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 #include <sys/mman.h>
25 
26 #include <endian.h>
27 #include <errno.h>
28 #include <stdint.h>
29 #include <imsg.h>
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <sha1.h>
34 #include <time.h>
35 #include <unistd.h>
36 #include <limits.h>
37 #include <zlib.h>
38 
39 #include "got_error.h"
40 #include "got_cancel.h"
41 #include "got_object.h"
42 #include "got_path.h"
43 #include "got_reference.h"
44 #include "got_repository_admin.h"
45 #include "got_opentemp.h"
46 
47 #include "got_lib_deltify.h"
48 #include "got_lib_delta.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_idset.h"
51 #include "got_lib_object_cache.h"
52 #include "got_lib_deflate.h"
53 #include "got_lib_pack.h"
54 #include "got_lib_pack_create.h"
55 #include "got_lib_privsep.h"
56 #include "got_lib_repository.h"
57 #include "got_lib_ratelimit.h"
58 #include "got_lib_inflate.h"
59 
60 #include "murmurhash2.h"
61 
62 #ifndef MIN
63 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
64 #endif
65 
66 #ifndef MAX
67 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
68 #endif
69 
70 #ifndef nitems
71 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
72 #endif
73 
74 struct got_pack_meta {
75 	struct got_object_id id;
76 	uint32_t path_hash;
77 	int	obj_type;
78 	off_t	size;
79 	time_t	mtime;
80 
81 	/* The best delta we picked */
82 	struct got_pack_meta *head;
83 	struct got_pack_meta *prev;
84 	unsigned char *delta_buf; /* if encoded in memory (compressed) */
85 	off_t	delta_offset;	/* offset in delta cache file (compressed) */
86 	off_t	delta_len;	/* encoded delta length */
87 	off_t	delta_compressed_len; /* encoded+compressed delta length */
88 	int	nchain;
89 
90 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
91 	struct got_object_id *base_obj_id;
92 
93 	/* Only used for delta window */
94 	struct got_delta_table *dtab;
95 
96 	/* Only used for writing offset deltas */
97 	off_t	off;
98 };
99 
100 struct got_pack_metavec {
101 	struct got_pack_meta **meta;
102 	int nmeta;
103 	int metasz;
104 };
105 
106 static const struct got_error *
107 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
108     const char *path, int obj_type, time_t mtime, uint32_t seed)
109 {
110 	struct got_pack_meta *m;
111 
112 	*new = NULL;
113 
114 	m = calloc(1, sizeof(*m));
115 	if (m == NULL)
116 		return got_error_from_errno("calloc");
117 
118 	memcpy(&m->id, id, sizeof(m->id));
119 
120 	m->path_hash = murmurhash2(path, strlen(path), seed);
121 	m->obj_type = obj_type;
122 	m->mtime = mtime;
123 	*new = m;
124 	return NULL;
125 }
126 
127 static void
128 clear_meta(struct got_pack_meta *meta)
129 {
130 	if (meta == NULL)
131 		return;
132 	meta->path_hash = 0;
133 	free(meta->delta_buf);
134 	meta->delta_buf = NULL;
135 	free(meta->base_obj_id);
136 	meta->base_obj_id = NULL;
137 	meta->reused_delta_offset = 0;
138 }
139 
140 static void
141 free_nmeta(struct got_pack_meta **meta, int nmeta)
142 {
143 	int i;
144 
145 	for (i = 0; i < nmeta; i++)
146 		clear_meta(meta[i]);
147 	free(meta);
148 }
149 
150 static int
151 delta_order_cmp(const void *pa, const void *pb)
152 {
153 	struct got_pack_meta *a, *b;
154 
155 	a = *(struct got_pack_meta **)pa;
156 	b = *(struct got_pack_meta **)pb;
157 
158 	if (a->obj_type != b->obj_type)
159 		return a->obj_type - b->obj_type;
160 	if (a->path_hash < b->path_hash)
161 		return -1;
162 	if (a->path_hash > b->path_hash)
163 		return 1;
164 	if (a->mtime < b->mtime)
165 		return -1;
166 	if (a->mtime > b->mtime)
167 		return 1;
168 	return got_object_id_cmp(&a->id, &b->id);
169 }
170 
171 static off_t
172 delta_size(struct got_delta_instruction *deltas, int ndeltas)
173 {
174 	int i;
175 	off_t size = 32;
176 	for (i = 0; i < ndeltas; i++) {
177 		if (deltas[i].copy)
178 			size += GOT_DELTA_SIZE_SHIFT;
179 		else
180 			size += deltas[i].len + 1;
181 	}
182 	return size;
183 }
184 
185 static const struct got_error *
186 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
187 {
188 	char *n;
189 
190 	if (*len + nseg >= *sz) {
191 		while (*len + nseg >= *sz)
192 			*sz += *sz / 2;
193 		n = realloc(*p, *sz);
194 		if (n == NULL)
195 			return got_error_from_errno("realloc");
196 		*p = n;
197 	}
198 	memcpy(*p + *len, seg, nseg);
199 	*len += nseg;
200 	return NULL;
201 }
202 
203 static const struct got_error *
204 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
205     struct got_delta_instruction *deltas, int ndeltas,
206     off_t delta_size, off_t base_size)
207 {
208 	const struct got_error *err;
209 	unsigned char buf[16], *bp;
210 	int i, j;
211 	size_t len = 0, compressed_len;
212 	off_t bufsize = delta_size;
213 	off_t n;
214 	struct got_delta_instruction *d;
215 	uint8_t *delta_buf;
216 
217 	delta_buf = malloc(bufsize);
218 	if (delta_buf == NULL)
219 		return got_error_from_errno("malloc");
220 
221 	/* base object size */
222 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
223 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
224 	for (i = 1; n > 0; i++) {
225 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
226 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
227 		n >>= GOT_DELTA_SIZE_SHIFT;
228 	}
229 	err = append(&delta_buf, &len, &bufsize, buf, i);
230 	if (err)
231 		goto done;
232 
233 	/* target object size */
234 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
235 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
236 	for (i = 1; n > 0; i++) {
237 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
238 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
239 		n >>= GOT_DELTA_SIZE_SHIFT;
240 	}
241 	err = append(&delta_buf, &len, &bufsize, buf, i);
242 	if (err)
243 		goto done;
244 
245 	for (j = 0; j < ndeltas; j++) {
246 		d = &deltas[j];
247 		if (d->copy) {
248 			n = d->offset;
249 			bp = &buf[1];
250 			buf[0] = GOT_DELTA_BASE_COPY;
251 			for (i = 0; i < 4; i++) {
252 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
253 				buf[0] |= 1 << i;
254 				*bp++ = n & 0xff;
255 				n >>= 8;
256 				if (n == 0)
257 					break;
258 			}
259 
260 			n = d->len;
261 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
262 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
263 				for (i = 0; i < 3 && n > 0; i++) {
264 					buf[0] |= 1 << (i + 4);
265 					*bp++ = n & 0xff;
266 					n >>= 8;
267 				}
268 			}
269 			err = append(&delta_buf, &len, &bufsize,
270 			    buf, bp - buf);
271 			if (err)
272 				goto done;
273 		} else if (o->f == NULL) {
274 			n = 0;
275 			while (n != d->len) {
276 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
277 				err = append(&delta_buf, &len, &bufsize,
278 				    buf, 1);
279 				if (err)
280 					goto done;
281 				err = append(&delta_buf, &len, &bufsize,
282 				    o->data + o->hdrlen + d->offset + n,
283 				    buf[0]);
284 				if (err)
285 					goto done;
286 				n += buf[0];
287 			}
288 		} else {
289 			char content[128];
290 			size_t r;
291 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
292 				err = got_error_from_errno("fseeko");
293 				goto done;
294 			}
295 			n = 0;
296 			while (n != d->len) {
297 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
298 				err = append(&delta_buf, &len, &bufsize,
299 				    buf, 1);
300 				if (err)
301 					goto done;
302 				r = fread(content, 1, buf[0], o->f);
303 				if (r != buf[0]) {
304 					err = got_ferror(o->f, GOT_ERR_IO);
305 					goto done;
306 				}
307 				err = append(&delta_buf, &len, &bufsize,
308 				    content, buf[0]);
309 				if (err)
310 					goto done;
311 				n += buf[0];
312 			}
313 		}
314 	}
315 
316 	err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
317 	    NULL, NULL, delta_buf, 0, len);
318 	if (err)
319 		goto done;
320 
321 	m->delta_len = len;
322 	m->delta_compressed_len = compressed_len;
323 done:
324 	free(delta_buf);
325 	return err;
326 }
327 
328 static const struct got_error *
329 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
330     struct got_delta_instruction *deltas, int ndeltas,
331     off_t base_size, FILE *f)
332 {
333 	const struct got_error *err;
334 	unsigned char buf[16], *bp;
335 	int i, j;
336 	off_t n;
337 	struct got_deflate_buf zb;
338 	struct got_delta_instruction *d;
339 	off_t delta_len = 0, compressed_len = 0;
340 
341 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
342 	if (err)
343 		return err;
344 
345 	/* base object size */
346 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
347 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
348 	for (i = 1; n > 0; i++) {
349 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
350 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
351 		n >>= GOT_DELTA_SIZE_SHIFT;
352 	}
353 
354 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
355 	    buf, 0, i, f, NULL);
356 	if (err)
357 		goto done;
358 	delta_len += i;
359 
360 	/* target object size */
361 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
362 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
363 	for (i = 1; n > 0; i++) {
364 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
365 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
366 		n >>= GOT_DELTA_SIZE_SHIFT;
367 	}
368 
369 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
370 	    buf, 0, i, f, NULL);
371 	if (err)
372 		goto done;
373 	delta_len += i;
374 
375 	for (j = 0; j < ndeltas; j++) {
376 		d = &deltas[j];
377 		if (d->copy) {
378 			n = d->offset;
379 			bp = &buf[1];
380 			buf[0] = GOT_DELTA_BASE_COPY;
381 			for (i = 0; i < 4; i++) {
382 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
383 				buf[0] |= 1 << i;
384 				*bp++ = n & 0xff;
385 				n >>= 8;
386 				if (n == 0)
387 					break;
388 			}
389 			n = d->len;
390 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
391 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
392 				for (i = 0; i < 3 && n > 0; i++) {
393 					buf[0] |= 1 << (i + 4);
394 					*bp++ = n & 0xff;
395 					n >>= 8;
396 				}
397 			}
398 			err = got_deflate_append_to_file_mmap(&zb,
399 			    &compressed_len, buf, 0, bp - buf, f, NULL);
400 			if (err)
401 				goto done;
402 			delta_len += (bp - buf);
403 		} else if (o->f == NULL) {
404 			n = 0;
405 			while (n != d->len) {
406 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
407 				err = got_deflate_append_to_file_mmap(&zb,
408 				    &compressed_len, buf, 0, 1, f, NULL);
409 				if (err)
410 					goto done;
411 				delta_len++;
412 				err = got_deflate_append_to_file_mmap(&zb,
413 				    &compressed_len,
414 				    o->data + o->hdrlen + d->offset + n, 0,
415 				    buf[0], f, NULL);
416 				if (err)
417 					goto done;
418 				delta_len += buf[0];
419 				n += buf[0];
420 			}
421 		} else {
422 			char content[128];
423 			size_t r;
424 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
425 				err = got_error_from_errno("fseeko");
426 				goto done;
427 			}
428 			n = 0;
429 			while (n != d->len) {
430 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
431 				err = got_deflate_append_to_file_mmap(&zb,
432 				    &compressed_len, buf, 0, 1, f, NULL);
433 				if (err)
434 					goto done;
435 				delta_len++;
436 				r = fread(content, 1, buf[0], o->f);
437 				if (r != buf[0]) {
438 					err = got_ferror(o->f, GOT_ERR_IO);
439 					goto done;
440 				}
441 				err = got_deflate_append_to_file_mmap(&zb,
442 				    &compressed_len, content, 0, buf[0], f,
443 				    NULL);
444 				if (err)
445 					goto done;
446 				delta_len += buf[0];
447 				n += buf[0];
448 			}
449 		}
450 	}
451 
452 	err = got_deflate_flush(&zb, f, NULL, &compressed_len);
453 	if (err)
454 		goto done;
455 
456 	/* sanity check */
457 	if (compressed_len != ftello(f) - m->delta_offset) {
458 		err = got_error(GOT_ERR_COMPRESSION);
459 		goto done;
460 	}
461 
462 	m->delta_len = delta_len;
463 	m->delta_compressed_len = compressed_len;
464 done:
465 	got_deflate_end(&zb);
466 	return err;
467 }
468 
469 static const struct got_error *
470 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
471     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
472     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
473     int nobj_written)
474 {
475 	const struct got_error *err;
476 	int elapsed;
477 
478 	if (progress_cb == NULL)
479 		return NULL;
480 
481 	err = got_ratelimit_check(&elapsed, rl);
482 	if (err || !elapsed)
483 		return err;
484 
485 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
486 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
487 }
488 
489 static const struct got_error *
490 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
491 {
492 	if (v->nmeta == v->metasz){
493 		size_t newsize = 2 * v->metasz;
494 		struct got_pack_meta **new;
495 		new = reallocarray(v->meta, newsize, sizeof(*new));
496 		if (new == NULL)
497 			return got_error_from_errno("reallocarray");
498 		v->meta = new;
499 		v->metasz = newsize; 
500 	}
501 
502 	v->meta[v->nmeta++] = m;
503 	return NULL;
504 }
505 
506 static const struct got_error *
507 find_pack_for_reuse(struct got_packidx **best_packidx,
508     struct got_repository *repo)
509 {
510 	const struct got_error *err = NULL;
511 	struct got_pathlist_entry *pe;
512 	const char *best_packidx_path = NULL;
513 	int nobj_max = 0;
514 
515 	*best_packidx = NULL;
516 
517 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
518 		const char *path_packidx = pe->path;
519 		struct got_packidx *packidx;
520 		int nobj;
521 
522 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
523 		if (err)
524 			break;
525 
526 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
527 		if (nobj > nobj_max) {
528 			best_packidx_path = path_packidx;
529 			nobj_max = nobj;
530 		}
531 	}
532 
533 	if (best_packidx_path) {
534 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
535 		    repo);
536 	}
537 
538 	return err;
539 }
540 
541 struct send_id_arg {
542 	struct imsgbuf *ibuf;
543 	struct got_object_id *ids[GOT_IMSG_OBJ_ID_LIST_MAX_NIDS];
544 	size_t nids;
545 };
546 
547 static const struct got_error *
548 send_id(struct got_object_id *id, void *data, void *arg)
549 {
550 	const struct got_error *err = NULL;
551 	struct send_id_arg *a = arg;
552 
553 	a->ids[a->nids++] = id;
554 
555 	if (a->nids >= GOT_IMSG_OBJ_ID_LIST_MAX_NIDS) {
556 		err = got_privsep_send_object_idlist(a->ibuf, a->ids, a->nids);
557 		if (err)
558 			return err;
559 		a->nids = 0;
560 	}
561 
562 	return NULL;
563 }
564 
565 static const struct got_error *
566 recv_reused_delta(struct got_imsg_reused_delta *delta,
567     struct got_object_idset *idset, struct got_pack_metavec *v)
568 {
569 	struct got_pack_meta *m, *base;
570 
571 	if (delta->delta_offset + delta->delta_size < delta->delta_offset ||
572 	    delta->delta_offset +
573 	    delta->delta_compressed_size < delta->delta_offset)
574 		return got_error(GOT_ERR_BAD_PACKFILE);
575 
576 	m = got_object_idset_get(idset, &delta->id);
577 	if (m == NULL)
578 		return got_error(GOT_ERR_NO_OBJ);
579 
580 	base = got_object_idset_get(idset, &delta->base_id);
581 	if (base == NULL)
582 		return got_error(GOT_ERR_NO_OBJ);
583 
584 	m->delta_len = delta->delta_size;
585 	m->delta_compressed_len = delta->delta_compressed_size;
586 	m->delta_offset = delta->delta_out_offset;
587 	m->prev = base;
588 	m->size = delta->result_size;
589 	m->reused_delta_offset = delta->delta_offset;
590 	m->base_obj_id = got_object_id_dup(&delta->base_id);
591 	if (m->base_obj_id == NULL)
592 		return got_error_from_errno("got_object_id_dup");
593 
594 	return add_meta(m, v);
595 }
596 
597 static const struct got_error *
598 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
599     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
600     struct got_repository *repo,
601     got_pack_progress_cb progress_cb, void *progress_arg,
602     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
603 {
604 	const struct got_error *err = NULL;
605 	struct got_packidx *packidx;
606 	struct got_pack *pack;
607 	struct send_id_arg sia;
608 	struct got_imsg_reused_delta deltas[GOT_IMSG_REUSED_DELTAS_MAX_NDELTAS];
609 	size_t ndeltas, i;
610 
611 	err = find_pack_for_reuse(&packidx, repo);
612 	if (err)
613 		return err;
614 
615 	if (packidx == NULL)
616 		return NULL;
617 
618 	err = got_object_prepare_delta_reuse(&pack, packidx,
619 	    delta_cache_fd, repo);
620 	if (err)
621 		return err;
622 
623 	memset(&sia, 0, sizeof(sia));
624 	sia.ibuf = pack->privsep_child->ibuf;
625 	err = got_object_idset_for_each(idset, send_id, &sia);
626 	if (err)
627 		return err;
628 	if (sia.nids > 0) {
629 		err = got_privsep_send_object_idlist(pack->privsep_child->ibuf,
630 		    sia.ids, sia.nids);
631 		if (err)
632 			return err;
633 	}
634 	err = got_privsep_send_object_idlist_done(pack->privsep_child->ibuf);
635 	if (err)
636 		return err;
637 
638 	for (;;) {
639 		int done = 0;
640 
641 		if (cancel_cb) {
642 			err = (*cancel_cb)(cancel_arg);
643 			if (err)
644 				break;
645 		}
646 
647 		err = got_privsep_recv_reused_deltas(&done, deltas, &ndeltas,
648 		    pack->privsep_child->ibuf);
649 		if (err || done)
650 			break;
651 
652 		for (i = 0; i < ndeltas; i++) {
653 			struct got_imsg_reused_delta *delta = &deltas[i];
654 			err = recv_reused_delta(delta, idset, v);
655 			if (err)
656 				goto done;
657 		}
658 
659 		err = report_progress(progress_cb, progress_arg, rl,
660 		    ncolored, nfound, ntrees, 0L, ncommits,
661 		    got_object_idset_num_elements(idset), v->nmeta, 0);
662 		if (err)
663 			break;
664 	}
665 done:
666 	return err;
667 }
668 
669 static const struct got_error *
670 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
671     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
672     struct got_repository *repo,
673     got_pack_progress_cb progress_cb, void *progress_arg,
674     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
675 {
676 	const struct got_error *err = NULL;
677 	struct got_pack_meta *m = NULL, *base = NULL;
678 	struct got_raw_object *raw = NULL, *base_raw = NULL;
679 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
680 	int i, j, ndeltas, best_ndeltas;
681 	off_t size, best_size;
682 	const int max_base_candidates = 3;
683 	size_t delta_memsize = 0;
684 	const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
685 	int outfd = -1;
686 	uint32_t delta_seed;
687 
688 	delta_seed = arc4random();
689 
690 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
691 	for (i = 0; i < nmeta; i++) {
692 		if (cancel_cb) {
693 			err = (*cancel_cb)(cancel_arg);
694 			if (err)
695 				break;
696 		}
697 		err = report_progress(progress_cb, progress_arg, rl,
698 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
699 		    nreused + i, 0);
700 		if (err)
701 			goto done;
702 		m = meta[i];
703 
704 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
705 		    m->obj_type == GOT_OBJ_TYPE_TAG)
706 			continue;
707 
708 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
709 		if (err)
710 			goto done;
711 		m->size = raw->size;
712 
713 		if (raw->f == NULL) {
714 			err = got_deltify_init_mem(&m->dtab, raw->data,
715 			    raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
716 		} else {
717 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
718 			    raw->size + raw->hdrlen, delta_seed);
719 		}
720 		if (err)
721 			goto done;
722 
723 		if (i > max_base_candidates) {
724 			struct got_pack_meta *n = NULL;
725 			n = meta[i - (max_base_candidates + 1)];
726 			got_deltify_free(n->dtab);
727 			n->dtab = NULL;
728 		}
729 
730 		best_size = raw->size;
731 		best_ndeltas = 0;
732 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
733 			if (cancel_cb) {
734 				err = (*cancel_cb)(cancel_arg);
735 				if (err)
736 					goto done;
737 			}
738 			base = meta[j];
739 			/* long chains make unpacking slow, avoid such bases */
740 			if (base->nchain >= 128 ||
741 			    base->obj_type != m->obj_type)
742 				continue;
743 
744 			err = got_object_raw_open(&base_raw, &outfd, repo,
745 			    &base->id);
746 			if (err)
747 				goto done;
748 
749 			if (raw->f == NULL && base_raw->f == NULL) {
750 				err = got_deltify_mem_mem(&deltas, &ndeltas,
751 				    raw->data, raw->hdrlen,
752 				    raw->size + raw->hdrlen, delta_seed,
753 				    base->dtab, base_raw->data,
754 				    base_raw->hdrlen,
755 				    base_raw->size + base_raw->hdrlen);
756 			} else if (raw->f == NULL) {
757 				err = got_deltify_mem_file(&deltas, &ndeltas,
758 				    raw->data, raw->hdrlen,
759 				    raw->size + raw->hdrlen, delta_seed,
760 				    base->dtab, base_raw->f,
761 				    base_raw->hdrlen,
762 				    base_raw->size + base_raw->hdrlen);
763 			} else if (base_raw->f == NULL) {
764 				err = got_deltify_file_mem(&deltas, &ndeltas,
765 				    raw->f, raw->hdrlen,
766 				    raw->size + raw->hdrlen, delta_seed,
767 				    base->dtab, base_raw->data,
768 				    base_raw->hdrlen,
769 				    base_raw->size + base_raw->hdrlen);
770 			} else {
771 				err = got_deltify(&deltas, &ndeltas,
772 				    raw->f, raw->hdrlen,
773 				    raw->size + raw->hdrlen, delta_seed,
774 				    base->dtab, base_raw->f, base_raw->hdrlen,
775 				    base_raw->size + base_raw->hdrlen);
776 			}
777 			got_object_raw_close(base_raw);
778 			base_raw = NULL;
779 			if (err)
780 				goto done;
781 
782 			size = delta_size(deltas, ndeltas);
783 			if (size + 32 < best_size){
784 				/*
785 				 * if we already picked a best delta,
786 				 * replace it.
787 				 */
788 				best_size = size;
789 				free(best_deltas);
790 				best_deltas = deltas;
791 				best_ndeltas = ndeltas;
792 				deltas = NULL;
793 				m->nchain = base->nchain + 1;
794 				m->prev = base;
795 				m->head = base->head;
796 				if (m->head == NULL)
797 					m->head = base;
798 			} else {
799 				free(deltas);
800 				deltas = NULL;
801 				ndeltas = 0;
802 			}
803 		}
804 
805 		if (best_ndeltas > 0) {
806 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
807 			    delta_memsize + best_size <= max_delta_memsize) {
808 				delta_memsize += best_size;
809 				err = encode_delta_in_mem(m, raw, best_deltas,
810 				    best_ndeltas, best_size, m->prev->size);
811 			} else {
812 				m->delta_offset = ftello(delta_cache);
813 				err = encode_delta(m, raw, best_deltas,
814 				    best_ndeltas, m->prev->size, delta_cache);
815 			}
816 			free(best_deltas);
817 			best_deltas = NULL;
818 			best_ndeltas = 0;
819 			if (err)
820 				goto done;
821 		}
822 
823 		got_object_raw_close(raw);
824 		raw = NULL;
825 	}
826 done:
827 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
828 		got_deltify_free(meta[i]->dtab);
829 		meta[i]->dtab = NULL;
830 	}
831 	if (raw)
832 		got_object_raw_close(raw);
833 	if (base_raw)
834 		got_object_raw_close(base_raw);
835 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
836 		err = got_error_from_errno("close");
837 	free(deltas);
838 	free(best_deltas);
839 	return err;
840 }
841 
842 static const struct got_error *
843 search_packidx(int *found, struct got_object_id *id,
844     struct got_repository *repo)
845 {
846 	const struct got_error *err = NULL;
847 	struct got_packidx *packidx = NULL;
848 	int idx;
849 
850 	*found = 0;
851 
852 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
853 	if (err == NULL)
854 		*found = 1; /* object is already packed */
855 	else if (err->code == GOT_ERR_NO_OBJ)
856 		err = NULL;
857 	return err;
858 }
859 
860 static const struct got_error *
861 add_object(int want_meta, struct got_object_idset *idset,
862     struct got_object_id *id, const char *path, int obj_type,
863     time_t mtime, uint32_t seed, int loose_obj_only,
864     struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
865     got_pack_progress_cb progress_cb, void *progress_arg,
866     struct got_ratelimit *rl)
867 {
868 	const struct got_error *err;
869 	struct got_pack_meta *m = NULL;
870 
871 	if (loose_obj_only) {
872 		int is_packed;
873 		err = search_packidx(&is_packed, id, repo);
874 		if (err)
875 			return err;
876 		if (is_packed && want_meta)
877 			return NULL;
878 	}
879 
880 	if (want_meta) {
881 		err = alloc_meta(&m, id, path, obj_type, mtime, seed);
882 		if (err)
883 			return err;
884 
885 		(*nfound)++;
886 		err = report_progress(progress_cb, progress_arg, rl,
887 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
888 		if (err) {
889 			clear_meta(m);
890 			free(m);
891 			return err;
892 		}
893 	}
894 
895 	err = got_object_idset_add(idset, id, m);
896 	if (err) {
897 		clear_meta(m);
898 		free(m);
899 	}
900 	return err;
901 }
902 
903 static const struct got_error *
904 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
905     struct got_object_idset *idset, struct got_object_idset *idset_exclude,
906     struct got_tree_object *tree,
907     const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
908     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
909     got_pack_progress_cb progress_cb, void *progress_arg,
910     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
911 {
912 	const struct got_error *err;
913 	char *p = NULL;
914 	int i;
915 
916 	(*ntrees)++;
917 	err = report_progress(progress_cb, progress_arg, rl,
918 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
919 	if (err)
920 		return err;
921 
922 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
923 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
924 		struct got_object_id *id = got_tree_entry_get_id(e);
925 		mode_t mode = got_tree_entry_get_mode(e);
926 
927 		if (cancel_cb) {
928 			err = (*cancel_cb)(cancel_arg);
929 			if (err)
930 				break;
931 		}
932 
933 		if (got_object_tree_entry_is_submodule(e) ||
934 		    got_object_idset_contains(idset, id) ||
935 		    got_object_idset_contains(idset_exclude, id))
936 			continue;
937 
938 		/*
939 		 * If got-read-pack is crawling trees for us then
940 		 * we are only here to collect blob IDs.
941 		 */
942 		if (ids == NULL && S_ISDIR(mode))
943 			continue;
944 
945 		if (asprintf(&p, "%s%s%s", dpath,
946 		    got_path_is_root_dir(dpath) ? "" : "/",
947 		    got_tree_entry_get_name(e)) == -1) {
948 			err = got_error_from_errno("asprintf");
949 			break;
950 		}
951 
952 		if (S_ISDIR(mode)) {
953 			struct got_object_qid *qid;
954 			err = got_object_qid_alloc(&qid, id);
955 			if (err)
956 				break;
957 			qid->data = p;
958 			p = NULL;
959 			STAILQ_INSERT_TAIL(ids, qid, entry);
960 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
961 			err = add_object(want_meta,
962 			    want_meta ? idset : idset_exclude, id, p,
963 			    GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
964 			    repo, ncolored, nfound, ntrees,
965 			    progress_cb, progress_arg, rl);
966 			if (err)
967 				break;
968 			free(p);
969 			p = NULL;
970 		} else {
971 			free(p);
972 			p = NULL;
973 		}
974 	}
975 
976 	free(p);
977 	return err;
978 }
979 
980 static const struct got_error *
981 load_tree(int want_meta, struct got_object_idset *idset,
982     struct got_object_idset *idset_exclude,
983     struct got_object_id *tree_id, const char *dpath, time_t mtime,
984     uint32_t seed, struct got_repository *repo, int loose_obj_only,
985     int *ncolored, int *nfound, int *ntrees,
986     got_pack_progress_cb progress_cb, void *progress_arg,
987     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
988 {
989 	const struct got_error *err = NULL;
990 	struct got_object_id_queue tree_ids;
991 	struct got_object_qid *qid;
992 	struct got_tree_object *tree = NULL;
993 
994 	if (got_object_idset_contains(idset, tree_id) ||
995 	    got_object_idset_contains(idset_exclude, tree_id))
996 		return NULL;
997 
998 	err = got_object_qid_alloc(&qid, tree_id);
999 	if (err)
1000 		return err;
1001 	qid->data = strdup(dpath);
1002 	if (qid->data == NULL) {
1003 		err = got_error_from_errno("strdup");
1004 		got_object_qid_free(qid);
1005 		return err;
1006 	}
1007 
1008 	STAILQ_INIT(&tree_ids);
1009 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
1010 
1011 	while (!STAILQ_EMPTY(&tree_ids)) {
1012 		const char *path;
1013 		if (cancel_cb) {
1014 			err = (*cancel_cb)(cancel_arg);
1015 			if (err)
1016 				break;
1017 		}
1018 
1019 		qid = STAILQ_FIRST(&tree_ids);
1020 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
1021 		path = qid->data;
1022 
1023 		if (got_object_idset_contains(idset, &qid->id) ||
1024 		    got_object_idset_contains(idset_exclude, &qid->id)) {
1025 			free(qid->data);
1026 			got_object_qid_free(qid);
1027 			continue;
1028 		}
1029 
1030 		err = add_object(want_meta, want_meta ? idset : idset_exclude,
1031 		    &qid->id, path, GOT_OBJ_TYPE_TREE,
1032 		    mtime, seed, loose_obj_only, repo,
1033 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1034 		if (err) {
1035 			free(qid->data);
1036 			got_object_qid_free(qid);
1037 			break;
1038 		}
1039 
1040 		err = got_object_open_as_tree(&tree, repo, &qid->id);
1041 		if (err) {
1042 			free(qid->data);
1043 			got_object_qid_free(qid);
1044 			break;
1045 		}
1046 
1047 		err = load_tree_entries(&tree_ids, want_meta, idset,
1048 		    idset_exclude, tree, path, mtime, seed, repo,
1049 		    loose_obj_only, ncolored, nfound, ntrees,
1050 		    progress_cb, progress_arg, rl,
1051 		    cancel_cb, cancel_arg);
1052 		free(qid->data);
1053 		got_object_qid_free(qid);
1054 		if (err)
1055 			break;
1056 
1057 		got_object_tree_close(tree);
1058 		tree = NULL;
1059 	}
1060 
1061 	STAILQ_FOREACH(qid, &tree_ids, entry)
1062 		free(qid->data);
1063 	got_object_id_queue_free(&tree_ids);
1064 	if (tree)
1065 		got_object_tree_close(tree);
1066 	return err;
1067 }
1068 
1069 static const struct got_error *
1070 load_commit(int want_meta, struct got_object_idset *idset,
1071     struct got_object_idset *idset_exclude,
1072     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1073     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1074     got_pack_progress_cb progress_cb, void *progress_arg,
1075     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1076 {
1077 	const struct got_error *err;
1078 	struct got_commit_object *commit;
1079 
1080 	if (got_object_idset_contains(idset, id) ||
1081 	    got_object_idset_contains(idset_exclude, id))
1082 		return NULL;
1083 
1084 	if (loose_obj_only) {
1085 		int is_packed;
1086 		err = search_packidx(&is_packed, id, repo);
1087 		if (err)
1088 			return err;
1089 		if (is_packed && want_meta)
1090 			return NULL;
1091 	}
1092 
1093 	err = got_object_open_as_commit(&commit, repo, id);
1094 	if (err)
1095 		return err;
1096 
1097 	err = add_object(want_meta, want_meta ? idset : idset_exclude,
1098 	    id, "", GOT_OBJ_TYPE_COMMIT,
1099 	    got_object_commit_get_committer_time(commit), seed,
1100 	    loose_obj_only, repo,
1101 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1102 	if (err)
1103 		goto done;
1104 
1105 	err = load_tree(want_meta, idset, idset_exclude,
1106 	    got_object_commit_get_tree_id(commit),
1107 	    "", got_object_commit_get_committer_time(commit), seed,
1108 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1109 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1110 done:
1111 	got_object_commit_close(commit);
1112 	return err;
1113 }
1114 
1115 static const struct got_error *
1116 load_tag(int want_meta, struct got_object_idset *idset,
1117     struct got_object_idset *idset_exclude,
1118     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1119     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1120     got_pack_progress_cb progress_cb, void *progress_arg,
1121     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1122 {
1123 	const struct got_error *err;
1124 	struct got_tag_object *tag = NULL;
1125 
1126 	if (got_object_idset_contains(idset, id) ||
1127 	    got_object_idset_contains(idset_exclude, id))
1128 		return NULL;
1129 
1130 	if (loose_obj_only) {
1131 		int is_packed;
1132 		err = search_packidx(&is_packed, id, repo);
1133 		if (err)
1134 			return err;
1135 		if (is_packed && want_meta)
1136 			return NULL;
1137 	}
1138 
1139 	err = got_object_open_as_tag(&tag, repo, id);
1140 	if (err)
1141 		return err;
1142 
1143 	err = add_object(want_meta, want_meta ? idset : idset_exclude,
1144 	    id, "", GOT_OBJ_TYPE_TAG,
1145 	    got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1146 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1147 	if (err)
1148 		goto done;
1149 
1150 	switch (got_object_tag_get_object_type(tag)) {
1151 	case GOT_OBJ_TYPE_COMMIT:
1152 		err = load_commit(want_meta, idset, idset_exclude,
1153 		    got_object_tag_get_object_id(tag), repo, seed,
1154 		    loose_obj_only, ncolored, nfound, ntrees,
1155 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1156 		break;
1157 	case GOT_OBJ_TYPE_TREE:
1158 		err = load_tree(want_meta, idset, idset_exclude,
1159 		    got_object_tag_get_object_id(tag), "",
1160 		    got_object_tag_get_tagger_time(tag), seed, repo,
1161 		    loose_obj_only, ncolored, nfound, ntrees,
1162 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1163 		break;
1164 	default:
1165 		break;
1166 	}
1167 
1168 done:
1169 	got_object_tag_close(tag);
1170 	return err;
1171 }
1172 
1173 enum findtwixt_color {
1174 	COLOR_KEEP = 0,
1175 	COLOR_DROP,
1176 	COLOR_BLANK,
1177 	COLOR_SKIP,
1178 };
1179 
1180 static const int findtwixt_colors[] = {
1181 	COLOR_KEEP,
1182 	COLOR_DROP,
1183 	COLOR_BLANK,
1184 	COLOR_SKIP,
1185 };
1186 
1187 static const struct got_error *
1188 paint_commit(struct got_object_qid *qid, int color)
1189 {
1190 	if (color < 0 || color >= nitems(findtwixt_colors))
1191 		return got_error(GOT_ERR_RANGE);
1192 
1193 	qid->data = (void *)&findtwixt_colors[color];
1194 	return NULL;
1195 }
1196 
1197 static const struct got_error *
1198 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1199     int color, struct got_repository *repo)
1200 {
1201 	const struct got_error *err;
1202 	struct got_object_qid *qid;
1203 
1204 	err = got_object_qid_alloc(&qid, id);
1205 	if (err)
1206 		return err;
1207 
1208 	STAILQ_INSERT_TAIL(ids, qid, entry);
1209 	return paint_commit(qid, color);
1210 }
1211 
1212 struct append_id_arg {
1213 	struct got_object_id **array;
1214 	int idx;
1215 	struct got_object_idset *drop;
1216 	struct got_object_idset *skip;
1217 };
1218 
1219 static const struct got_error *
1220 append_id(struct got_object_id *id, void *data, void *arg)
1221 {
1222 	struct append_id_arg *a = arg;
1223 
1224 	if (got_object_idset_contains(a->skip, id) ||
1225 	    got_object_idset_contains(a->drop, id))
1226 		return NULL;
1227 
1228 	a->array[++a->idx] = got_object_id_dup(id);
1229 	if (a->array[a->idx] == NULL)
1230 		return got_error_from_errno("got_object_id_dup");
1231 
1232 	return NULL;
1233 }
1234 
1235 static const struct got_error *
1236 queue_commit_or_tag_id(struct got_object_id *id, int color,
1237     struct got_object_id_queue *ids, struct got_repository *repo)
1238 {
1239 	const struct got_error *err;
1240 	struct got_tag_object *tag = NULL;
1241 	int obj_type;
1242 
1243 	err = got_object_get_type(&obj_type, repo, id);
1244 	if (err)
1245 		return err;
1246 
1247 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1248 		err = got_object_open_as_tag(&tag, repo, id);
1249 		if (err)
1250 			return err;
1251 		obj_type = got_object_tag_get_object_type(tag);
1252 		id = got_object_tag_get_object_id(tag);
1253 	}
1254 
1255 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1256 		err = queue_commit_id(ids, id, color, repo);
1257 		if (err)
1258 			goto done;
1259 	}
1260 done:
1261 	if (tag)
1262 		got_object_tag_close(tag);
1263 	return err;
1264 }
1265 
1266 static const struct got_error *
1267 paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1268     struct got_object_idset *keep, struct got_object_idset *drop,
1269     struct got_object_idset *skip, struct got_repository *repo,
1270     got_pack_progress_cb progress_cb, void *progress_arg,
1271     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1272 {
1273 	const struct got_error *err = NULL;
1274 	struct got_commit_object *commit = NULL;
1275 	const struct got_object_id_queue *parents;
1276 	struct got_object_qid *qid;
1277 	int nqueued = nids, nskip = 0;
1278 
1279 	while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1280 		int color;
1281 
1282 		if (cancel_cb) {
1283 			err = cancel_cb(cancel_arg);
1284 			if (err)
1285 				break;
1286 		}
1287 
1288 		qid = STAILQ_FIRST(ids);
1289 		STAILQ_REMOVE_HEAD(ids, entry);
1290 		nqueued--;
1291 		color = *((int *)qid->data);
1292 		if (color == COLOR_SKIP)
1293 			nskip--;
1294 
1295 		if (got_object_idset_contains(skip, &qid->id)) {
1296 			got_object_qid_free(qid);
1297 			continue;
1298 		}
1299 
1300 		switch (color) {
1301 		case COLOR_KEEP:
1302 			if (got_object_idset_contains(keep, &qid->id)) {
1303 				got_object_qid_free(qid);
1304 				continue;
1305 			}
1306 			if (got_object_idset_contains(drop, &qid->id)) {
1307 				err = paint_commit(qid, COLOR_SKIP);
1308 				if (err)
1309 					goto done;
1310 			} else
1311 				(*ncolored)++;
1312 			err = got_object_idset_add(keep, &qid->id, NULL);
1313 			if (err)
1314 				goto done;
1315 			break;
1316 		case COLOR_DROP:
1317 			if (got_object_idset_contains(drop, &qid->id)) {
1318 				got_object_qid_free(qid);
1319 				continue;
1320 			}
1321 			if (got_object_idset_contains(keep, &qid->id)) {
1322 				err = paint_commit(qid, COLOR_SKIP);
1323 				if (err)
1324 					goto done;
1325 			} else
1326 				(*ncolored)++;
1327 			err = got_object_idset_add(drop, &qid->id, NULL);
1328 			if (err)
1329 				goto done;
1330 			break;
1331 		case COLOR_SKIP:
1332 			if (!got_object_idset_contains(skip, &qid->id)) {
1333 				err = got_object_idset_add(skip, &qid->id,
1334 				    NULL);
1335 				if (err)
1336 					goto done;
1337 			}
1338 			break;
1339 		default:
1340 			/* should not happen */
1341 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1342 			    "%s invalid commit color %d", __func__, color);
1343 			goto done;
1344 		}
1345 
1346 		err = report_progress(progress_cb, progress_arg, rl,
1347 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1348 		if (err)
1349 			break;
1350 
1351 
1352 		err = got_object_open_as_commit(&commit, repo, &qid->id);
1353 		if (err)
1354 			break;
1355 
1356 		parents = got_object_commit_get_parent_ids(commit);
1357 		if (parents) {
1358 			struct got_object_qid *pid;
1359 			color = *((int *)qid->data);
1360 			STAILQ_FOREACH(pid, parents, entry) {
1361 				err = queue_commit_id(ids, &pid->id, color,
1362 				    repo);
1363 				if (err)
1364 					break;
1365 				nqueued++;
1366 				if (color == COLOR_SKIP)
1367 					nskip++;
1368 			}
1369 		}
1370 
1371 		got_object_commit_close(commit);
1372 		commit = NULL;
1373 		got_object_qid_free(qid);
1374 	}
1375 done:
1376 	if (commit)
1377 		got_object_commit_close(commit);
1378 	return err;
1379 }
1380 
1381 static const struct got_error *
1382 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1383     struct got_object_id **head, int nhead,
1384     struct got_object_id **tail, int ntail,
1385     struct got_repository *repo,
1386     got_pack_progress_cb progress_cb, void *progress_arg,
1387     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1388 {
1389 	const struct got_error *err = NULL;
1390 	struct got_object_id_queue ids;
1391 	struct got_object_idset *keep, *drop, *skip = NULL;
1392 	int i, nkeep;
1393 
1394 	STAILQ_INIT(&ids);
1395 	*res = NULL;
1396 	*nres = 0;
1397 	*ncolored = 0;
1398 
1399 	keep = got_object_idset_alloc();
1400 	if (keep == NULL)
1401 		return got_error_from_errno("got_object_idset_alloc");
1402 
1403 	drop = got_object_idset_alloc();
1404 	if (drop == NULL) {
1405 		err = got_error_from_errno("got_object_idset_alloc");
1406 		goto done;
1407 	}
1408 
1409 	skip = got_object_idset_alloc();
1410 	if (skip == NULL) {
1411 		err = got_error_from_errno("got_object_idset_alloc");
1412 		goto done;
1413 	}
1414 
1415 	for (i = 0; i < nhead; i++) {
1416 		struct got_object_id *id = head[i];
1417 		if (id == NULL)
1418 			continue;
1419 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1420 		if (err)
1421 			goto done;
1422 	}		
1423 
1424 	for (i = 0; i < ntail; i++) {
1425 		struct got_object_id *id = tail[i];
1426 		if (id == NULL)
1427 			continue;
1428 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1429 		if (err)
1430 			goto done;
1431 	}
1432 
1433 	err = paint_commits(ncolored, &ids, nhead + ntail,
1434 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1435 	    cancel_cb, cancel_arg);
1436 	if (err)
1437 		goto done;
1438 
1439 	nkeep = got_object_idset_num_elements(keep);
1440 	if (nkeep > 0) {
1441 		struct append_id_arg arg;
1442 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1443 		if (arg.array == NULL) {
1444 			err = got_error_from_errno("calloc");
1445 			goto done;
1446 		}
1447 		arg.idx = -1;
1448 		arg.skip = skip;
1449 		arg.drop = drop;
1450 		err = got_object_idset_for_each(keep, append_id, &arg);
1451 		if (err) {
1452 			free(arg.array);
1453 			goto done;
1454 		}
1455 		*res = arg.array;
1456 		*nres = arg.idx + 1;
1457 	}
1458 done:
1459 	got_object_idset_free(keep);
1460 	got_object_idset_free(drop);
1461 	if (skip)
1462 		got_object_idset_free(skip);
1463 	got_object_id_queue_free(&ids);
1464 	return err;
1465 }
1466 
1467 struct load_packed_obj_arg {
1468 	/* output parameters: */
1469 	struct got_object_id *id;
1470 	char *dpath;
1471 	time_t mtime;
1472 
1473 	/* input parameters: */
1474 	uint32_t seed;
1475 	int want_meta;
1476 	struct got_object_idset *idset;
1477 	struct got_object_idset *idset_exclude;
1478 	int loose_obj_only;
1479 	int *ncolored;
1480 	int *nfound;
1481 	int *ntrees;
1482 	got_pack_progress_cb progress_cb;
1483 	void *progress_arg;
1484 	struct got_ratelimit *rl;
1485 	got_cancel_cb cancel_cb;
1486 	void *cancel_arg;
1487 };
1488 
1489 static const struct got_error *
1490 load_packed_commit_id(void *arg, time_t mtime, struct got_object_id *id,
1491     struct got_repository *repo)
1492 {
1493 	struct load_packed_obj_arg *a = arg;
1494 
1495 	if (got_object_idset_contains(a->idset, id) ||
1496 	    got_object_idset_contains(a->idset_exclude, id))
1497 		return NULL;
1498 
1499 	return add_object(a->want_meta,
1500 	    a->want_meta ? a->idset : a->idset_exclude,
1501 	    id, "", GOT_OBJ_TYPE_COMMIT, mtime, a->seed, a->loose_obj_only,
1502 	    repo, a->ncolored, a->nfound, a->ntrees,
1503 	    a->progress_cb, a->progress_arg, a->rl);
1504 }
1505 
1506 static const struct got_error *
1507 load_packed_tree_ids(void *arg, struct got_tree_object *tree, time_t mtime,
1508     struct got_object_id *id, const char *dpath, struct got_repository *repo)
1509 {
1510 	const struct got_error *err;
1511 	struct load_packed_obj_arg *a = arg;
1512 	const char *relpath;
1513 
1514 	/*
1515 	 * When we receive a tree's ID and path but not the tree itself,
1516 	 * this tree object was not found in the pack file. This is the
1517 	 * last time we are being called for this optimized traversal.
1518 	 * Return from here and switch to loading objects the slow way.
1519 	 */
1520 	if (tree == NULL) {
1521 		free(a->id);
1522 		a->id = got_object_id_dup(id);
1523 		if (a->id == NULL) {
1524 			err = got_error_from_errno("got_object_id_dup");
1525 			free(a->dpath);
1526 			a->dpath = NULL;
1527 			return err;
1528 		}
1529 
1530 		free(a->dpath);
1531 		a->dpath = strdup(dpath);
1532 		if (a->dpath == NULL) {
1533 			err = got_error_from_errno("strdup");
1534 			free(a->id);
1535 			a->id = NULL;
1536 			return err;
1537 		}
1538 
1539 		a->mtime = mtime;
1540 		return NULL;
1541 	}
1542 
1543 	if (got_object_idset_contains(a->idset, id) ||
1544 	    got_object_idset_contains(a->idset_exclude, id))
1545 		return NULL;
1546 
1547 	relpath = dpath;
1548 	while (relpath[0] == '/')
1549 		relpath++;
1550 
1551 	err = add_object(a->want_meta,
1552 	    a->want_meta ? a->idset : a->idset_exclude,
1553 	    id, relpath, GOT_OBJ_TYPE_TREE, mtime, a->seed,
1554 	    a->loose_obj_only, repo, a->ncolored, a->nfound, a->ntrees,
1555 	    a->progress_cb, a->progress_arg, a->rl);
1556 	if (err)
1557 		return err;
1558 
1559 	return load_tree_entries(NULL, a->want_meta, a->idset,
1560 	    a->idset_exclude, tree, dpath, mtime, a->seed, repo,
1561 	    a->loose_obj_only, a->ncolored, a->nfound, a->ntrees,
1562 	    a->progress_cb, a->progress_arg, a->rl,
1563 	    a->cancel_cb, a->cancel_arg);
1564 }
1565 
1566 static const struct got_error *
1567 load_packed_object_ids(int *found_all_objects,
1568     struct got_object_id **ours, int nours,
1569     struct got_object_id **theirs, int ntheirs,
1570     int want_meta, uint32_t seed, struct got_object_idset *idset,
1571     struct got_object_idset *idset_exclude, int loose_obj_only,
1572     struct got_repository *repo, struct got_packidx *packidx,
1573     int *ncolored, int *nfound, int *ntrees,
1574     got_pack_progress_cb progress_cb, void *progress_arg,
1575     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1576 {
1577 	const struct got_error *err = NULL;
1578 	struct load_packed_obj_arg lpa;
1579 
1580 	memset(&lpa, 0, sizeof(lpa));
1581 	lpa.seed = seed;
1582 	lpa.want_meta = want_meta;
1583 	lpa.idset = idset;
1584 	lpa.idset_exclude = idset_exclude;
1585 	lpa.loose_obj_only = loose_obj_only;
1586 	lpa.ncolored = ncolored;
1587 	lpa.nfound = nfound;
1588 	lpa.ntrees = ntrees;
1589 	lpa.progress_cb = progress_cb;
1590 	lpa.progress_arg = progress_arg;
1591 	lpa.rl = rl;
1592 	lpa.cancel_cb = cancel_cb;
1593 	lpa.cancel_arg = cancel_arg;
1594 
1595 	/* Attempt to load objects via got-read-pack, as far as possible. */
1596 	err = got_object_enumerate(found_all_objects, load_packed_commit_id,
1597 	   load_packed_tree_ids, &lpa, ours, nours, theirs, ntheirs,
1598 	   packidx, repo);
1599 	if (err)
1600 		return err;
1601 
1602 	if (lpa.id == NULL)
1603 		return NULL;
1604 
1605 	/*
1606 	 * An incomplete tree hierarchy was present in the pack file
1607 	 * and caused loading to be aborted.
1608 	 * Continue loading trees the slow way.
1609 	 */
1610 	err = load_tree(want_meta, idset, idset_exclude,
1611 	    lpa.id, lpa.dpath, lpa.mtime, seed, repo, loose_obj_only,
1612 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1613 	    cancel_cb, cancel_arg);
1614 	free(lpa.id);
1615 	free(lpa.dpath);
1616 	return err;
1617 }
1618 
1619 static const struct got_error *
1620 find_pack_for_enumeration(struct got_packidx **best_packidx,
1621     struct got_object_id **ids, int nids, struct got_repository *repo)
1622 {
1623 	const struct got_error *err = NULL;
1624 	struct got_pathlist_entry *pe;
1625 	const char *best_packidx_path = NULL;
1626 	int nobj_max = 0;
1627 	int ncommits_max = 0;
1628 
1629 	*best_packidx = NULL;
1630 
1631 	/*
1632 	 * Find the largest pack which contains at least some of the
1633 	 * commits and tags we are interested in.
1634 	 */
1635 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1636 		const char *path_packidx = pe->path;
1637 		struct got_packidx *packidx;
1638 		int nobj, i, idx, ncommits = 0;
1639 
1640 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1641 		if (err)
1642 			break;
1643 
1644 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1645 		if (nobj <= nobj_max)
1646 			continue;
1647 
1648 		for (i = 0; i < nids; i++) {
1649 			idx = got_packidx_get_object_idx(packidx, ids[i]);
1650 			if (idx != -1)
1651 				ncommits++;
1652 		}
1653 		if (ncommits > ncommits_max) {
1654 			best_packidx_path = path_packidx;
1655 			nobj_max = nobj;
1656 			ncommits_max = ncommits;
1657 		}
1658 	}
1659 
1660 	if (best_packidx_path && err == NULL) {
1661 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1662 		    repo);
1663 	}
1664 
1665 	return err;
1666 }
1667 
1668 static const struct got_error *
1669 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1670     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1671     struct got_object_id **ours, int nours, struct got_repository *repo,
1672     uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1673     void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1674     void *cancel_arg)
1675 {
1676 	const struct got_error *err = NULL;
1677 	struct got_object_id **ids = NULL;
1678 	struct got_packidx *packidx = NULL;
1679 	int i, nobj = 0, obj_type, found_all_objects = 0;
1680 	struct got_object_idset *idset_exclude;
1681 
1682 	idset_exclude = got_object_idset_alloc();
1683 	if (idset_exclude == NULL)
1684 		return got_error_from_errno("got_object_idset_alloc");
1685 
1686 	*ncolored = 0;
1687 	*nfound = 0;
1688 	*ntrees = 0;
1689 
1690 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1691 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1692 	if (err)
1693 		goto done;
1694 
1695 	err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
1696 	if (err)
1697 		goto done;
1698 	if (packidx) {
1699 		err = load_packed_object_ids(&found_all_objects,
1700 		    theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
1701 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1702 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1703 		if (err)
1704 			goto done;
1705 	}
1706 
1707 	for (i = 0; i < ntheirs; i++) {
1708 		struct got_object_id *id = theirs[i];
1709 		if (id == NULL)
1710 			continue;
1711 		err = got_object_get_type(&obj_type, repo, id);
1712 		if (err)
1713 			return err;
1714 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1715 			if (!found_all_objects) {
1716 				err = load_commit(0, idset, idset_exclude,
1717 				    id, repo, seed, loose_obj_only,
1718 				    ncolored, nfound, ntrees,
1719 				    progress_cb, progress_arg, rl,
1720 				    cancel_cb, cancel_arg);
1721 				if (err)
1722 					goto done;
1723 			}
1724 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1725 			err = load_tag(0, idset, idset_exclude, id, repo,
1726 			    seed, loose_obj_only, ncolored, nfound, ntrees,
1727 			    progress_cb, progress_arg, rl,
1728 			    cancel_cb, cancel_arg);
1729 			if (err)
1730 				goto done;
1731 		}
1732 	}
1733 
1734 	found_all_objects = 0;
1735 	err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
1736 	if (err)
1737 		goto done;
1738 	if (packidx) {
1739 		err = load_packed_object_ids(&found_all_objects, ids,
1740 		    nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
1741 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1742 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1743 		if (err)
1744 			goto done;
1745 	}
1746 
1747 	if (!found_all_objects) {
1748 		for (i = 0; i < nobj; i++) {
1749 			err = load_commit(1, idset, idset_exclude, ids[i],
1750 			    repo, seed, loose_obj_only, ncolored, nfound,
1751 			    ntrees, progress_cb, progress_arg, rl,
1752 			    cancel_cb, cancel_arg);
1753 			if (err)
1754 				goto done;
1755 		}
1756 	}
1757 
1758 	for (i = 0; i < nours; i++) {
1759 		struct got_object_id *id = ours[i];
1760 		struct got_pack_meta *m;
1761 		if (id == NULL)
1762 			continue;
1763 		m = got_object_idset_get(idset, id);
1764 		if (m == NULL) {
1765 			err = got_object_get_type(&obj_type, repo, id);
1766 			if (err)
1767 				goto done;
1768 		} else
1769 			obj_type = m->obj_type;
1770 		if (obj_type != GOT_OBJ_TYPE_TAG)
1771 			continue;
1772 		err = load_tag(1, idset, idset_exclude, id, repo,
1773 		    seed, loose_obj_only, ncolored, nfound, ntrees,
1774 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1775 		if (err)
1776 			goto done;
1777 	}
1778 done:
1779 	for (i = 0; i < nobj; i++) {
1780 		free(ids[i]);
1781 	}
1782 	free(ids);
1783 	got_object_idset_free(idset_exclude);
1784 	return err;
1785 }
1786 
1787 const struct got_error *
1788 hwrite(FILE *f, void *buf, off_t len, SHA1_CTX *ctx)
1789 {
1790 	size_t n;
1791 
1792 	SHA1Update(ctx, buf, len);
1793 	n = fwrite(buf, 1, len, f);
1794 	if (n != len)
1795 		return got_ferror(f, GOT_ERR_IO);
1796 	return NULL;
1797 }
1798 
1799 const struct got_error *
1800 hcopy(FILE *fsrc, FILE *fdst, off_t len, SHA1_CTX *ctx)
1801 {
1802 	unsigned char buf[65536];
1803 	off_t remain = len;
1804 	size_t n;
1805 
1806 	while (remain > 0) {
1807 		size_t copylen = MIN(sizeof(buf), remain);
1808 		n = fread(buf, 1, copylen, fsrc);
1809 		if (n != copylen)
1810 			return got_ferror(fsrc, GOT_ERR_IO);
1811 		SHA1Update(ctx, buf, copylen);
1812 		n = fwrite(buf, 1, copylen, fdst);
1813 		if (n != copylen)
1814 			return got_ferror(fdst, GOT_ERR_IO);
1815 		remain -= copylen;
1816 	}
1817 
1818 	return NULL;
1819 }
1820 
1821 const struct got_error *
1822 hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
1823     FILE *fdst, off_t len, SHA1_CTX *ctx)
1824 {
1825 	size_t n;
1826 
1827 	if (src_offset + len > src_size)
1828 		return got_error(GOT_ERR_RANGE);
1829 
1830 	SHA1Update(ctx, src + src_offset, len);
1831 	n = fwrite(src + src_offset, 1, len, fdst);
1832 	if (n != len)
1833 		return got_ferror(fdst, GOT_ERR_IO);
1834 
1835 	return NULL;
1836 }
1837 
1838 static void
1839 putbe32(char *b, uint32_t n)
1840 {
1841 	b[0] = n >> 24;
1842 	b[1] = n >> 16;
1843 	b[2] = n >> 8;
1844 	b[3] = n >> 0;
1845 }
1846 
1847 static int
1848 write_order_cmp(const void *pa, const void *pb)
1849 {
1850 	struct got_pack_meta *a, *b, *ahd, *bhd;
1851 
1852 	a = *(struct got_pack_meta **)pa;
1853 	b = *(struct got_pack_meta **)pb;
1854 	ahd = (a->head == NULL) ? a : a->head;
1855 	bhd = (b->head == NULL) ? b : b->head;
1856 	if (bhd->mtime < ahd->mtime)
1857 		return -1;
1858 	if (bhd->mtime > ahd->mtime)
1859 		return 1;
1860 	if (bhd < ahd)
1861 		return -1;
1862 	if (bhd > ahd)
1863 		return 1;
1864 	if (a->nchain != b->nchain)
1865 		return a->nchain - b->nchain;
1866 	if (a->mtime < b->mtime)
1867 		return -1;
1868 	if (a->mtime > b->mtime)
1869 		return 1;
1870 	return got_object_id_cmp(&a->id, &b->id);
1871 }
1872 
1873 static int
1874 reuse_write_order_cmp(const void *pa, const void *pb)
1875 {
1876 	struct got_pack_meta *a, *b;
1877 
1878 	a = *(struct got_pack_meta **)pa;
1879 	b = *(struct got_pack_meta **)pb;
1880 
1881 	if (a->reused_delta_offset < b->reused_delta_offset)
1882 		return -1;
1883 	if (a->reused_delta_offset > b->reused_delta_offset)
1884 		return 1;
1885 	return 0;
1886 }
1887 
1888 static const struct got_error *
1889 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1890 {
1891 	size_t i;
1892 
1893 	*hdrlen = 0;
1894 
1895 	hdr[0] = obj_type << 4;
1896 	hdr[0] |= len & 0xf;
1897 	len >>= 4;
1898 	for (i = 1; len != 0; i++){
1899 		if (i >= bufsize)
1900 			return got_error(GOT_ERR_NO_SPACE);
1901 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1902 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1903 		len >>= GOT_DELTA_SIZE_SHIFT;
1904 	}
1905 
1906 	*hdrlen = i;
1907 	return NULL;
1908 }
1909 
1910 static int
1911 packoff(char *hdr, off_t off)
1912 {
1913 	int i, j;
1914 	char rbuf[8];
1915 
1916 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1917 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1918 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1919 		    GOT_DELTA_SIZE_MORE;
1920 	}
1921 
1922 	j = 0;
1923 	while (i > 0)
1924 		hdr[j++] = rbuf[--i];
1925 	return j;
1926 }
1927 
1928 static const struct got_error *
1929 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1930     struct got_pack_meta *m)
1931 {
1932 	const struct got_error *err;
1933 	char buf[32];
1934 	int nh;
1935 
1936 	if (m->prev->off != 0) {
1937 		err = packhdr(&nh, buf, sizeof(buf),
1938 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1939 		if (err)
1940 			return err;
1941 		nh += packoff(buf + nh, m->off - m->prev->off);
1942 		err = hwrite(packfile, buf, nh, ctx);
1943 		if (err)
1944 			return err;
1945 		*packfile_size += nh;
1946 	} else {
1947 		err = packhdr(&nh, buf, sizeof(buf),
1948 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1949 		if (err)
1950 			return err;
1951 		err = hwrite(packfile, buf, nh, ctx);
1952 		if (err)
1953 			return err;
1954 		*packfile_size += nh;
1955 		err = hwrite(packfile, m->prev->id.sha1,
1956 		    sizeof(m->prev->id.sha1), ctx);
1957 		if (err)
1958 			return err;
1959 		*packfile_size += sizeof(m->prev->id.sha1);
1960 	}
1961 
1962 	return NULL;
1963 }
1964 
1965 static const struct got_error *
1966 write_packed_object(off_t *packfile_size, FILE *packfile,
1967     FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
1968     struct got_pack_meta *m, int *outfd, SHA1_CTX *ctx,
1969     struct got_repository *repo)
1970 {
1971 	const struct got_error *err = NULL;
1972 	struct got_deflate_checksum csum;
1973 	char buf[32];
1974 	int nh;
1975 	struct got_raw_object *raw = NULL;
1976 	off_t outlen;
1977 
1978 	csum.output_sha1 = ctx;
1979 	csum.output_crc = NULL;
1980 
1981 	m->off = ftello(packfile);
1982 	if (m->delta_len == 0) {
1983 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1984 		if (err)
1985 			goto done;
1986 		err = packhdr(&nh, buf, sizeof(buf),
1987 		    m->obj_type, raw->size);
1988 		if (err)
1989 			goto done;
1990 		err = hwrite(packfile, buf, nh, ctx);
1991 		if (err)
1992 			goto done;
1993 		*packfile_size += nh;
1994 		if (raw->f == NULL) {
1995 			err = got_deflate_to_file_mmap(&outlen,
1996 			    raw->data + raw->hdrlen, 0, raw->size,
1997 			    packfile, &csum);
1998 			if (err)
1999 				goto done;
2000 		} else {
2001 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
2002 			    == -1) {
2003 				err = got_error_from_errno("fseeko");
2004 				goto done;
2005 			}
2006 			err = got_deflate_to_file(&outlen, raw->f,
2007 			    raw->size, packfile, &csum);
2008 			if (err)
2009 				goto done;
2010 		}
2011 		*packfile_size += outlen;
2012 		got_object_raw_close(raw);
2013 		raw = NULL;
2014 	} else if (m->delta_buf) {
2015 		err = deltahdr(packfile_size, ctx, packfile, m);
2016 		if (err)
2017 			goto done;
2018 		err = hwrite(packfile, m->delta_buf,
2019 		    m->delta_compressed_len, ctx);
2020 		if (err)
2021 			goto done;
2022 		*packfile_size += m->delta_compressed_len;
2023 		free(m->delta_buf);
2024 		m->delta_buf = NULL;
2025 	} else if (delta_cache_map) {
2026 		err = deltahdr(packfile_size, ctx, packfile, m);
2027 		if (err)
2028 			goto done;
2029 		err = hcopy_mmap(delta_cache_map, m->delta_offset,
2030 		    delta_cache_size, packfile, m->delta_compressed_len,
2031 		    ctx);
2032 		if (err)
2033 			goto done;
2034 		*packfile_size += m->delta_compressed_len;
2035 	} else {
2036 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
2037 		    == -1) {
2038 			err = got_error_from_errno("fseeko");
2039 			goto done;
2040 		}
2041 		err = deltahdr(packfile_size, ctx, packfile, m);
2042 		if (err)
2043 			goto done;
2044 		err = hcopy(delta_cache, packfile,
2045 		    m->delta_compressed_len, ctx);
2046 		if (err)
2047 			goto done;
2048 		*packfile_size += m->delta_compressed_len;
2049 	}
2050 done:
2051 	if (raw)
2052 		got_object_raw_close(raw);
2053 	return err;
2054 }
2055 
2056 static const struct got_error *
2057 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
2058     struct got_pack_meta **deltify, int ndeltify,
2059     struct got_pack_meta **reuse, int nreuse,
2060     int ncolored, int nfound, int ntrees, int nours,
2061     struct got_repository *repo,
2062     got_pack_progress_cb progress_cb, void *progress_arg,
2063     struct got_ratelimit *rl,
2064     got_cancel_cb cancel_cb, void *cancel_arg)
2065 {
2066 	const struct got_error *err = NULL;
2067 	int i;
2068 	SHA1_CTX ctx;
2069 	struct got_pack_meta *m;
2070 	char buf[32];
2071 	size_t n;
2072 	off_t packfile_size = 0;
2073 	int outfd = -1;
2074 	int delta_cache_fd = -1;
2075 	uint8_t *delta_cache_map = NULL;
2076 	size_t delta_cache_size = 0;
2077 
2078 	SHA1Init(&ctx);
2079 
2080 #ifndef GOT_PACK_NO_MMAP
2081 	delta_cache_fd = dup(fileno(delta_cache));
2082 	if (delta_cache_fd != -1) {
2083 		struct stat sb;
2084 		if (fstat(delta_cache_fd, &sb) == -1) {
2085 			err = got_error_from_errno("fstat");
2086 			goto done;
2087 		}
2088 		if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
2089 			delta_cache_map = mmap(NULL, sb.st_size,
2090 			    PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
2091 			if (delta_cache_map == MAP_FAILED) {
2092 				if (errno != ENOMEM) {
2093 					err = got_error_from_errno("mmap");
2094 					goto done;
2095 				}
2096 				delta_cache_map = NULL; /* fallback on stdio */
2097 			} else
2098 				delta_cache_size = (size_t)sb.st_size;
2099 		}
2100 	}
2101 #endif
2102 	err = hwrite(packfile, "PACK", 4, &ctx);
2103 	if (err)
2104 		goto done;
2105 	putbe32(buf, GOT_PACKFILE_VERSION);
2106 	err = hwrite(packfile, buf, 4, &ctx);
2107 	if (err)
2108 		goto done;
2109 	putbe32(buf, ndeltify + nreuse);
2110 	err = hwrite(packfile, buf, 4, &ctx);
2111 	if (err)
2112 		goto done;
2113 
2114 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
2115 	    write_order_cmp);
2116 	for (i = 0; i < ndeltify; i++) {
2117 		err = report_progress(progress_cb, progress_arg, rl,
2118 		    ncolored, nfound, ntrees, packfile_size, nours,
2119 		    ndeltify + nreuse, ndeltify + nreuse, i);
2120 		if (err)
2121 			goto done;
2122 		m = deltify[i];
2123 		err = write_packed_object(&packfile_size, packfile,
2124 		    delta_cache, delta_cache_map, delta_cache_size,
2125 		    m, &outfd, &ctx, repo);
2126 		if (err)
2127 			goto done;
2128 	}
2129 
2130 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
2131 	    reuse_write_order_cmp);
2132 	for (i = 0; i < nreuse; i++) {
2133 		err = report_progress(progress_cb, progress_arg, rl,
2134 		    ncolored, nfound, ntrees, packfile_size, nours,
2135 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
2136 		if (err)
2137 			goto done;
2138 		m = reuse[i];
2139 		err = write_packed_object(&packfile_size, packfile,
2140 		    delta_cache, delta_cache_map, delta_cache_size,
2141 		    m, &outfd, &ctx, repo);
2142 		if (err)
2143 			goto done;
2144 	}
2145 
2146 	SHA1Final(pack_sha1, &ctx);
2147 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
2148 	if (n != SHA1_DIGEST_LENGTH)
2149 		err = got_ferror(packfile, GOT_ERR_IO);
2150 	packfile_size += SHA1_DIGEST_LENGTH;
2151 	packfile_size += sizeof(struct got_packfile_hdr);
2152 	if (progress_cb) {
2153 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2154 		    packfile_size, nours, ndeltify + nreuse,
2155 		    ndeltify + nreuse, ndeltify + nreuse);
2156 		if (err)
2157 			goto done;
2158 	}
2159 done:
2160 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
2161 		err = got_error_from_errno("close");
2162 	if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
2163 		err = got_error_from_errno("munmap");
2164 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2165 		err = got_error_from_errno("close");
2166 	return err;
2167 }
2168 
2169 static const struct got_error *
2170 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
2171 {
2172 	struct got_pack_meta *m = data;
2173 	struct got_pack_metavec *v = arg;
2174 
2175 	if (m->reused_delta_offset != 0)
2176 		return NULL;
2177 
2178 	return add_meta(m, v);
2179 }
2180 
2181 const struct got_error *
2182 got_pack_create(uint8_t *packsha1, FILE *packfile,
2183     struct got_object_id **theirs, int ntheirs,
2184     struct got_object_id **ours, int nours,
2185     struct got_repository *repo, int loose_obj_only, int allow_empty,
2186     got_pack_progress_cb progress_cb, void *progress_arg,
2187     got_cancel_cb cancel_cb, void *cancel_arg)
2188 {
2189 	const struct got_error *err;
2190 	int delta_cache_fd = -1;
2191 	FILE *delta_cache = NULL;
2192 	struct got_object_idset *idset;
2193 	struct got_ratelimit rl;
2194 	struct got_pack_metavec deltify, reuse;
2195 	int ncolored = 0, nfound = 0, ntrees = 0;
2196 	size_t ndeltify;
2197 	uint32_t seed;
2198 
2199 	seed = arc4random();
2200 
2201 	memset(&deltify, 0, sizeof(deltify));
2202 	memset(&reuse, 0, sizeof(reuse));
2203 
2204 	got_ratelimit_init(&rl, 0, 500);
2205 
2206 	idset = got_object_idset_alloc();
2207 	if (idset == NULL)
2208 		return got_error_from_errno("got_object_idset_alloc");
2209 
2210 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
2211 	    ntheirs, ours, nours, repo, seed, loose_obj_only,
2212 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
2213 	if (err)
2214 		goto done;
2215 
2216 	if (progress_cb) {
2217 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2218 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
2219 		if (err)
2220 			goto done;
2221 	}
2222 
2223 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
2224 		err = got_error(GOT_ERR_CANNOT_PACK);
2225 		goto done;
2226 	}
2227 
2228 	delta_cache_fd = got_opentempfd();
2229 	if (delta_cache_fd == -1) {
2230 		err = got_error_from_errno("got_opentemp");
2231 		goto done;
2232 	}
2233 
2234 	reuse.metasz = 64;
2235 	reuse.meta = calloc(reuse.metasz,
2236 	    sizeof(struct got_pack_meta *));
2237 	if (reuse.meta == NULL) {
2238 		err = got_error_from_errno("calloc");
2239 		goto done;
2240 	}
2241 
2242 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
2243 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
2244 	    cancel_cb, cancel_arg);
2245 	if (err)
2246 		goto done;
2247 
2248 	delta_cache = fdopen(delta_cache_fd, "a+");
2249 	if (delta_cache == NULL) {
2250 		err = got_error_from_errno("fdopen");
2251 		goto done;
2252 	}
2253 	delta_cache_fd = -1;
2254 
2255 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
2256 		err = got_error_from_errno("fseeko");
2257 		goto done;
2258 	}
2259 
2260 	ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
2261 	if (ndeltify > 0) {
2262 		deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
2263 		if (deltify.meta == NULL) {
2264 			err = got_error_from_errno("calloc");
2265 			goto done;
2266 		}
2267 		deltify.metasz = ndeltify;
2268 
2269 		err = got_object_idset_for_each(idset, add_meta_idset_cb,
2270 		    &deltify);
2271 		if (err)
2272 			goto done;
2273 		if (deltify.nmeta > 0) {
2274 			err = pick_deltas(deltify.meta, deltify.nmeta,
2275 			    ncolored, nfound, ntrees, nours, reuse.nmeta,
2276 			    delta_cache, repo, progress_cb, progress_arg, &rl,
2277 			    cancel_cb, cancel_arg);
2278 			if (err)
2279 				goto done;
2280 		}
2281 	}
2282 
2283 	if (fflush(delta_cache) == EOF) {
2284 		err = got_error_from_errno("fflush");
2285 		goto done;
2286 	}
2287 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
2288 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
2289 	    nours, repo, progress_cb, progress_arg, &rl,
2290 	    cancel_cb, cancel_arg);
2291 	if (err)
2292 		goto done;
2293 done:
2294 	free_nmeta(deltify.meta, deltify.nmeta);
2295 	free_nmeta(reuse.meta, reuse.nmeta);
2296 	got_object_idset_free(idset);
2297 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2298 		err = got_error_from_errno("close");
2299 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
2300 		err = got_error_from_errno("fclose");
2301 	return err;
2302 }