
1278 		STAILQ_REMOVE_HEAD(ids, entry);
1279 		nqueued--;
1280 		color = *((int *)qid->data);
1281 		if (color == COLOR_SKIP)
1282 			nskip--;
1283 
1284 		if (got_object_idset_contains(skip, &qid->id)) {
1285 			got_object_qid_free(qid);
1286 			continue;
1287 		}
1288 
1289 		switch (color) {
1290 		case COLOR_KEEP:
1291 			if (got_object_idset_contains(keep, &qid->id)) {
1292 				got_object_qid_free(qid);
1293 				continue;
1294 			}
1295 			if (got_object_idset_contains(drop, &qid->id)) {
1296 				err = paint_commit(qid, COLOR_SKIP);
1297 				if (err)
1298 					goto done;
1299 				nskip++;
1300 			} else
1301 				(*ncolored)++;
1302 			err = got_object_idset_add(keep, &qid->id, NULL);
1303 			if (err)
1304 				goto done;
1305 			break;
1306 		case COLOR_DROP:
1307 			if (got_object_idset_contains(drop, &qid->id)) {
1308 				got_object_qid_free(qid);
1309 				continue;
1310 			}
1311 			if (got_object_idset_contains(keep, &qid->id)) {
1312 				err = paint_commit(qid, COLOR_SKIP);
1313 				if (err)
1314 					goto done;
1315 				nskip++;
1316 			} else
1317 				(*ncolored)++;
1318 			err = got_object_idset_add(drop, &qid->id, NULL);
1319 			if (err)
1320 				goto done;
1321 			break;
1322 		case COLOR_SKIP:
1323 			if (!got_object_idset_contains(skip, &qid->id)) {
1324 				err = got_object_idset_add(skip, &qid->id,
1325 				    NULL);
1326 				if (err)
1327 					goto done;
1328 			}
1329 			break;
1330 		default:
1331 			/* should not happen */
1332 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1333 			    "%s invalid commit color %d", __func__, color);
1334 			goto done;
1335 		}
1336 
1337 		err = report_progress(progress_cb, progress_arg, rl,
1338 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1339 		if (err)
1340 			break;
1341 
1342 
1343 		err = got_object_open_as_commit(&commit, repo, &qid->id);
1344 		if (err)
1345 			break;
1346 
1347 		parents = got_object_commit_get_parent_ids(commit);
1348 		if (parents) {
1349 			struct got_object_qid *pid;
1350 			color = *((int *)qid->data);
1351 			STAILQ_FOREACH(pid, parents, entry) {
1352 				err = queue_commit_id(ids, &pid->id, color,
1353 				    repo);
1354 				if (err)
1355 					break;
1356 				nqueued++;
1357 				if (color == COLOR_SKIP)
1358 					nskip++;
1359 			}
1360 		}
1361 
1362 		got_object_commit_close(commit);
1363 		commit = NULL;
1364 		got_object_qid_free(qid);
1365 	}
1366 done:
1367 	if (commit)
1368 		got_object_commit_close(commit);
1369 	return err;
1370 }
1371 
1372 static const struct got_error *
1373 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1374     struct got_object_id **head, int nhead,
1375     struct got_object_id **tail, int ntail,
1376     struct got_repository *repo,
1377     got_pack_progress_cb progress_cb, void *progress_arg,
1378     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1379 {
1380 	const struct got_error *err = NULL;
1381 	struct got_object_id_queue ids;
1382 	struct got_object_idset *keep, *drop, *skip = NULL;
1383 	int i, nkeep;
1384 
1385 	STAILQ_INIT(&ids);
1386 	*res = NULL;
1387 	*nres = 0;
1388 	*ncolored = 0;
1389 
1390 	keep = got_object_idset_alloc();
1391 	if (keep == NULL)
1392 		return got_error_from_errno("got_object_idset_alloc");
1393 
1394 	drop = got_object_idset_alloc();
1395 	if (drop == NULL) {
1396 		err = got_error_from_errno("got_object_idset_alloc");
1397 		goto done;
1398 	}
1399 
1400 	skip = got_object_idset_alloc();
1401 	if (skip == NULL) {
1402 		err = got_error_from_errno("got_object_idset_alloc");
1403 		goto done;
1404 	}
1405 
1406 	for (i = 0; i < nhead; i++) {
1407 		struct got_object_id *id = head[i];
1408 		if (id == NULL)
1409 			continue;
1410 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1411 		if (err)
1412 			goto done;
1413 	}		
1414 
1415 	for (i = 0; i < ntail; i++) {
1416 		struct got_object_id *id = tail[i];
1417 		if (id == NULL)
1418 			continue;
1419 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1420 		if (err)
1421 			goto done;
1422 	}
1423 
1424 	err = paint_commits(ncolored, &ids, nhead + ntail,
1425 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1426 	    cancel_cb, cancel_arg);
1427 	if (err)
1428 		goto done;
1429 
1430 	nkeep = got_object_idset_num_elements(keep);
1431 	if (nkeep > 0) {
1432 		struct append_id_arg arg;
1433 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1434 		if (arg.array == NULL) {
1435 			err = got_error_from_errno("calloc");
1436 			goto done;
1437 		}
1438 		arg.idx = -1;
1439 		arg.skip = skip;
1440 		arg.drop = drop;
1441 		err = got_object_idset_for_each(keep, append_id, &arg);
1442 		if (err) {
1443 			free(arg.array);
1444 			goto done;
1445 		}
1446 		*res = arg.array;
1447 		*nres = arg.idx + 1;
1448 	}
1449 done:
1450 	got_object_idset_free(keep);
1451 	got_object_idset_free(drop);
1452 	if (skip)
1453 		got_object_idset_free(skip);
1454 	got_object_id_queue_free(&ids);
1455 	return err;
1456 }
1457 
1458 static const struct got_error *
1459 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1460     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1461     struct got_object_id **ours, int nours, struct got_repository *repo,
1462     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1463     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1464 {
1465 	const struct got_error *err = NULL;
1466 	struct got_object_id **ids = NULL;
1467 	int i, nobj = 0, obj_type;
1468 
1469 	*ncolored = 0;
1470 	*nfound = 0;
1471 	*ntrees = 0;
1472 
1473 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1474 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1475 	if (err || nobj == 0)
1476 		goto done;
1477 
1478 	for (i = 0; i < ntheirs; i++) {
1479 		struct got_object_id *id = theirs[i];
1480 		if (id == NULL)
1481 			continue;
1482 		err = got_object_get_type(&obj_type, repo, id);
1483 		if (err)
1484 			return err;
1485 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1486 			err = load_commit(0, idset, id, repo,
1487 			    loose_obj_only, ncolored, nfound, ntrees,
1488 			    progress_cb, progress_arg, rl,
1489 			    cancel_cb, cancel_arg);
1490 			if (err)
1491 				goto done;
1492 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1493 			err = load_tag(0, idset, id, repo,
1494 			    loose_obj_only, ncolored, nfound, ntrees,
1495 			    progress_cb, progress_arg, rl,
1496 			    cancel_cb, cancel_arg);
1497 			if (err)
1498 				goto done;
1499 		}
1500 	}
1501 
1502 	for (i = 0; i < nobj; i++) {
1503 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1504 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1505 		    cancel_cb, cancel_arg);
1506 		if (err)
1507 			goto done;
1508 	}
1509 
1510 	for (i = 0; i < nours; i++) {
1511 		struct got_object_id *id = ours[i];
1512 		struct got_pack_meta *m;
1513 		if (id == NULL)
1514 			continue;
1515 		m = got_object_idset_get(idset, id);
1516 		if (m == NULL) {
1517 			err = got_object_get_type(&obj_type, repo, id);
1518 			if (err)
1519 				goto done;
1520 		} else
1521 			obj_type = m->obj_type;
1522 		if (obj_type != GOT_OBJ_TYPE_TAG)
1523 			continue;
1524 		err = load_tag(1, idset, id, repo, loose_obj_only,
1525 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1526 		    cancel_cb, cancel_arg);
1527 		if (err)
1528 			goto done;
1529 	}
1530 done:
1531 	for (i = 0; i < nobj; i++) {
1532 		free(ids[i]);
1533 	}
1534 	free(ids);
1535 	return err;
1536 }
1537 
1538 const struct got_error *
1539 hwrite(FILE *f, void *buf, off_t len, SHA1_CTX *ctx)
1540 {
1541 	size_t n;
1542 
1543 	SHA1Update(ctx, buf, len);
1544 	n = fwrite(buf, 1, len, f);
1545 	if (n != len)
1546 		return got_ferror(f, GOT_ERR_IO);
1547 	return NULL;
1548 }
1549 
1550 const struct got_error *
1551 hcopy(FILE *fsrc, FILE *fdst, off_t len, SHA1_CTX *ctx)
1552 {
1553 	unsigned char buf[65536];
1554 	off_t remain = len;
1555 	size_t n;
1556 
1557 	while (remain > 0) {
1558 		size_t copylen = MIN(sizeof(buf), remain);
1559 		n = fread(buf, 1, copylen, fsrc);
1560 		if (n != copylen)
1561 			return got_ferror(fsrc, GOT_ERR_IO);
1562 		SHA1Update(ctx, buf, copylen);
1563 		n = fwrite(buf, 1, copylen, fdst);
1564 		if (n != copylen)
1565 			return got_ferror(fdst, GOT_ERR_IO);
1566 		remain -= copylen;
1567 	}
1568 
1569 	return NULL;
1570 }
1571 
1572 static void
1573 putbe32(char *b, uint32_t n)
1574 {
1575 	b[0] = n >> 24;
1576 	b[1] = n >> 16;
1577 	b[2] = n >> 8;
1578 	b[3] = n >> 0;
1579 }
1580 
1581 static int
1582 write_order_cmp(const void *pa, const void *pb)
1583 {
1584 	struct got_pack_meta *a, *b, *ahd, *bhd;
1585 
1586 	a = *(struct got_pack_meta **)pa;
1587 	b = *(struct got_pack_meta **)pb;
1588 	ahd = (a->head == NULL) ? a : a->head;
1589 	bhd = (b->head == NULL) ? b : b->head;
1590 	if (bhd->mtime < ahd->mtime)
1591 		return -1;
1592 	if (bhd->mtime > ahd->mtime)
1593 		return 1;
1594 	if (bhd < ahd)
1595 		return -1;
1596 	if (bhd > ahd)
1597 		return 1;
1598 	if (a->nchain != b->nchain)
1599 		return a->nchain - b->nchain;
1600 	if (a->mtime < b->mtime)
1601 		return -1;
1602 	if (a->mtime > b->mtime)
1603 		return 1;
1604 	return got_object_id_cmp(&a->id, &b->id);
1605 }
1606 
1607 static int
1608 reuse_write_order_cmp(const void *pa, const void *pb)
1609 {
1610 	struct got_pack_meta *a, *b;
1611 
1612 	a = *(struct got_pack_meta **)pa;
1613 	b = *(struct got_pack_meta **)pb;
1614 
1615 	if (a->reused_delta_offset < b->reused_delta_offset)
1616 		return -1;
1617 	if (a->reused_delta_offset > b->reused_delta_offset)
1618 		return 1;
1619 	return 0;
1620 }
1621 
1622 static const struct got_error *
1623 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1624 {
1625 	size_t i;
1626 
1627 	*hdrlen = 0;
1628 
1629 	hdr[0] = obj_type << 4;
1630 	hdr[0] |= len & 0xf;
1631 	len >>= 4;
1632 	for (i = 1; len != 0; i++){
1633 		if (i >= bufsize)
1634 			return got_error(GOT_ERR_NO_SPACE);
1635 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1636 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1637 		len >>= GOT_DELTA_SIZE_SHIFT;
1638 	}
1639 
1640 	*hdrlen = i;
1641 	return NULL;
1642 }
1643 
1644 static int
1645 packoff(char *hdr, off_t off)
1646 {
1647 	int i, j;
1648 	char rbuf[8];
1649 
1650 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1651 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1652 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1653 		    GOT_DELTA_SIZE_MORE;
1654 	}
1655 
1656 	j = 0;
1657 	while (i > 0)
1658 		hdr[j++] = rbuf[--i];
1659 	return j;
1660 }
1661 
1662 static const struct got_error *
1663 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1664     struct got_pack_meta *m)
1665 {
1666 	const struct got_error *err;
1667 	char buf[32];
1668 	int nh;
1669 
1670 	if (m->prev->off != 0) {
1671 		err = packhdr(&nh, buf, sizeof(buf),
1672 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1673 		if (err)
1674 			return err;
1675 		nh += packoff(buf + nh, m->off - m->prev->off);
1676 		err = hwrite(packfile, buf, nh, ctx);
1677 		if (err)
1678 			return err;
1679 		*packfile_size += nh;
1680 	} else {
1681 		err = packhdr(&nh, buf, sizeof(buf),
1682 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1683 		if (err)
1684 			return err;
1685 		err = hwrite(packfile, buf, nh, ctx);
1686 		if (err)
1687 			return err;
1688 		*packfile_size += nh;
1689 		err = hwrite(packfile, m->prev->id.sha1,
1690 		    sizeof(m->prev->id.sha1), ctx);
1691 		if (err)
1692 			return err;
1693 		*packfile_size += sizeof(m->prev->id.sha1);
1694 	}
1695 
1696 	return NULL;
1697 }
1698 
1699 static const struct got_error *
1700 write_packed_object(off_t *packfile_size, FILE *packfile,
1701     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1702     SHA1_CTX *ctx, struct got_repository *repo)
1703 {
1704 	const struct got_error *err = NULL;
1705 	struct got_deflate_checksum csum;
1706 	char buf[32];
1707 	int nh;
1708 	struct got_raw_object *raw = NULL;
1709 	off_t outlen;
1710 
1711 	csum.output_sha1 = ctx;
1712 	csum.output_crc = NULL;
1713 
1714 	m->off = ftello(packfile);
1715 	if (m->delta_len == 0) {
1716 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1717 		if (err)
1718 			goto done;
1719 		err = packhdr(&nh, buf, sizeof(buf),
1720 		    m->obj_type, raw->size);
1721 		if (err)
1722 			goto done;
1723 		err = hwrite(packfile, buf, nh, ctx);
1724 		if (err)
1725 			goto done;
1726 		*packfile_size += nh;
1727 		if (raw->f == NULL) {
1728 			err = got_deflate_to_file_mmap(&outlen,
1729 			    raw->data + raw->hdrlen, 0, raw->size,
1730 			    packfile, &csum);
1731 			if (err)
1732 				goto done;
1733 		} else {
1734 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1735 			    == -1) {
1736 				err = got_error_from_errno("fseeko");
1737 				goto done;
1738 			}
1739 			err = got_deflate_to_file(&outlen, raw->f,
1740 			    raw->size, packfile, &csum);
1741 			if (err)
1742 				goto done;
1743 		}
1744 		*packfile_size += outlen;
1745 		got_object_raw_close(raw);
1746 		raw = NULL;
1747 	} else if (m->delta_buf) {
1748 		err = deltahdr(packfile_size, ctx, packfile, m);
1749 		if (err)
1750 			goto done;
1751 		err = hwrite(packfile, m->delta_buf,
1752 		    m->delta_compressed_len, ctx);
1753 		if (err)
1754 			goto done;
1755 		*packfile_size += m->delta_compressed_len;
1756 		free(m->delta_buf);
1757 		m->delta_buf = NULL;
1758 	} else {
1759 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1760 		    == -1) {
1761 			err = got_error_from_errno("fseeko");
1762 			goto done;
1763 		}
1764 		err = deltahdr(packfile_size, ctx, packfile, m);
1765 		if (err)
1766 			goto done;
1767 		err = hcopy(delta_cache, packfile,
1768 		    m->delta_compressed_len, ctx);
1769 		if (err)
1770 			goto done;
1771 		*packfile_size += m->delta_compressed_len;
1772 	}
1773 done:
1774 	if (raw)
1775 		got_object_raw_close(raw);
1776 	return err;
1777 }
1778 
1779 static const struct got_error *
1780 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1781     struct got_pack_meta **deltify, int ndeltify,
1782     struct got_pack_meta **reuse, int nreuse,
1783     int ncolored, int nfound, int ntrees, int nours,
1784     struct got_repository *repo,
1785     got_pack_progress_cb progress_cb, void *progress_arg,
1786     struct got_ratelimit *rl,
1787     got_cancel_cb cancel_cb, void *cancel_arg)
1788 {
1789 	const struct got_error *err = NULL;
1790 	int i;
1791 	SHA1_CTX ctx;
1792 	struct got_pack_meta *m;
1793 	char buf[32];
1794 	size_t n;
1795 	off_t packfile_size = 0;
1796 	int outfd = -1;
1797 
1798 	SHA1Init(&ctx);
1799 
1800 	err = hwrite(packfile, "PACK", 4, &ctx);
1801 	if (err)
1802 		return err;
1803 	putbe32(buf, GOT_PACKFILE_VERSION);
1804 	err = hwrite(packfile, buf, 4, &ctx);
1805 	if (err)
1806 		goto done;
1807 	putbe32(buf, ndeltify + nreuse);
1808 	err = hwrite(packfile, buf, 4, &ctx);
1809 	if (err)
1810 		goto done;
1811 
1812 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1813 	    write_order_cmp);
1814 	for (i = 0; i < ndeltify; i++) {
1815 		err = report_progress(progress_cb, progress_arg, rl,
1816 		    ncolored, nfound, ntrees, packfile_size, nours,
1817 		    ndeltify + nreuse, ndeltify + nreuse, i);
1818 		if (err)
1819 			goto done;
1820 		m = deltify[i];
1821 		err = write_packed_object(&packfile_size, packfile,
1822 		    delta_cache, m, &outfd, &ctx, repo);
1823 		if (err)
1824 			goto done;
1825 	}
1826 
1827 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1828 	    reuse_write_order_cmp);
1829 	for (i = 0; i < nreuse; i++) {
1830 		err = report_progress(progress_cb, progress_arg, rl,
1831 		    ncolored, nfound, ntrees, packfile_size, nours,
1832 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1833 		if (err)
1834 			goto done;
1835 		m = reuse[i];
1836 		err = write_packed_object(&packfile_size, packfile,
1837 		    delta_cache, m, &outfd, &ctx, repo);
1838 		if (err)
1839 			goto done;
1840 	}
1841 
1842 	SHA1Final(pack_sha1, &ctx);
1843 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1844 	if (n != SHA1_DIGEST_LENGTH)
1845 		err = got_ferror(packfile, GOT_ERR_IO);
1846 	packfile_size += SHA1_DIGEST_LENGTH;
1847 	packfile_size += sizeof(struct got_packfile_hdr);
1848 	if (progress_cb) {
1849 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1850 		    packfile_size, nours, ndeltify + nreuse,
1851 		    ndeltify + nreuse, ndeltify + nreuse);
1852 		if (err)
1853 			goto done;
1854 	}
1855 done:
1856 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1857 		err = got_error_from_errno("close");
1858 	return err;
1859 }
1860 
1861 static const struct got_error *
1862 remove_unused_object(struct got_object_id *id, void *data, void *arg)
1863 {
1864 	struct got_object_idset *idset = arg;
1865 
1866 	if (data == NULL)
1867 		got_object_idset_remove(NULL, idset, id);
1868 
1869 	return NULL;
1870 }
1871 
1872 static const struct got_error *
1873 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1874 {
1875 	struct got_pack_meta *m = data;
1876 	struct got_pack_metavec *v = arg;
1877 
1878 	if (m->have_reused_delta)
1879 		return NULL;
1880 
1881 	return add_meta(m, v);
1882 }
1883 
1884 const struct got_error *
1885 got_pack_create(uint8_t *packsha1, FILE *packfile,
1886     struct got_object_id **theirs, int ntheirs,
1887     struct got_object_id **ours, int nours,
1888     struct got_repository *repo, int loose_obj_only, int allow_empty,
1889     got_pack_progress_cb progress_cb, void *progress_arg,
1890     got_cancel_cb cancel_cb, void *cancel_arg)
1891 {
1892 	const struct got_error *err;
1893 	int delta_cache_fd = -1;
1894 	FILE *delta_cache = NULL;
1895 	struct got_object_idset *idset;
1896 	struct got_ratelimit rl;
1897 	struct got_pack_metavec deltify, reuse;
1898 	int ncolored = 0, nfound = 0, ntrees = 0;
1899 	size_t ndeltify;
1900 
1901 	memset(&deltify, 0, sizeof(deltify));
1902 	memset(&reuse, 0, sizeof(reuse));
1903 
1904 	got_ratelimit_init(&rl, 0, 500);
1905 
1906 	idset = got_object_idset_alloc();
1907 	if (idset == NULL)
1908 		return got_error_from_errno("got_object_idset_alloc");
1909 
1910 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1911 	    ntheirs, ours, nours, repo, loose_obj_only,
1912 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1913 	if (err)
1914 		return err;
1915 
1916 	err = got_object_idset_for_each(idset, remove_unused_object, idset);
1917 	if (err)
1918 		goto done;
1919 
1920 	if (progress_cb) {
1921 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1922 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1923 		if (err)
1924 			goto done;
1925 	}
1926 
1927 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1928 		err = got_error(GOT_ERR_CANNOT_PACK);
1929 		goto done;
1930 	}
1931 
1932 	delta_cache_fd = got_opentempfd();
1933 	if (delta_cache_fd == -1) {
1934 		err = got_error_from_errno("got_opentemp");
1935 		goto done;
1936 	}
1937 
1938 	reuse.metasz = 64;
1939 	reuse.meta = calloc(reuse.metasz,
1940 	    sizeof(struct got_pack_meta *));
1941 	if (reuse.meta == NULL) {
1942 		err = got_error_from_errno("calloc");
1943 		goto done;
1944 	}
1945 
1946 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1947 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1948 	    cancel_cb, cancel_arg);
1949 	if (err)
1950 		goto done;
1951 
1952 	delta_cache = fdopen(delta_cache_fd, "a+");
1953 	if (delta_cache == NULL) {
1954 		err = got_error_from_errno("fdopen");
1955 		goto done;
1956 	}
1957 	delta_cache_fd = -1;
1958 
1959 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1960 		err = got_error_from_errno("fseeko");
1961 		goto done;
1962 	}
1963 
1964 	ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
1965 	if (ndeltify > 0) {
1966 		deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
1967 		if (deltify.meta == NULL) {
1968 			err = got_error_from_errno("calloc");
1969 			goto done;
1970 		}
1971 		deltify.metasz = ndeltify;
1972 
1973 		err = got_object_idset_for_each(idset, add_meta_idset_cb,
1974 		    &deltify);
1975 		if (err)
1976 			goto done;
1977 		if (deltify.nmeta > 0) {
1978 			err = pick_deltas(deltify.meta, deltify.nmeta,
1979 			    ncolored, nfound, ntrees, nours, reuse.nmeta,
1980 			    delta_cache, repo, progress_cb, progress_arg, &rl,
1981 			    cancel_cb, cancel_arg);
1982 			if (err)
1983 				goto done;
1984 		}
1985 	}
1986 
1987 	if (fflush(delta_cache) == EOF) {
1988 		err = got_error_from_errno("fflush");
1989 		goto done;
1990 	}
1991 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1992 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1993 	    nours, repo, progress_cb, progress_arg, &rl,
1994 	    cancel_cb, cancel_arg);
1995 	if (err)
1996 		goto done;
1997 done:
1998 	free_nmeta(deltify.meta, deltify.nmeta);
1999 	free_nmeta(reuse.meta, reuse.nmeta);
2000 	got_object_idset_free(idset);
2001 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2002 		err = got_error_from_errno("close");
2003 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
2004 		err = got_error_from_errno("fclose");
2005 	return err;
2006 }