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 <inttypes.h>
31 #include <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <sha1.h>
35 #include <time.h>
36 #include <unistd.h>
37 #include <limits.h>
38 #include <zlib.h>
39 
40 #include "got_error.h"
41 #include "got_cancel.h"
42 #include "got_object.h"
43 #include "got_path.h"
44 #include "got_reference.h"
45 #include "got_repository_admin.h"
46 #include "got_opentemp.h"
47 
48 #include "got_lib_deltify.h"
49 #include "got_lib_delta.h"
50 #include "got_lib_object.h"
51 #include "got_lib_object_idset.h"
52 #include "got_lib_object_cache.h"
53 #include "got_lib_deflate.h"
54 #include "got_lib_pack.h"
55 #include "got_lib_pack_create.h"
56 #include "got_lib_privsep.h"
57 #include "got_lib_repository.h"
58 #include "got_lib_ratelimit.h"
59 #include "got_lib_inflate.h"
60 
61 #include "murmurhash2.h"
62 
63 #ifndef MIN
64 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
65 #endif
66 
67 #ifndef MAX
68 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
69 #endif
70 
71 #ifndef nitems
72 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
73 #endif
74 
75 struct got_pack_meta {
76 	struct got_object_id id;
77 	uint32_t path_hash;
78 	int	obj_type;
79 	off_t	size;
80 	time_t	mtime;
81 
82 	/* The best delta we picked */
83 	struct got_pack_meta *head;
84 	struct got_pack_meta *prev;
85 	unsigned char *delta_buf; /* if encoded in memory (compressed) */
86 	off_t	delta_offset;	/* offset in delta cache file (compressed) */
87 	off_t	delta_len;	/* encoded delta length */
88 	off_t	delta_compressed_len; /* encoded+compressed delta length */
89 	int	nchain;
90 
91 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
92 	struct got_object_id *base_obj_id;
93 
94 	/* Only used for delta window */
95 	struct got_delta_table *dtab;
96 
97 	/* Only used for writing offset deltas */
98 	off_t	off;
99 };
100 
101 struct got_pack_metavec {
102 	struct got_pack_meta **meta;
103 	int nmeta;
104 	int metasz;
105 };
106 
107 static const struct got_error *
108 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
109     const char *path, int obj_type, time_t mtime, uint32_t seed)
110 {
111 	struct got_pack_meta *m;
112 
113 	*new = NULL;
114 
115 	m = calloc(1, sizeof(*m));
116 	if (m == NULL)
117 		return got_error_from_errno("calloc");
118 
119 	memcpy(&m->id, id, sizeof(m->id));
120 
121 	m->path_hash = murmurhash2(path, strlen(path), seed);
122 	m->obj_type = obj_type;
123 	m->mtime = mtime;
124 	*new = m;
125 	return NULL;
126 }
127 
128 static void
129 clear_meta(struct got_pack_meta *meta)
130 {
131 	if (meta == NULL)
132 		return;
133 	meta->path_hash = 0;
134 	free(meta->delta_buf);
135 	meta->delta_buf = NULL;
136 	free(meta->base_obj_id);
137 	meta->base_obj_id = NULL;
138 	meta->reused_delta_offset = 0;
139 }
140 
141 static void
142 free_nmeta(struct got_pack_meta **meta, int nmeta)
143 {
144 	int i;
145 
146 	for (i = 0; i < nmeta; i++)
147 		clear_meta(meta[i]);
148 	free(meta);
149 }
150 
151 static int
152 delta_order_cmp(const void *pa, const void *pb)
153 {
154 	struct got_pack_meta *a, *b;
155 
156 	a = *(struct got_pack_meta **)pa;
157 	b = *(struct got_pack_meta **)pb;
158 
159 	if (a->obj_type != b->obj_type)
160 		return a->obj_type - b->obj_type;
161 	if (a->path_hash < b->path_hash)
162 		return -1;
163 	if (a->path_hash > b->path_hash)
164 		return 1;
165 	if (a->mtime < b->mtime)
166 		return -1;
167 	if (a->mtime > b->mtime)
168 		return 1;
169 	return got_object_id_cmp(&a->id, &b->id);
170 }
171 
172 static off_t
173 delta_size(struct got_delta_instruction *deltas, int ndeltas)
174 {
175 	int i;
176 	off_t size = 32;
177 	for (i = 0; i < ndeltas; i++) {
178 		if (deltas[i].copy)
179 			size += GOT_DELTA_SIZE_SHIFT;
180 		else
181 			size += deltas[i].len + 1;
182 	}
183 	return size;
184 }
185 
186 static const struct got_error *
187 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
188 {
189 	char *n;
190 
191 	if (*len + nseg >= *sz) {
192 		while (*len + nseg >= *sz)
193 			*sz += *sz / 2;
194 		n = realloc(*p, *sz);
195 		if (n == NULL)
196 			return got_error_from_errno("realloc");
197 		*p = n;
198 	}
199 	memcpy(*p + *len, seg, nseg);
200 	*len += nseg;
201 	return NULL;
202 }
203 
204 static const struct got_error *
205 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
206     struct got_delta_instruction *deltas, int ndeltas,
207     off_t delta_size, off_t base_size)
208 {
209 	const struct got_error *err;
210 	unsigned char buf[16], *bp;
211 	int i, j;
212 	size_t len = 0, compressed_len;
213 	off_t bufsize = delta_size;
214 	off_t n;
215 	struct got_delta_instruction *d;
216 	uint8_t *delta_buf;
217 
218 	delta_buf = malloc(bufsize);
219 	if (delta_buf == NULL)
220 		return got_error_from_errno("malloc");
221 
222 	/* base object size */
223 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
224 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
225 	for (i = 1; n > 0; i++) {
226 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
227 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
228 		n >>= GOT_DELTA_SIZE_SHIFT;
229 	}
230 	err = append(&delta_buf, &len, &bufsize, buf, i);
231 	if (err)
232 		goto done;
233 
234 	/* target object size */
235 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
236 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
237 	for (i = 1; n > 0; i++) {
238 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
239 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
240 		n >>= GOT_DELTA_SIZE_SHIFT;
241 	}
242 	err = append(&delta_buf, &len, &bufsize, buf, i);
243 	if (err)
244 		goto done;
245 
246 	for (j = 0; j < ndeltas; j++) {
247 		d = &deltas[j];
248 		if (d->copy) {
249 			n = d->offset;
250 			bp = &buf[1];
251 			buf[0] = GOT_DELTA_BASE_COPY;
252 			for (i = 0; i < 4; i++) {
253 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
254 				buf[0] |= 1 << i;
255 				*bp++ = n & 0xff;
256 				n >>= 8;
257 				if (n == 0)
258 					break;
259 			}
260 
261 			n = d->len;
262 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
263 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
264 				for (i = 0; i < 3 && n > 0; i++) {
265 					buf[0] |= 1 << (i + 4);
266 					*bp++ = n & 0xff;
267 					n >>= 8;
268 				}
269 			}
270 			err = append(&delta_buf, &len, &bufsize,
271 			    buf, bp - buf);
272 			if (err)
273 				goto done;
274 		} else if (o->f == NULL) {
275 			n = 0;
276 			while (n != d->len) {
277 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
278 				err = append(&delta_buf, &len, &bufsize,
279 				    buf, 1);
280 				if (err)
281 					goto done;
282 				err = append(&delta_buf, &len, &bufsize,
283 				    o->data + o->hdrlen + d->offset + n,
284 				    buf[0]);
285 				if (err)
286 					goto done;
287 				n += buf[0];
288 			}
289 		} else {
290 			char content[128];
291 			size_t r;
292 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
293 				err = got_error_from_errno("fseeko");
294 				goto done;
295 			}
296 			n = 0;
297 			while (n != d->len) {
298 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
299 				err = append(&delta_buf, &len, &bufsize,
300 				    buf, 1);
301 				if (err)
302 					goto done;
303 				r = fread(content, 1, buf[0], o->f);
304 				if (r != buf[0]) {
305 					err = got_ferror(o->f, GOT_ERR_IO);
306 					goto done;
307 				}
308 				err = append(&delta_buf, &len, &bufsize,
309 				    content, buf[0]);
310 				if (err)
311 					goto done;
312 				n += buf[0];
313 			}
314 		}
315 	}
316 
317 	err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
318 	    NULL, NULL, delta_buf, 0, len);
319 	if (err)
320 		goto done;
321 
322 	m->delta_len = len;
323 	m->delta_compressed_len = compressed_len;
324 done:
325 	free(delta_buf);
326 	return err;
327 }
328 
329 static const struct got_error *
330 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
331     struct got_delta_instruction *deltas, int ndeltas,
332     off_t base_size, FILE *f)
333 {
334 	const struct got_error *err;
335 	unsigned char buf[16], *bp;
336 	int i, j;
337 	off_t n;
338 	struct got_deflate_buf zb;
339 	struct got_delta_instruction *d;
340 	off_t delta_len = 0, compressed_len = 0;
341 
342 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
343 	if (err)
344 		return err;
345 
346 	/* base object size */
347 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
348 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
349 	for (i = 1; n > 0; i++) {
350 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
351 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
352 		n >>= GOT_DELTA_SIZE_SHIFT;
353 	}
354 
355 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
356 	    buf, 0, i, f, NULL);
357 	if (err)
358 		goto done;
359 	delta_len += i;
360 
361 	/* target object size */
362 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
363 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
364 	for (i = 1; n > 0; i++) {
365 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
366 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
367 		n >>= GOT_DELTA_SIZE_SHIFT;
368 	}
369 
370 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
371 	    buf, 0, i, f, NULL);
372 	if (err)
373 		goto done;
374 	delta_len += i;
375 
376 	for (j = 0; j < ndeltas; j++) {
377 		d = &deltas[j];
378 		if (d->copy) {
379 			n = d->offset;
380 			bp = &buf[1];
381 			buf[0] = GOT_DELTA_BASE_COPY;
382 			for (i = 0; i < 4; i++) {
383 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
384 				buf[0] |= 1 << i;
385 				*bp++ = n & 0xff;
386 				n >>= 8;
387 				if (n == 0)
388 					break;
389 			}
390 			n = d->len;
391 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
392 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
393 				for (i = 0; i < 3 && n > 0; i++) {
394 					buf[0] |= 1 << (i + 4);
395 					*bp++ = n & 0xff;
396 					n >>= 8;
397 				}
398 			}
399 			err = got_deflate_append_to_file_mmap(&zb,
400 			    &compressed_len, buf, 0, bp - buf, f, NULL);
401 			if (err)
402 				goto done;
403 			delta_len += (bp - buf);
404 		} else if (o->f == NULL) {
405 			n = 0;
406 			while (n != d->len) {
407 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
408 				err = got_deflate_append_to_file_mmap(&zb,
409 				    &compressed_len, buf, 0, 1, f, NULL);
410 				if (err)
411 					goto done;
412 				delta_len++;
413 				err = got_deflate_append_to_file_mmap(&zb,
414 				    &compressed_len,
415 				    o->data + o->hdrlen + d->offset + n, 0,
416 				    buf[0], f, NULL);
417 				if (err)
418 					goto done;
419 				delta_len += buf[0];
420 				n += buf[0];
421 			}
422 		} else {
423 			char content[128];
424 			size_t r;
425 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
426 				err = got_error_from_errno("fseeko");
427 				goto done;
428 			}
429 			n = 0;
430 			while (n != d->len) {
431 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
432 				err = got_deflate_append_to_file_mmap(&zb,
433 				    &compressed_len, buf, 0, 1, f, NULL);
434 				if (err)
435 					goto done;
436 				delta_len++;
437 				r = fread(content, 1, buf[0], o->f);
438 				if (r != buf[0]) {
439 					err = got_ferror(o->f, GOT_ERR_IO);
440 					goto done;
441 				}
442 				err = got_deflate_append_to_file_mmap(&zb,
443 				    &compressed_len, content, 0, buf[0], f,
444 				    NULL);
445 				if (err)
446 					goto done;
447 				delta_len += buf[0];
448 				n += buf[0];
449 			}
450 		}
451 	}
452 
453 	err = got_deflate_flush(&zb, f, NULL, &compressed_len);
454 	if (err)
455 		goto done;
456 
457 	/* sanity check */
458 	if (compressed_len != ftello(f) - m->delta_offset) {
459 		err = got_error(GOT_ERR_COMPRESSION);
460 		goto done;
461 	}
462 
463 	m->delta_len = delta_len;
464 	m->delta_compressed_len = compressed_len;
465 done:
466 	got_deflate_end(&zb);
467 	return err;
468 }
469 
470 static const struct got_error *
471 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
472     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
473     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
474     int nobj_written)
475 {
476 	const struct got_error *err;
477 	int elapsed;
478 
479 	if (progress_cb == NULL)
480 		return NULL;
481 
482 	err = got_ratelimit_check(&elapsed, rl);
483 	if (err || !elapsed)
484 		return err;
485 
486 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
487 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
488 }
489 
490 static const struct got_error *
491 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
492 {
493 	if (v->nmeta == v->metasz){
494 		size_t newsize = 2 * v->metasz;
495 		struct got_pack_meta **new;
496 		new = reallocarray(v->meta, newsize, sizeof(*new));
497 		if (new == NULL)
498 			return got_error_from_errno("reallocarray");
499 		v->meta = new;
500 		v->metasz = newsize;
501 	}
502 
503 	v->meta[v->nmeta++] = m;
504 	return NULL;
505 }
506 
507 static const struct got_error *
508 find_pack_for_reuse(struct got_packidx **best_packidx,
509     struct got_repository *repo)
510 {
511 	const struct got_error *err = NULL;
512 	struct got_pathlist_entry *pe;
513 	const char *best_packidx_path = NULL;
514 	int nobj_max = 0;
515 
516 	*best_packidx = NULL;
517 
518 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
519 		const char *path_packidx = pe->path;
520 		struct got_packidx *packidx;
521 		int nobj;
522 
523 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
524 		if (err)
525 			break;
526 
527 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
528 		if (nobj > nobj_max) {
529 			best_packidx_path = path_packidx;
530 			nobj_max = nobj;
531 		}
532 	}
533 
534 	if (best_packidx_path) {
535 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
536 		    repo);
537 	}
538 
539 	return err;
540 }
541 
542 struct send_id_arg {
543 	struct imsgbuf *ibuf;
544 	struct got_object_id *ids[GOT_IMSG_OBJ_ID_LIST_MAX_NIDS];
545 	size_t nids;
546 };
547 
548 static const struct got_error *
549 send_id(struct got_object_id *id, void *data, void *arg)
550 {
551 	const struct got_error *err = NULL;
552 	struct send_id_arg *a = arg;
553 
554 	a->ids[a->nids++] = id;
555 
556 	if (a->nids >= GOT_IMSG_OBJ_ID_LIST_MAX_NIDS) {
557 		err = got_privsep_send_object_idlist(a->ibuf, a->ids, a->nids);
558 		if (err)
559 			return err;
560 		a->nids = 0;
561 	}
562 
563 	return NULL;
564 }
565 
566 static const struct got_error *
567 send_idset(struct imsgbuf *ibuf, struct got_object_idset *idset)
568 {
569 	const struct got_error *err;
570 	struct send_id_arg sia;
571 
572 	memset(&sia, 0, sizeof(sia));
573 	sia.ibuf = ibuf;
574 	err = got_object_idset_for_each(idset, send_id, &sia);
575 	if (err)
576 		return err;
577 
578 	if (sia.nids > 0) {
579 		err = got_privsep_send_object_idlist(ibuf, sia.ids, sia.nids);
580 		if (err)
581 			return err;
582 	}
583 
584 	return got_privsep_send_object_idlist_done(ibuf);
585 }
586 
587 
588 static const struct got_error *
589 recv_reused_delta(struct got_imsg_reused_delta *delta,
590     struct got_object_idset *idset, struct got_pack_metavec *v)
591 {
592 	struct got_pack_meta *m, *base;
593 
594 	if (delta->delta_offset + delta->delta_size < delta->delta_offset ||
595 	    delta->delta_offset +
596 	    delta->delta_compressed_size < delta->delta_offset)
597 		return got_error(GOT_ERR_BAD_PACKFILE);
598 
599 	m = got_object_idset_get(idset, &delta->id);
600 	if (m == NULL)
601 		return got_error(GOT_ERR_NO_OBJ);
602 
603 	base = got_object_idset_get(idset, &delta->base_id);
604 	if (base == NULL)
605 		return got_error(GOT_ERR_NO_OBJ);
606 
607 	m->delta_len = delta->delta_size;
608 	m->delta_compressed_len = delta->delta_compressed_size;
609 	m->delta_offset = delta->delta_out_offset;
610 	m->prev = base;
611 	m->size = delta->result_size;
612 	m->reused_delta_offset = delta->delta_offset;
613 	m->base_obj_id = got_object_id_dup(&delta->base_id);
614 	if (m->base_obj_id == NULL)
615 		return got_error_from_errno("got_object_id_dup");
616 
617 	return add_meta(m, v);
618 }
619 
620 static const struct got_error *
621 cache_pack_for_packidx(struct got_pack **pack, struct got_packidx *packidx,
622     struct got_repository *repo)
623 {
624 	const struct got_error *err;
625 	char *path_packfile = NULL;
626 
627 	err = got_packidx_get_packfile_path(&path_packfile,
628 	    packidx->path_packidx);
629 	if (err)
630 		return err;
631 
632 	*pack = got_repo_get_cached_pack(repo, path_packfile);
633 	if (*pack == NULL) {
634 		err = got_repo_cache_pack(pack, repo, path_packfile, packidx);
635 		if (err)
636 			goto done;
637 	}
638 	if ((*pack)->privsep_child == NULL) {
639 		err = got_pack_start_privsep_child(*pack, packidx);
640 		if (err)
641 			goto done;
642 	}
643 done:
644 	free(path_packfile);
645 	return err;
646 }
647 
648 static const struct got_error *
649 prepare_delta_reuse(struct got_pack *pack, struct got_packidx *packidx,
650     int delta_outfd, struct got_repository *repo)
651 {
652 	const struct got_error *err = NULL;
653 
654 	if (!pack->child_has_delta_outfd) {
655 		int outfd_child;
656 		outfd_child = dup(delta_outfd);
657 		if (outfd_child == -1) {
658 			err = got_error_from_errno("dup");
659 			goto done;
660 		}
661 		err = got_privsep_send_raw_delta_outfd(
662 		    pack->privsep_child->ibuf, outfd_child);
663 		if (err)
664 			goto done;
665 		pack->child_has_delta_outfd = 1;
666 	}
667 
668 	err = got_privsep_send_delta_reuse_req(pack->privsep_child->ibuf);
669 done:
670 	return err;
671 }
672 
673 
674 static const struct got_error *
675 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
676     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
677     struct got_repository *repo,
678     got_pack_progress_cb progress_cb, void *progress_arg,
679     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
680 {
681 	const struct got_error *err = NULL;
682 	struct got_packidx *packidx;
683 	struct got_pack *pack;
684 	struct got_imsg_reused_delta deltas[GOT_IMSG_REUSED_DELTAS_MAX_NDELTAS];
685 	size_t ndeltas, i;
686 
687 	err = find_pack_for_reuse(&packidx, repo);
688 	if (err)
689 		return err;
690 
691 	if (packidx == NULL)
692 		return NULL;
693 
694 	err = cache_pack_for_packidx(&pack, packidx, repo);
695 	if (err)
696 		return err;
697 
698 	err = prepare_delta_reuse(pack, packidx, delta_cache_fd, repo);
699 	if (err)
700 		return err;
701 
702 	err = send_idset(pack->privsep_child->ibuf, idset);
703 	if (err)
704 		return err;
705 
706 	for (;;) {
707 		int done = 0;
708 
709 		if (cancel_cb) {
710 			err = (*cancel_cb)(cancel_arg);
711 			if (err)
712 				break;
713 		}
714 
715 		err = got_privsep_recv_reused_deltas(&done, deltas, &ndeltas,
716 		    pack->privsep_child->ibuf);
717 		if (err || done)
718 			break;
719 
720 		for (i = 0; i < ndeltas; i++) {
721 			struct got_imsg_reused_delta *delta = &deltas[i];
722 			err = recv_reused_delta(delta, idset, v);
723 			if (err)
724 				goto done;
725 		}
726 
727 		err = report_progress(progress_cb, progress_arg, rl,
728 		    ncolored, nfound, ntrees, 0L, ncommits,
729 		    got_object_idset_num_elements(idset), v->nmeta, 0);
730 		if (err)
731 			break;
732 	}
733 done:
734 	return err;
735 }
736 
737 static const struct got_error *
738 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
739     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
740     struct got_repository *repo,
741     got_pack_progress_cb progress_cb, void *progress_arg,
742     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
743 {
744 	const struct got_error *err = NULL;
745 	struct got_pack_meta *m = NULL, *base = NULL;
746 	struct got_raw_object *raw = NULL, *base_raw = NULL;
747 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
748 	int i, j, ndeltas, best_ndeltas;
749 	off_t size, best_size;
750 	const int max_base_candidates = 3;
751 	size_t delta_memsize = 0;
752 	const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
753 	int outfd = -1;
754 	uint32_t delta_seed;
755 
756 	delta_seed = arc4random();
757 
758 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
759 	for (i = 0; i < nmeta; i++) {
760 		if (cancel_cb) {
761 			err = (*cancel_cb)(cancel_arg);
762 			if (err)
763 				break;
764 		}
765 		err = report_progress(progress_cb, progress_arg, rl,
766 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
767 		    nreused + i, 0);
768 		if (err)
769 			goto done;
770 		m = meta[i];
771 
772 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
773 		    m->obj_type == GOT_OBJ_TYPE_TAG)
774 			continue;
775 
776 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
777 		if (err)
778 			goto done;
779 		m->size = raw->size;
780 
781 		if (raw->f == NULL) {
782 			err = got_deltify_init_mem(&m->dtab, raw->data,
783 			    raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
784 		} else {
785 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
786 			    raw->size + raw->hdrlen, delta_seed);
787 		}
788 		if (err)
789 			goto done;
790 
791 		if (i > max_base_candidates) {
792 			struct got_pack_meta *n = NULL;
793 			n = meta[i - (max_base_candidates + 1)];
794 			got_deltify_free(n->dtab);
795 			n->dtab = NULL;
796 		}
797 
798 		best_size = raw->size;
799 		best_ndeltas = 0;
800 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
801 			if (cancel_cb) {
802 				err = (*cancel_cb)(cancel_arg);
803 				if (err)
804 					goto done;
805 			}
806 			base = meta[j];
807 			/* long chains make unpacking slow, avoid such bases */
808 			if (base->nchain >= 128 ||
809 			    base->obj_type != m->obj_type)
810 				continue;
811 
812 			err = got_object_raw_open(&base_raw, &outfd, repo,
813 			    &base->id);
814 			if (err)
815 				goto done;
816 
817 			if (raw->f == NULL && base_raw->f == NULL) {
818 				err = got_deltify_mem_mem(&deltas, &ndeltas,
819 				    raw->data, raw->hdrlen,
820 				    raw->size + raw->hdrlen, delta_seed,
821 				    base->dtab, base_raw->data,
822 				    base_raw->hdrlen,
823 				    base_raw->size + base_raw->hdrlen);
824 			} else if (raw->f == NULL) {
825 				err = got_deltify_mem_file(&deltas, &ndeltas,
826 				    raw->data, raw->hdrlen,
827 				    raw->size + raw->hdrlen, delta_seed,
828 				    base->dtab, base_raw->f,
829 				    base_raw->hdrlen,
830 				    base_raw->size + base_raw->hdrlen);
831 			} else if (base_raw->f == NULL) {
832 				err = got_deltify_file_mem(&deltas, &ndeltas,
833 				    raw->f, raw->hdrlen,
834 				    raw->size + raw->hdrlen, delta_seed,
835 				    base->dtab, base_raw->data,
836 				    base_raw->hdrlen,
837 				    base_raw->size + base_raw->hdrlen);
838 			} else {
839 				err = got_deltify(&deltas, &ndeltas,
840 				    raw->f, raw->hdrlen,
841 				    raw->size + raw->hdrlen, delta_seed,
842 				    base->dtab, base_raw->f, base_raw->hdrlen,
843 				    base_raw->size + base_raw->hdrlen);
844 			}
845 			got_object_raw_close(base_raw);
846 			base_raw = NULL;
847 			if (err)
848 				goto done;
849 
850 			size = delta_size(deltas, ndeltas);
851 			if (size + 32 < best_size){
852 				/*
853 				 * if we already picked a best delta,
854 				 * replace it.
855 				 */
856 				best_size = size;
857 				free(best_deltas);
858 				best_deltas = deltas;
859 				best_ndeltas = ndeltas;
860 				deltas = NULL;
861 				m->nchain = base->nchain + 1;
862 				m->prev = base;
863 				m->head = base->head;
864 				if (m->head == NULL)
865 					m->head = base;
866 			} else {
867 				free(deltas);
868 				deltas = NULL;
869 				ndeltas = 0;
870 			}
871 		}
872 
873 		if (best_ndeltas > 0) {
874 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
875 			    delta_memsize + best_size <= max_delta_memsize) {
876 				delta_memsize += best_size;
877 				err = encode_delta_in_mem(m, raw, best_deltas,
878 				    best_ndeltas, best_size, m->prev->size);
879 			} else {
880 				m->delta_offset = ftello(delta_cache);
881 				err = encode_delta(m, raw, best_deltas,
882 				    best_ndeltas, m->prev->size, delta_cache);
883 			}
884 			free(best_deltas);
885 			best_deltas = NULL;
886 			best_ndeltas = 0;
887 			if (err)
888 				goto done;
889 		}
890 
891 		got_object_raw_close(raw);
892 		raw = NULL;
893 	}
894 done:
895 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
896 		got_deltify_free(meta[i]->dtab);
897 		meta[i]->dtab = NULL;
898 	}
899 	if (raw)
900 		got_object_raw_close(raw);
901 	if (base_raw)
902 		got_object_raw_close(base_raw);
903 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
904 		err = got_error_from_errno("close");
905 	free(deltas);
906 	free(best_deltas);
907 	return err;
908 }
909 
910 static const struct got_error *
911 search_packidx(int *found, struct got_object_id *id,
912     struct got_repository *repo)
913 {
914 	const struct got_error *err = NULL;
915 	struct got_packidx *packidx = NULL;
916 	int idx;
917 
918 	*found = 0;
919 
920 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
921 	if (err == NULL)
922 		*found = 1; /* object is already packed */
923 	else if (err->code == GOT_ERR_NO_OBJ)
924 		err = NULL;
925 	return err;
926 }
927 
928 static const struct got_error *
929 add_object(int want_meta, struct got_object_idset *idset,
930     struct got_object_id *id, const char *path, int obj_type,
931     time_t mtime, uint32_t seed, int loose_obj_only,
932     struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
933     got_pack_progress_cb progress_cb, void *progress_arg,
934     struct got_ratelimit *rl)
935 {
936 	const struct got_error *err;
937 	struct got_pack_meta *m = NULL;
938 
939 	if (loose_obj_only) {
940 		int is_packed;
941 		err = search_packidx(&is_packed, id, repo);
942 		if (err)
943 			return err;
944 		if (is_packed && want_meta)
945 			return NULL;
946 	}
947 
948 	if (want_meta) {
949 		err = alloc_meta(&m, id, path, obj_type, mtime, seed);
950 		if (err)
951 			return err;
952 
953 		(*nfound)++;
954 		err = report_progress(progress_cb, progress_arg, rl,
955 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
956 		if (err) {
957 			clear_meta(m);
958 			free(m);
959 			return err;
960 		}
961 	}
962 
963 	err = got_object_idset_add(idset, id, m);
964 	if (err) {
965 		clear_meta(m);
966 		free(m);
967 	}
968 	return err;
969 }
970 
971 static const struct got_error *
972 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
973     struct got_object_idset *idset, struct got_object_idset *idset_exclude,
974     struct got_tree_object *tree,
975     const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
976     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
977     got_pack_progress_cb progress_cb, void *progress_arg,
978     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
979 {
980 	const struct got_error *err;
981 	char *p = NULL;
982 	int i;
983 
984 	(*ntrees)++;
985 	err = report_progress(progress_cb, progress_arg, rl,
986 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
987 	if (err)
988 		return err;
989 
990 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
991 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
992 		struct got_object_id *id = got_tree_entry_get_id(e);
993 		mode_t mode = got_tree_entry_get_mode(e);
994 
995 		if (cancel_cb) {
996 			err = (*cancel_cb)(cancel_arg);
997 			if (err)
998 				break;
999 		}
1000 
1001 		if (got_object_tree_entry_is_submodule(e) ||
1002 		    got_object_idset_contains(idset, id) ||
1003 		    got_object_idset_contains(idset_exclude, id))
1004 			continue;
1005 
1006 		/*
1007 		 * If got-read-pack is crawling trees for us then
1008 		 * we are only here to collect blob IDs.
1009 		 */
1010 		if (ids == NULL && S_ISDIR(mode))
1011 			continue;
1012 
1013 		if (asprintf(&p, "%s%s%s", dpath,
1014 		    got_path_is_root_dir(dpath) ? "" : "/",
1015 		    got_tree_entry_get_name(e)) == -1) {
1016 			err = got_error_from_errno("asprintf");
1017 			break;
1018 		}
1019 
1020 		if (S_ISDIR(mode)) {
1021 			struct got_object_qid *qid;
1022 			err = got_object_qid_alloc(&qid, id);
1023 			if (err)
1024 				break;
1025 			qid->data = p;
1026 			p = NULL;
1027 			STAILQ_INSERT_TAIL(ids, qid, entry);
1028 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
1029 			err = add_object(want_meta,
1030 			    want_meta ? idset : idset_exclude, id, p,
1031 			    GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
1032 			    repo, ncolored, nfound, ntrees,
1033 			    progress_cb, progress_arg, rl);
1034 			if (err)
1035 				break;
1036 			free(p);
1037 			p = NULL;
1038 		} else {
1039 			free(p);
1040 			p = NULL;
1041 		}
1042 	}
1043 
1044 	free(p);
1045 	return err;
1046 }
1047 
1048 static const struct got_error *
1049 load_tree(int want_meta, struct got_object_idset *idset,
1050     struct got_object_idset *idset_exclude,
1051     struct got_object_id *tree_id, const char *dpath, time_t mtime,
1052     uint32_t seed, struct got_repository *repo, int loose_obj_only,
1053     int *ncolored, int *nfound, int *ntrees,
1054     got_pack_progress_cb progress_cb, void *progress_arg,
1055     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1056 {
1057 	const struct got_error *err = NULL;
1058 	struct got_object_id_queue tree_ids;
1059 	struct got_object_qid *qid;
1060 	struct got_tree_object *tree = NULL;
1061 
1062 	if (got_object_idset_contains(idset, tree_id) ||
1063 	    got_object_idset_contains(idset_exclude, tree_id))
1064 		return NULL;
1065 
1066 	err = got_object_qid_alloc(&qid, tree_id);
1067 	if (err)
1068 		return err;
1069 	qid->data = strdup(dpath);
1070 	if (qid->data == NULL) {
1071 		err = got_error_from_errno("strdup");
1072 		got_object_qid_free(qid);
1073 		return err;
1074 	}
1075 
1076 	STAILQ_INIT(&tree_ids);
1077 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
1078 
1079 	while (!STAILQ_EMPTY(&tree_ids)) {
1080 		const char *path;
1081 		if (cancel_cb) {
1082 			err = (*cancel_cb)(cancel_arg);
1083 			if (err)
1084 				break;
1085 		}
1086 
1087 		qid = STAILQ_FIRST(&tree_ids);
1088 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
1089 		path = qid->data;
1090 
1091 		if (got_object_idset_contains(idset, &qid->id) ||
1092 		    got_object_idset_contains(idset_exclude, &qid->id)) {
1093 			free(qid->data);
1094 			got_object_qid_free(qid);
1095 			continue;
1096 		}
1097 
1098 		err = add_object(want_meta, want_meta ? idset : idset_exclude,
1099 		    &qid->id, path, GOT_OBJ_TYPE_TREE,
1100 		    mtime, seed, loose_obj_only, repo,
1101 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1102 		if (err) {
1103 			free(qid->data);
1104 			got_object_qid_free(qid);
1105 			break;
1106 		}
1107 
1108 		err = got_object_open_as_tree(&tree, repo, &qid->id);
1109 		if (err) {
1110 			free(qid->data);
1111 			got_object_qid_free(qid);
1112 			break;
1113 		}
1114 
1115 		err = load_tree_entries(&tree_ids, want_meta, idset,
1116 		    idset_exclude, tree, path, mtime, seed, repo,
1117 		    loose_obj_only, ncolored, nfound, ntrees,
1118 		    progress_cb, progress_arg, rl,
1119 		    cancel_cb, cancel_arg);
1120 		free(qid->data);
1121 		got_object_qid_free(qid);
1122 		if (err)
1123 			break;
1124 
1125 		got_object_tree_close(tree);
1126 		tree = NULL;
1127 	}
1128 
1129 	STAILQ_FOREACH(qid, &tree_ids, entry)
1130 		free(qid->data);
1131 	got_object_id_queue_free(&tree_ids);
1132 	if (tree)
1133 		got_object_tree_close(tree);
1134 	return err;
1135 }
1136 
1137 static const struct got_error *
1138 load_commit(int want_meta, struct got_object_idset *idset,
1139     struct got_object_idset *idset_exclude,
1140     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1141     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1142     got_pack_progress_cb progress_cb, void *progress_arg,
1143     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1144 {
1145 	const struct got_error *err;
1146 	struct got_commit_object *commit;
1147 
1148 	if (got_object_idset_contains(idset, id) ||
1149 	    got_object_idset_contains(idset_exclude, id))
1150 		return NULL;
1151 
1152 	if (loose_obj_only) {
1153 		int is_packed;
1154 		err = search_packidx(&is_packed, id, repo);
1155 		if (err)
1156 			return err;
1157 		if (is_packed && want_meta)
1158 			return NULL;
1159 	}
1160 
1161 	err = got_object_open_as_commit(&commit, repo, id);
1162 	if (err)
1163 		return err;
1164 
1165 	err = add_object(want_meta, want_meta ? idset : idset_exclude,
1166 	    id, "", GOT_OBJ_TYPE_COMMIT,
1167 	    got_object_commit_get_committer_time(commit), seed,
1168 	    loose_obj_only, repo,
1169 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1170 	if (err)
1171 		goto done;
1172 
1173 	err = load_tree(want_meta, idset, idset_exclude,
1174 	    got_object_commit_get_tree_id(commit),
1175 	    "", got_object_commit_get_committer_time(commit), seed,
1176 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1177 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1178 done:
1179 	got_object_commit_close(commit);
1180 	return err;
1181 }
1182 
1183 static const struct got_error *
1184 load_tag(int want_meta, struct got_object_idset *idset,
1185     struct got_object_idset *idset_exclude,
1186     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1187     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1188     got_pack_progress_cb progress_cb, void *progress_arg,
1189     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1190 {
1191 	const struct got_error *err;
1192 	struct got_tag_object *tag = NULL;
1193 
1194 	if (got_object_idset_contains(idset, id) ||
1195 	    got_object_idset_contains(idset_exclude, id))
1196 		return NULL;
1197 
1198 	if (loose_obj_only) {
1199 		int is_packed;
1200 		err = search_packidx(&is_packed, id, repo);
1201 		if (err)
1202 			return err;
1203 		if (is_packed && want_meta)
1204 			return NULL;
1205 	}
1206 
1207 	err = got_object_open_as_tag(&tag, repo, id);
1208 	if (err)
1209 		return err;
1210 
1211 	err = add_object(want_meta, want_meta ? idset : idset_exclude,
1212 	    id, "", GOT_OBJ_TYPE_TAG,
1213 	    got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1214 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1215 	if (err)
1216 		goto done;
1217 
1218 	switch (got_object_tag_get_object_type(tag)) {
1219 	case GOT_OBJ_TYPE_COMMIT:
1220 		err = load_commit(want_meta, idset, idset_exclude,
1221 		    got_object_tag_get_object_id(tag), repo, seed,
1222 		    loose_obj_only, ncolored, nfound, ntrees,
1223 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1224 		break;
1225 	case GOT_OBJ_TYPE_TREE:
1226 		err = load_tree(want_meta, idset, idset_exclude,
1227 		    got_object_tag_get_object_id(tag), "",
1228 		    got_object_tag_get_tagger_time(tag), seed, repo,
1229 		    loose_obj_only, ncolored, nfound, ntrees,
1230 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1231 		break;
1232 	default:
1233 		break;
1234 	}
1235 
1236 done:
1237 	got_object_tag_close(tag);
1238 	return err;
1239 }
1240 
1241 enum findtwixt_color {
1242 	COLOR_KEEP = 0,
1243 	COLOR_DROP,
1244 	COLOR_SKIP,
1245 	COLOR_MAX,
1246 };
1247 
1248 static const struct got_error *
1249 paint_commit(struct got_object_qid *qid, intptr_t color)
1250 {
1251 	if (color < 0 || color >= COLOR_MAX)
1252 		return got_error(GOT_ERR_RANGE);
1253 
1254 	qid->data = (void *)color;
1255 	return NULL;
1256 }
1257 
1258 static const struct got_error *
1259 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1260     intptr_t color, struct got_repository *repo)
1261 {
1262 	const struct got_error *err;
1263 	struct got_object_qid *qid;
1264 
1265 	err = got_object_qid_alloc(&qid, id);
1266 	if (err)
1267 		return err;
1268 
1269 	STAILQ_INSERT_TAIL(ids, qid, entry);
1270 	return paint_commit(qid, color);
1271 }
1272 
1273 struct append_id_arg {
1274 	struct got_object_id **array;
1275 	int idx;
1276 	struct got_object_idset *drop;
1277 	struct got_object_idset *skip;
1278 };
1279 
1280 static const struct got_error *
1281 append_id(struct got_object_id *id, void *data, void *arg)
1282 {
1283 	struct append_id_arg *a = arg;
1284 
1285 	if (got_object_idset_contains(a->skip, id) ||
1286 	    got_object_idset_contains(a->drop, id))
1287 		return NULL;
1288 
1289 	a->array[++a->idx] = got_object_id_dup(id);
1290 	if (a->array[a->idx] == NULL)
1291 		return got_error_from_errno("got_object_id_dup");
1292 
1293 	return NULL;
1294 }
1295 
1296 static const struct got_error *
1297 queue_commit_or_tag_id(struct got_object_id *id, intptr_t color,
1298     struct got_object_id_queue *ids, struct got_repository *repo)
1299 {
1300 	const struct got_error *err;
1301 	struct got_tag_object *tag = NULL;
1302 	int obj_type;
1303 
1304 	err = got_object_get_type(&obj_type, repo, id);
1305 	if (err)
1306 		return err;
1307 
1308 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1309 		err = got_object_open_as_tag(&tag, repo, id);
1310 		if (err)
1311 			return err;
1312 		obj_type = got_object_tag_get_object_type(tag);
1313 		id = got_object_tag_get_object_id(tag);
1314 	}
1315 
1316 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1317 		err = queue_commit_id(ids, id, color, repo);
1318 		if (err)
1319 			goto done;
1320 	}
1321 done:
1322 	if (tag)
1323 		got_object_tag_close(tag);
1324 	return err;
1325 }
1326 
1327 struct recv_painted_commit_arg {
1328 	int *ncolored;
1329 	int *nqueued;
1330 	int *nskip;
1331 	struct got_object_id_queue *ids;
1332 	struct got_object_idset *keep;
1333 	struct got_object_idset *drop;
1334 	struct got_object_idset *skip;
1335 	got_pack_progress_cb progress_cb;
1336 	void *progress_arg;
1337 	struct got_ratelimit *rl;
1338 	got_cancel_cb cancel_cb;
1339 	void *cancel_arg;
1340 };
1341 
1342 static const struct got_error *
1343 recv_painted_commit(void *arg, struct got_object_id *id, intptr_t color)
1344 {
1345 	const struct got_error *err = NULL;
1346 	struct recv_painted_commit_arg *a = arg;
1347 	struct got_object_qid *qid, *tmp;
1348 
1349 	if (a->cancel_cb) {
1350 		err = a->cancel_cb(a->cancel_arg);
1351 		if (err)
1352 			return err;
1353 	}
1354 
1355 	switch (color) {
1356 	case COLOR_KEEP:
1357 		err = got_object_idset_add(a->keep, id, NULL);
1358 		if (err)
1359 			return err;
1360 		(*a->ncolored)++;
1361 		break;
1362 	case COLOR_DROP:
1363 		err = got_object_idset_add(a->drop, id, NULL);
1364 		if (err)
1365 			return err;
1366 		(*a->ncolored)++;
1367 		break;
1368 	case COLOR_SKIP:
1369 		err = got_object_idset_add(a->skip, id, NULL);
1370 		if (err)
1371 			return err;
1372 		break;
1373 	default:
1374 		/* should not happen */
1375 		return got_error_fmt(GOT_ERR_NOT_IMPL,
1376 		    "%s invalid commit color %"PRIdPTR, __func__, color);
1377 	}
1378 
1379 	STAILQ_FOREACH_SAFE(qid, a->ids, entry, tmp) {
1380 		if (got_object_id_cmp(&qid->id, id) != 0)
1381 			continue;
1382 		STAILQ_REMOVE(a->ids, qid, got_object_qid, entry);
1383 		color = (intptr_t)qid->data;
1384 		got_object_qid_free(qid);
1385 		(*a->nqueued)--;
1386 		if (color == COLOR_SKIP)
1387 			(*a->nskip)--;
1388 		break;
1389 	}
1390 
1391 	return report_progress(a->progress_cb, a->progress_arg, a->rl,
1392 	    *a->ncolored, 0, 0, 0L, 0, 0, 0, 0);
1393 }
1394 
1395 static const struct got_error *
1396 paint_packed_commits(struct got_pack *pack, struct got_object_id *id,
1397     int idx, intptr_t color, int *ncolored, int *nqueued, int *nskip,
1398     struct got_object_id_queue *ids,
1399     struct got_object_idset *keep, struct got_object_idset *drop,
1400     struct got_object_idset *skip, struct got_repository *repo,
1401     got_pack_progress_cb progress_cb, void *progress_arg,
1402     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1403 {
1404 	const struct got_error *err = NULL;
1405 	struct got_object_id_queue next_ids;
1406 	struct got_object_qid *qid, *tmp;
1407 	struct recv_painted_commit_arg arg;
1408 
1409 	STAILQ_INIT(&next_ids);
1410 
1411 	err = got_privsep_send_painting_request(pack->privsep_child->ibuf,
1412 	    idx, id, color);
1413 	if (err)
1414 		return err;
1415 
1416 	arg.ncolored = ncolored;
1417 	arg.nqueued = nqueued;
1418 	arg.nskip = nskip;
1419 	arg.ids = ids;
1420 	arg.keep = keep;
1421 	arg.drop = drop;
1422 	arg.skip = skip;
1423 	arg.progress_cb = progress_cb;
1424 	arg.progress_arg = progress_arg;
1425 	arg.rl = rl;
1426 	arg.cancel_cb = cancel_cb;
1427 	arg.cancel_arg = cancel_arg;
1428 	err = got_privsep_recv_painted_commits(&next_ids,
1429 	    recv_painted_commit, &arg, pack->privsep_child->ibuf);
1430 	if (err)
1431 		return err;
1432 
1433 	STAILQ_FOREACH_SAFE(qid, &next_ids, entry, tmp) {
1434 		struct got_object_qid *old_id;
1435 		intptr_t qcolor, ocolor;
1436 		STAILQ_FOREACH(old_id, ids, entry) {
1437 			if (got_object_id_cmp(&qid->id, &old_id->id))
1438 				continue;
1439 			qcolor = (intptr_t)qid->data;
1440 			ocolor = (intptr_t)old_id->data;
1441 			STAILQ_REMOVE(&next_ids, qid, got_object_qid, entry);
1442 			got_object_qid_free(qid);
1443 			qid = NULL;
1444 			if (qcolor != ocolor) {
1445 				paint_commit(old_id, qcolor);
1446 				if (ocolor == COLOR_SKIP)
1447 					(*nskip)--;
1448 				else if (qcolor == COLOR_SKIP)
1449 					(*nskip)++;
1450 			}
1451 			break;
1452 		}
1453 	}
1454 	while (!STAILQ_EMPTY(&next_ids)) {
1455 		qid = STAILQ_FIRST(&next_ids);
1456 		STAILQ_REMOVE_HEAD(&next_ids, entry);
1457 		paint_commit(qid, color);
1458 		STAILQ_INSERT_TAIL(ids, qid, entry);
1459 		(*nqueued)++;
1460 		if (color == COLOR_SKIP)
1461 			(*nskip)++;
1462 	}
1463 
1464 	return err;
1465 }
1466 
1467 static const struct got_error *
1468 find_pack_for_commit_painting(struct got_packidx **best_packidx,
1469     struct got_object_id_queue *ids, int nids, struct got_repository *repo)
1470 {
1471 	const struct got_error *err = NULL;
1472 	struct got_pathlist_entry *pe;
1473 	const char *best_packidx_path = NULL;
1474 	int nobj_max = 0;
1475 	int ncommits_max = 0;
1476 
1477 	*best_packidx = NULL;
1478 
1479 	/*
1480 	 * Find the largest pack which contains at least some of the
1481 	 * commits we are interested in.
1482 	 */
1483 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1484 		const char *path_packidx = pe->path;
1485 		struct got_packidx *packidx;
1486 		int nobj, idx, ncommits = 0;
1487 		struct got_object_qid *qid;
1488 
1489 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1490 		if (err)
1491 			break;
1492 
1493 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1494 		if (nobj <= nobj_max)
1495 			continue;
1496 
1497 		STAILQ_FOREACH(qid, ids, entry) {
1498 			idx = got_packidx_get_object_idx(packidx, &qid->id);
1499 			if (idx != -1)
1500 				ncommits++;
1501 		}
1502 		if (ncommits > ncommits_max) {
1503 			best_packidx_path = path_packidx;
1504 			nobj_max = nobj;
1505 			ncommits_max = ncommits;
1506 		}
1507 	}
1508 
1509 	if (best_packidx_path && err == NULL) {
1510 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1511 		    repo);
1512 	}
1513 
1514 	return err;
1515 }
1516 
1517 static const struct got_error *
1518 paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1519     struct got_object_idset *keep, struct got_object_idset *drop,
1520     struct got_object_idset *skip, struct got_repository *repo,
1521     got_pack_progress_cb progress_cb, void *progress_arg,
1522     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1523 {
1524 	const struct got_error *err = NULL;
1525 	struct got_commit_object *commit = NULL;
1526 	struct got_packidx *packidx = NULL;
1527 	struct got_pack *pack = NULL;
1528 	const struct got_object_id_queue *parents;
1529 	struct got_object_qid *qid = NULL;
1530 	int nqueued = nids, nskip = 0;
1531 	int idx;
1532 
1533 	while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1534 		intptr_t color;
1535 
1536 		if (cancel_cb) {
1537 			err = cancel_cb(cancel_arg);
1538 			if (err)
1539 				break;
1540 		}
1541 
1542 		qid = STAILQ_FIRST(ids);
1543 		STAILQ_REMOVE_HEAD(ids, entry);
1544 		nqueued--;
1545 		color = (intptr_t)qid->data;
1546 		if (color == COLOR_SKIP)
1547 			nskip--;
1548 
1549 		if (got_object_idset_contains(skip, &qid->id)) {
1550 			got_object_qid_free(qid);
1551 			qid = NULL;
1552 			continue;
1553 		}
1554 		if (color == COLOR_KEEP &&
1555 		    got_object_idset_contains(keep, &qid->id)) {
1556 			got_object_qid_free(qid);
1557 			qid = NULL;
1558 			continue;
1559 		}
1560 		if (color == COLOR_DROP &&
1561 		    got_object_idset_contains(drop, &qid->id)) {
1562 			got_object_qid_free(qid);
1563 			qid = NULL;
1564 			continue;
1565 		}
1566 
1567 		/* Pinned pack may have moved to different cache slot. */
1568 		pack = got_repo_get_pinned_pack(repo);
1569 
1570 		if (packidx && pack) {
1571 			idx = got_packidx_get_object_idx(packidx, &qid->id);
1572 			if (idx != -1) {
1573 				err = paint_packed_commits(pack, &qid->id,
1574 				    idx, color, ncolored, &nqueued, &nskip,
1575 				    ids, keep, drop, skip, repo,
1576 				    progress_cb, progress_arg, rl,
1577 				    cancel_cb, cancel_arg);
1578 				if (err)
1579 					break;
1580 				got_object_qid_free(qid);
1581 				qid = NULL;
1582 				continue;
1583 			}
1584 		}
1585 
1586 		switch (color) {
1587 		case COLOR_KEEP:
1588 			if (got_object_idset_contains(drop, &qid->id)) {
1589 				err = paint_commit(qid, COLOR_SKIP);
1590 				if (err)
1591 					goto done;
1592 			} else
1593 				(*ncolored)++;
1594 			err = got_object_idset_add(keep, &qid->id, NULL);
1595 			if (err)
1596 				goto done;
1597 			break;
1598 		case COLOR_DROP:
1599 			if (got_object_idset_contains(keep, &qid->id)) {
1600 				err = paint_commit(qid, COLOR_SKIP);
1601 				if (err)
1602 					goto done;
1603 			} else
1604 				(*ncolored)++;
1605 			err = got_object_idset_add(drop, &qid->id, NULL);
1606 			if (err)
1607 				goto done;
1608 			break;
1609 		case COLOR_SKIP:
1610 			if (!got_object_idset_contains(skip, &qid->id)) {
1611 				err = got_object_idset_add(skip, &qid->id,
1612 				    NULL);
1613 				if (err)
1614 					goto done;
1615 			}
1616 			break;
1617 		default:
1618 			/* should not happen */
1619 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1620 			    "%s invalid commit color %"PRIdPTR, __func__,
1621 			    color);
1622 			goto done;
1623 		}
1624 
1625 		err = report_progress(progress_cb, progress_arg, rl,
1626 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1627 		if (err)
1628 			break;
1629 
1630 		err = got_object_open_as_commit(&commit, repo, &qid->id);
1631 		if (err)
1632 			break;
1633 
1634 		parents = got_object_commit_get_parent_ids(commit);
1635 		if (parents) {
1636 			struct got_object_qid *pid;
1637 			color = (intptr_t)qid->data;
1638 			STAILQ_FOREACH(pid, parents, entry) {
1639 				err = queue_commit_id(ids, &pid->id,
1640 				    color, repo);
1641 				if (err)
1642 					break;
1643 				nqueued++;
1644 				if (color == COLOR_SKIP)
1645 					nskip++;
1646 			}
1647 		}
1648 
1649 		if (pack == NULL && (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
1650 			if (packidx == NULL) {
1651 				err = find_pack_for_commit_painting(&packidx,
1652 				    ids, nqueued, repo);
1653 				if (err)
1654 					goto done;
1655 			}
1656 			if (packidx != NULL) {
1657 				err = cache_pack_for_packidx(&pack, packidx,
1658 				    repo);
1659 				if (err)
1660 					goto done;
1661 				err = got_privsep_init_commit_painting(
1662 				    pack->privsep_child->ibuf);
1663 				if (err)
1664 					goto done;
1665 				err = send_idset(pack->privsep_child->ibuf,
1666 				    keep);
1667 				if (err)
1668 					goto done;
1669 				err = send_idset(pack->privsep_child->ibuf, drop);
1670 				if (err)
1671 					goto done;
1672 				err = send_idset(pack->privsep_child->ibuf, skip);
1673 				if (err)
1674 					goto done;
1675 				err = got_repo_pin_pack(repo, packidx, pack);
1676 				if (err)
1677 					goto done;
1678 			}
1679 		}
1680 
1681 		got_object_commit_close(commit);
1682 		commit = NULL;
1683 
1684 		got_object_qid_free(qid);
1685 		qid = NULL;
1686 	}
1687 done:
1688 	if (pack) {
1689 		const struct got_error *pack_err;
1690 		pack_err = got_privsep_send_painting_commits_done(
1691 		    pack->privsep_child->ibuf);
1692 		if (err == NULL)
1693 			err = pack_err;
1694 	}
1695 	if (commit)
1696 		got_object_commit_close(commit);
1697 	got_object_qid_free(qid);
1698 	got_repo_unpin_pack(repo);
1699 	return err;
1700 }
1701 
1702 static const struct got_error *
1703 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1704     struct got_object_id **head, int nhead,
1705     struct got_object_id **tail, int ntail,
1706     struct got_repository *repo,
1707     got_pack_progress_cb progress_cb, void *progress_arg,
1708     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1709 {
1710 	const struct got_error *err = NULL;
1711 	struct got_object_id_queue ids;
1712 	struct got_object_idset *keep, *drop, *skip = NULL;
1713 	int i, nkeep;
1714 
1715 	STAILQ_INIT(&ids);
1716 	*res = NULL;
1717 	*nres = 0;
1718 	*ncolored = 0;
1719 
1720 	keep = got_object_idset_alloc();
1721 	if (keep == NULL)
1722 		return got_error_from_errno("got_object_idset_alloc");
1723 
1724 	drop = got_object_idset_alloc();
1725 	if (drop == NULL) {
1726 		err = got_error_from_errno("got_object_idset_alloc");
1727 		goto done;
1728 	}
1729 
1730 	skip = got_object_idset_alloc();
1731 	if (skip == NULL) {
1732 		err = got_error_from_errno("got_object_idset_alloc");
1733 		goto done;
1734 	}
1735 
1736 	for (i = 0; i < nhead; i++) {
1737 		struct got_object_id *id = head[i];
1738 		if (id == NULL)
1739 			continue;
1740 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1741 		if (err)
1742 			goto done;
1743 	}
1744 
1745 	for (i = 0; i < ntail; i++) {
1746 		struct got_object_id *id = tail[i];
1747 		if (id == NULL)
1748 			continue;
1749 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1750 		if (err)
1751 			goto done;
1752 	}
1753 
1754 	err = paint_commits(ncolored, &ids, nhead + ntail,
1755 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1756 	    cancel_cb, cancel_arg);
1757 	if (err)
1758 		goto done;
1759 
1760 	nkeep = got_object_idset_num_elements(keep);
1761 	if (nkeep > 0) {
1762 		struct append_id_arg arg;
1763 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1764 		if (arg.array == NULL) {
1765 			err = got_error_from_errno("calloc");
1766 			goto done;
1767 		}
1768 		arg.idx = -1;
1769 		arg.skip = skip;
1770 		arg.drop = drop;
1771 		err = got_object_idset_for_each(keep, append_id, &arg);
1772 		if (err) {
1773 			free(arg.array);
1774 			goto done;
1775 		}
1776 		*res = arg.array;
1777 		*nres = arg.idx + 1;
1778 	}
1779 done:
1780 	got_object_idset_free(keep);
1781 	got_object_idset_free(drop);
1782 	if (skip)
1783 		got_object_idset_free(skip);
1784 	got_object_id_queue_free(&ids);
1785 	return err;
1786 }
1787 
1788 struct load_packed_obj_arg {
1789 	/* output parameters: */
1790 	struct got_object_id *id;
1791 	char *dpath;
1792 	time_t mtime;
1793 
1794 	/* input parameters: */
1795 	uint32_t seed;
1796 	int want_meta;
1797 	struct got_object_idset *idset;
1798 	struct got_object_idset *idset_exclude;
1799 	int loose_obj_only;
1800 	int *ncolored;
1801 	int *nfound;
1802 	int *ntrees;
1803 	got_pack_progress_cb progress_cb;
1804 	void *progress_arg;
1805 	struct got_ratelimit *rl;
1806 	got_cancel_cb cancel_cb;
1807 	void *cancel_arg;
1808 };
1809 
1810 static const struct got_error *
1811 load_packed_commit_id(void *arg, time_t mtime, struct got_object_id *id,
1812     struct got_repository *repo)
1813 {
1814 	struct load_packed_obj_arg *a = arg;
1815 
1816 	if (got_object_idset_contains(a->idset, id) ||
1817 	    got_object_idset_contains(a->idset_exclude, id))
1818 		return NULL;
1819 
1820 	return add_object(a->want_meta,
1821 	    a->want_meta ? a->idset : a->idset_exclude,
1822 	    id, "", GOT_OBJ_TYPE_COMMIT, mtime, a->seed, a->loose_obj_only,
1823 	    repo, a->ncolored, a->nfound, a->ntrees,
1824 	    a->progress_cb, a->progress_arg, a->rl);
1825 }
1826 
1827 static const struct got_error *
1828 load_packed_tree_ids(void *arg, struct got_tree_object *tree, time_t mtime,
1829     struct got_object_id *id, const char *dpath, struct got_repository *repo)
1830 {
1831 	const struct got_error *err;
1832 	struct load_packed_obj_arg *a = arg;
1833 	const char *relpath;
1834 
1835 	/*
1836 	 * When we receive a tree's ID and path but not the tree itself,
1837 	 * this tree object was not found in the pack file. This is the
1838 	 * last time we are being called for this optimized traversal.
1839 	 * Return from here and switch to loading objects the slow way.
1840 	 */
1841 	if (tree == NULL) {
1842 		free(a->id);
1843 		a->id = got_object_id_dup(id);
1844 		if (a->id == NULL) {
1845 			err = got_error_from_errno("got_object_id_dup");
1846 			free(a->dpath);
1847 			a->dpath = NULL;
1848 			return err;
1849 		}
1850 
1851 		free(a->dpath);
1852 		a->dpath = strdup(dpath);
1853 		if (a->dpath == NULL) {
1854 			err = got_error_from_errno("strdup");
1855 			free(a->id);
1856 			a->id = NULL;
1857 			return err;
1858 		}
1859 
1860 		a->mtime = mtime;
1861 		return NULL;
1862 	}
1863 
1864 	if (got_object_idset_contains(a->idset, id) ||
1865 	    got_object_idset_contains(a->idset_exclude, id))
1866 		return NULL;
1867 
1868 	relpath = dpath;
1869 	while (relpath[0] == '/')
1870 		relpath++;
1871 
1872 	err = add_object(a->want_meta,
1873 	    a->want_meta ? a->idset : a->idset_exclude,
1874 	    id, relpath, GOT_OBJ_TYPE_TREE, mtime, a->seed,
1875 	    a->loose_obj_only, repo, a->ncolored, a->nfound, a->ntrees,
1876 	    a->progress_cb, a->progress_arg, a->rl);
1877 	if (err)
1878 		return err;
1879 
1880 	return load_tree_entries(NULL, a->want_meta, a->idset,
1881 	    a->idset_exclude, tree, dpath, mtime, a->seed, repo,
1882 	    a->loose_obj_only, a->ncolored, a->nfound, a->ntrees,
1883 	    a->progress_cb, a->progress_arg, a->rl,
1884 	    a->cancel_cb, a->cancel_arg);
1885 }
1886 
1887 static const struct got_error *
1888 load_packed_object_ids(int *found_all_objects,
1889     struct got_object_id **ours, int nours,
1890     struct got_object_id **theirs, int ntheirs,
1891     int want_meta, uint32_t seed, struct got_object_idset *idset,
1892     struct got_object_idset *idset_exclude, int loose_obj_only,
1893     struct got_repository *repo, struct got_packidx *packidx,
1894     int *ncolored, int *nfound, int *ntrees,
1895     got_pack_progress_cb progress_cb, void *progress_arg,
1896     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1897 {
1898 	const struct got_error *err = NULL;
1899 	struct load_packed_obj_arg lpa;
1900 
1901 	memset(&lpa, 0, sizeof(lpa));
1902 	lpa.seed = seed;
1903 	lpa.want_meta = want_meta;
1904 	lpa.idset = idset;
1905 	lpa.idset_exclude = idset_exclude;
1906 	lpa.loose_obj_only = loose_obj_only;
1907 	lpa.ncolored = ncolored;
1908 	lpa.nfound = nfound;
1909 	lpa.ntrees = ntrees;
1910 	lpa.progress_cb = progress_cb;
1911 	lpa.progress_arg = progress_arg;
1912 	lpa.rl = rl;
1913 	lpa.cancel_cb = cancel_cb;
1914 	lpa.cancel_arg = cancel_arg;
1915 
1916 	/* Attempt to load objects via got-read-pack, as far as possible. */
1917 	err = got_object_enumerate(found_all_objects, load_packed_commit_id,
1918 	   load_packed_tree_ids, &lpa, ours, nours, theirs, ntheirs,
1919 	   packidx, repo);
1920 	if (err)
1921 		return err;
1922 
1923 	if (lpa.id == NULL)
1924 		return NULL;
1925 
1926 	/*
1927 	 * An incomplete tree hierarchy was present in the pack file
1928 	 * and caused loading to be aborted.
1929 	 * Continue loading trees the slow way.
1930 	 */
1931 	err = load_tree(want_meta, idset, idset_exclude,
1932 	    lpa.id, lpa.dpath, lpa.mtime, seed, repo, loose_obj_only,
1933 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1934 	    cancel_cb, cancel_arg);
1935 	free(lpa.id);
1936 	free(lpa.dpath);
1937 	return err;
1938 }
1939 
1940 static const struct got_error *
1941 find_pack_for_enumeration(struct got_packidx **best_packidx,
1942     struct got_object_id **ids, int nids, struct got_repository *repo)
1943 {
1944 	const struct got_error *err = NULL;
1945 	struct got_pathlist_entry *pe;
1946 	const char *best_packidx_path = NULL;
1947 	int nobj_max = 0;
1948 	int ncommits_max = 0;
1949 
1950 	*best_packidx = NULL;
1951 
1952 	/*
1953 	 * Find the largest pack which contains at least some of the
1954 	 * commits and tags we are interested in.
1955 	 */
1956 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1957 		const char *path_packidx = pe->path;
1958 		struct got_packidx *packidx;
1959 		int nobj, i, idx, ncommits = 0;
1960 
1961 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1962 		if (err)
1963 			break;
1964 
1965 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1966 		if (nobj <= nobj_max)
1967 			continue;
1968 
1969 		for (i = 0; i < nids; i++) {
1970 			idx = got_packidx_get_object_idx(packidx, ids[i]);
1971 			if (idx != -1)
1972 				ncommits++;
1973 		}
1974 		if (ncommits > ncommits_max) {
1975 			best_packidx_path = path_packidx;
1976 			nobj_max = nobj;
1977 			ncommits_max = ncommits;
1978 		}
1979 	}
1980 
1981 	if (best_packidx_path && err == NULL) {
1982 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1983 		    repo);
1984 	}
1985 
1986 	return err;
1987 }
1988 
1989 static const struct got_error *
1990 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1991     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1992     struct got_object_id **ours, int nours, struct got_repository *repo,
1993     uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1994     void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1995     void *cancel_arg)
1996 {
1997 	const struct got_error *err = NULL;
1998 	struct got_object_id **ids = NULL;
1999 	struct got_packidx *packidx = NULL;
2000 	int i, nobj = 0, obj_type, found_all_objects = 0;
2001 	struct got_object_idset *idset_exclude;
2002 
2003 	idset_exclude = got_object_idset_alloc();
2004 	if (idset_exclude == NULL)
2005 		return got_error_from_errno("got_object_idset_alloc");
2006 
2007 	*ncolored = 0;
2008 	*nfound = 0;
2009 	*ntrees = 0;
2010 
2011 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
2012 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2013 	if (err)
2014 		goto done;
2015 
2016 	err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
2017 	if (err)
2018 		goto done;
2019 	if (packidx) {
2020 		err = load_packed_object_ids(&found_all_objects,
2021 		    theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
2022 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
2023 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2024 		if (err)
2025 			goto done;
2026 	}
2027 
2028 	for (i = 0; i < ntheirs; i++) {
2029 		struct got_object_id *id = theirs[i];
2030 		if (id == NULL)
2031 			continue;
2032 		err = got_object_get_type(&obj_type, repo, id);
2033 		if (err)
2034 			return err;
2035 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
2036 			if (!found_all_objects) {
2037 				err = load_commit(0, idset, idset_exclude,
2038 				    id, repo, seed, loose_obj_only,
2039 				    ncolored, nfound, ntrees,
2040 				    progress_cb, progress_arg, rl,
2041 				    cancel_cb, cancel_arg);
2042 				if (err)
2043 					goto done;
2044 			}
2045 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
2046 			err = load_tag(0, idset, idset_exclude, id, repo,
2047 			    seed, loose_obj_only, ncolored, nfound, ntrees,
2048 			    progress_cb, progress_arg, rl,
2049 			    cancel_cb, cancel_arg);
2050 			if (err)
2051 				goto done;
2052 		}
2053 	}
2054 
2055 	found_all_objects = 0;
2056 	err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
2057 	if (err)
2058 		goto done;
2059 	if (packidx) {
2060 		err = load_packed_object_ids(&found_all_objects, ids,
2061 		    nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
2062 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
2063 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2064 		if (err)
2065 			goto done;
2066 	}
2067 
2068 	if (!found_all_objects) {
2069 		for (i = 0; i < nobj; i++) {
2070 			err = load_commit(1, idset, idset_exclude, ids[i],
2071 			    repo, seed, loose_obj_only, ncolored, nfound,
2072 			    ntrees, progress_cb, progress_arg, rl,
2073 			    cancel_cb, cancel_arg);
2074 			if (err)
2075 				goto done;
2076 		}
2077 	}
2078 
2079 	for (i = 0; i < nours; i++) {
2080 		struct got_object_id *id = ours[i];
2081 		struct got_pack_meta *m;
2082 		if (id == NULL)
2083 			continue;
2084 		m = got_object_idset_get(idset, id);
2085 		if (m == NULL) {
2086 			err = got_object_get_type(&obj_type, repo, id);
2087 			if (err)
2088 				goto done;
2089 		} else
2090 			obj_type = m->obj_type;
2091 		if (obj_type != GOT_OBJ_TYPE_TAG)
2092 			continue;
2093 		err = load_tag(1, idset, idset_exclude, id, repo,
2094 		    seed, loose_obj_only, ncolored, nfound, ntrees,
2095 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
2096 		if (err)
2097 			goto done;
2098 	}
2099 done:
2100 	for (i = 0; i < nobj; i++) {
2101 		free(ids[i]);
2102 	}
2103 	free(ids);
2104 	got_object_idset_free(idset_exclude);
2105 	return err;
2106 }
2107 
2108 static const struct got_error *
2109 hwrite(FILE *f, const void *buf, off_t len, SHA1_CTX *ctx)
2110 {
2111 	size_t n;
2112 
2113 	SHA1Update(ctx, buf, len);
2114 	n = fwrite(buf, 1, len, f);
2115 	if (n != len)
2116 		return got_ferror(f, GOT_ERR_IO);
2117 	return NULL;
2118 }
2119 
2120 static const struct got_error *
2121 hcopy(FILE *fsrc, FILE *fdst, off_t len, SHA1_CTX *ctx)
2122 {
2123 	unsigned char buf[65536];
2124 	off_t remain = len;
2125 	size_t n;
2126 
2127 	while (remain > 0) {
2128 		size_t copylen = MIN(sizeof(buf), remain);
2129 		n = fread(buf, 1, copylen, fsrc);
2130 		if (n != copylen)
2131 			return got_ferror(fsrc, GOT_ERR_IO);
2132 		SHA1Update(ctx, buf, copylen);
2133 		n = fwrite(buf, 1, copylen, fdst);
2134 		if (n != copylen)
2135 			return got_ferror(fdst, GOT_ERR_IO);
2136 		remain -= copylen;
2137 	}
2138 
2139 	return NULL;
2140 }
2141 
2142 static const struct got_error *
2143 hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
2144     FILE *fdst, off_t len, SHA1_CTX *ctx)
2145 {
2146 	size_t n;
2147 
2148 	if (src_offset + len > src_size)
2149 		return got_error(GOT_ERR_RANGE);
2150 
2151 	SHA1Update(ctx, src + src_offset, len);
2152 	n = fwrite(src + src_offset, 1, len, fdst);
2153 	if (n != len)
2154 		return got_ferror(fdst, GOT_ERR_IO);
2155 
2156 	return NULL;
2157 }
2158 
2159 static void
2160 putbe32(char *b, uint32_t n)
2161 {
2162 	b[0] = n >> 24;
2163 	b[1] = n >> 16;
2164 	b[2] = n >> 8;
2165 	b[3] = n >> 0;
2166 }
2167 
2168 static int
2169 write_order_cmp(const void *pa, const void *pb)
2170 {
2171 	struct got_pack_meta *a, *b, *ahd, *bhd;
2172 
2173 	a = *(struct got_pack_meta **)pa;
2174 	b = *(struct got_pack_meta **)pb;
2175 	ahd = (a->head == NULL) ? a : a->head;
2176 	bhd = (b->head == NULL) ? b : b->head;
2177 	if (bhd->mtime < ahd->mtime)
2178 		return -1;
2179 	if (bhd->mtime > ahd->mtime)
2180 		return 1;
2181 	if (bhd < ahd)
2182 		return -1;
2183 	if (bhd > ahd)
2184 		return 1;
2185 	if (a->nchain != b->nchain)
2186 		return a->nchain - b->nchain;
2187 	if (a->mtime < b->mtime)
2188 		return -1;
2189 	if (a->mtime > b->mtime)
2190 		return 1;
2191 	return got_object_id_cmp(&a->id, &b->id);
2192 }
2193 
2194 static int
2195 reuse_write_order_cmp(const void *pa, const void *pb)
2196 {
2197 	struct got_pack_meta *a, *b;
2198 
2199 	a = *(struct got_pack_meta **)pa;
2200 	b = *(struct got_pack_meta **)pb;
2201 
2202 	if (a->reused_delta_offset < b->reused_delta_offset)
2203 		return -1;
2204 	if (a->reused_delta_offset > b->reused_delta_offset)
2205 		return 1;
2206 	return 0;
2207 }
2208 
2209 static const struct got_error *
2210 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
2211 {
2212 	size_t i;
2213 
2214 	*hdrlen = 0;
2215 
2216 	hdr[0] = obj_type << 4;
2217 	hdr[0] |= len & 0xf;
2218 	len >>= 4;
2219 	for (i = 1; len != 0; i++){
2220 		if (i >= bufsize)
2221 			return got_error(GOT_ERR_NO_SPACE);
2222 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
2223 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
2224 		len >>= GOT_DELTA_SIZE_SHIFT;
2225 	}
2226 
2227 	*hdrlen = i;
2228 	return NULL;
2229 }
2230 
2231 static int
2232 packoff(char *hdr, off_t off)
2233 {
2234 	int i, j;
2235 	char rbuf[8];
2236 
2237 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
2238 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
2239 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
2240 		    GOT_DELTA_SIZE_MORE;
2241 	}
2242 
2243 	j = 0;
2244 	while (i > 0)
2245 		hdr[j++] = rbuf[--i];
2246 	return j;
2247 }
2248 
2249 static const struct got_error *
2250 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
2251     struct got_pack_meta *m)
2252 {
2253 	const struct got_error *err;
2254 	char buf[32];
2255 	int nh;
2256 
2257 	if (m->prev->off != 0) {
2258 		err = packhdr(&nh, buf, sizeof(buf),
2259 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
2260 		if (err)
2261 			return err;
2262 		nh += packoff(buf + nh, m->off - m->prev->off);
2263 		err = hwrite(packfile, buf, nh, ctx);
2264 		if (err)
2265 			return err;
2266 		*packfile_size += nh;
2267 	} else {
2268 		err = packhdr(&nh, buf, sizeof(buf),
2269 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
2270 		if (err)
2271 			return err;
2272 		err = hwrite(packfile, buf, nh, ctx);
2273 		if (err)
2274 			return err;
2275 		*packfile_size += nh;
2276 		err = hwrite(packfile, m->prev->id.sha1,
2277 		    sizeof(m->prev->id.sha1), ctx);
2278 		if (err)
2279 			return err;
2280 		*packfile_size += sizeof(m->prev->id.sha1);
2281 	}
2282 
2283 	return NULL;
2284 }
2285 
2286 static const struct got_error *
2287 write_packed_object(off_t *packfile_size, FILE *packfile,
2288     FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
2289     struct got_pack_meta *m, int *outfd, SHA1_CTX *ctx,
2290     struct got_repository *repo)
2291 {
2292 	const struct got_error *err = NULL;
2293 	struct got_deflate_checksum csum;
2294 	char buf[32];
2295 	int nh;
2296 	struct got_raw_object *raw = NULL;
2297 	off_t outlen;
2298 
2299 	csum.output_sha1 = ctx;
2300 	csum.output_crc = NULL;
2301 
2302 	m->off = ftello(packfile);
2303 	if (m->delta_len == 0) {
2304 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
2305 		if (err)
2306 			goto done;
2307 		err = packhdr(&nh, buf, sizeof(buf),
2308 		    m->obj_type, raw->size);
2309 		if (err)
2310 			goto done;
2311 		err = hwrite(packfile, buf, nh, ctx);
2312 		if (err)
2313 			goto done;
2314 		*packfile_size += nh;
2315 		if (raw->f == NULL) {
2316 			err = got_deflate_to_file_mmap(&outlen,
2317 			    raw->data + raw->hdrlen, 0, raw->size,
2318 			    packfile, &csum);
2319 			if (err)
2320 				goto done;
2321 		} else {
2322 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
2323 			    == -1) {
2324 				err = got_error_from_errno("fseeko");
2325 				goto done;
2326 			}
2327 			err = got_deflate_to_file(&outlen, raw->f,
2328 			    raw->size, packfile, &csum);
2329 			if (err)
2330 				goto done;
2331 		}
2332 		*packfile_size += outlen;
2333 		got_object_raw_close(raw);
2334 		raw = NULL;
2335 	} else if (m->delta_buf) {
2336 		err = deltahdr(packfile_size, ctx, packfile, m);
2337 		if (err)
2338 			goto done;
2339 		err = hwrite(packfile, m->delta_buf,
2340 		    m->delta_compressed_len, ctx);
2341 		if (err)
2342 			goto done;
2343 		*packfile_size += m->delta_compressed_len;
2344 		free(m->delta_buf);
2345 		m->delta_buf = NULL;
2346 	} else if (delta_cache_map) {
2347 		err = deltahdr(packfile_size, ctx, packfile, m);
2348 		if (err)
2349 			goto done;
2350 		err = hcopy_mmap(delta_cache_map, m->delta_offset,
2351 		    delta_cache_size, packfile, m->delta_compressed_len,
2352 		    ctx);
2353 		if (err)
2354 			goto done;
2355 		*packfile_size += m->delta_compressed_len;
2356 	} else {
2357 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
2358 		    == -1) {
2359 			err = got_error_from_errno("fseeko");
2360 			goto done;
2361 		}
2362 		err = deltahdr(packfile_size, ctx, packfile, m);
2363 		if (err)
2364 			goto done;
2365 		err = hcopy(delta_cache, packfile,
2366 		    m->delta_compressed_len, ctx);
2367 		if (err)
2368 			goto done;
2369 		*packfile_size += m->delta_compressed_len;
2370 	}
2371 done:
2372 	if (raw)
2373 		got_object_raw_close(raw);
2374 	return err;
2375 }
2376 
2377 static const struct got_error *
2378 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
2379     struct got_pack_meta **deltify, int ndeltify,
2380     struct got_pack_meta **reuse, int nreuse,
2381     int ncolored, int nfound, int ntrees, int nours,
2382     struct got_repository *repo,
2383     got_pack_progress_cb progress_cb, void *progress_arg,
2384     struct got_ratelimit *rl,
2385     got_cancel_cb cancel_cb, void *cancel_arg)
2386 {
2387 	const struct got_error *err = NULL;
2388 	int i;
2389 	SHA1_CTX ctx;
2390 	struct got_pack_meta *m;
2391 	char buf[32];
2392 	size_t n;
2393 	off_t packfile_size = 0;
2394 	int outfd = -1;
2395 	int delta_cache_fd = -1;
2396 	uint8_t *delta_cache_map = NULL;
2397 	size_t delta_cache_size = 0;
2398 
2399 	SHA1Init(&ctx);
2400 
2401 #ifndef GOT_PACK_NO_MMAP
2402 	delta_cache_fd = dup(fileno(delta_cache));
2403 	if (delta_cache_fd != -1) {
2404 		struct stat sb;
2405 		if (fstat(delta_cache_fd, &sb) == -1) {
2406 			err = got_error_from_errno("fstat");
2407 			goto done;
2408 		}
2409 		if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
2410 			delta_cache_map = mmap(NULL, sb.st_size,
2411 			    PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
2412 			if (delta_cache_map == MAP_FAILED) {
2413 				if (errno != ENOMEM) {
2414 					err = got_error_from_errno("mmap");
2415 					goto done;
2416 				}
2417 				delta_cache_map = NULL; /* fallback on stdio */
2418 			} else
2419 				delta_cache_size = (size_t)sb.st_size;
2420 		}
2421 	}
2422 #endif
2423 	err = hwrite(packfile, "PACK", 4, &ctx);
2424 	if (err)
2425 		goto done;
2426 	putbe32(buf, GOT_PACKFILE_VERSION);
2427 	err = hwrite(packfile, buf, 4, &ctx);
2428 	if (err)
2429 		goto done;
2430 	putbe32(buf, ndeltify + nreuse);
2431 	err = hwrite(packfile, buf, 4, &ctx);
2432 	if (err)
2433 		goto done;
2434 
2435 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
2436 	    write_order_cmp);
2437 	for (i = 0; i < ndeltify; i++) {
2438 		err = report_progress(progress_cb, progress_arg, rl,
2439 		    ncolored, nfound, ntrees, packfile_size, nours,
2440 		    ndeltify + nreuse, ndeltify + nreuse, i);
2441 		if (err)
2442 			goto done;
2443 		m = deltify[i];
2444 		err = write_packed_object(&packfile_size, packfile,
2445 		    delta_cache, delta_cache_map, delta_cache_size,
2446 		    m, &outfd, &ctx, repo);
2447 		if (err)
2448 			goto done;
2449 	}
2450 
2451 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
2452 	    reuse_write_order_cmp);
2453 	for (i = 0; i < nreuse; i++) {
2454 		err = report_progress(progress_cb, progress_arg, rl,
2455 		    ncolored, nfound, ntrees, packfile_size, nours,
2456 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
2457 		if (err)
2458 			goto done;
2459 		m = reuse[i];
2460 		err = write_packed_object(&packfile_size, packfile,
2461 		    delta_cache, delta_cache_map, delta_cache_size,
2462 		    m, &outfd, &ctx, repo);
2463 		if (err)
2464 			goto done;
2465 	}
2466 
2467 	SHA1Final(pack_sha1, &ctx);
2468 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
2469 	if (n != SHA1_DIGEST_LENGTH)
2470 		err = got_ferror(packfile, GOT_ERR_IO);
2471 	packfile_size += SHA1_DIGEST_LENGTH;
2472 	packfile_size += sizeof(struct got_packfile_hdr);
2473 	if (progress_cb) {
2474 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2475 		    packfile_size, nours, ndeltify + nreuse,
2476 		    ndeltify + nreuse, ndeltify + nreuse);
2477 		if (err)
2478 			goto done;
2479 	}
2480 done:
2481 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
2482 		err = got_error_from_errno("close");
2483 	if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
2484 		err = got_error_from_errno("munmap");
2485 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2486 		err = got_error_from_errno("close");
2487 	return err;
2488 }
2489 
2490 static const struct got_error *
2491 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
2492 {
2493 	struct got_pack_meta *m = data;
2494 	struct got_pack_metavec *v = arg;
2495 
2496 	if (m->reused_delta_offset != 0)
2497 		return NULL;
2498 
2499 	return add_meta(m, v);
2500 }
2501 
2502 const struct got_error *
2503 got_pack_create(uint8_t *packsha1, FILE *packfile,
2504     struct got_object_id **theirs, int ntheirs,
2505     struct got_object_id **ours, int nours,
2506     struct got_repository *repo, int loose_obj_only, int allow_empty,
2507     got_pack_progress_cb progress_cb, void *progress_arg,
2508     got_cancel_cb cancel_cb, void *cancel_arg)
2509 {
2510 	const struct got_error *err;
2511 	int delta_cache_fd = -1;
2512 	FILE *delta_cache = NULL;
2513 	struct got_object_idset *idset;
2514 	struct got_ratelimit rl;
2515 	struct got_pack_metavec deltify, reuse;
2516 	int ncolored = 0, nfound = 0, ntrees = 0;
2517 	size_t ndeltify;
2518 	uint32_t seed;
2519 
2520 	seed = arc4random();
2521 
2522 	memset(&deltify, 0, sizeof(deltify));
2523 	memset(&reuse, 0, sizeof(reuse));
2524 
2525 	got_ratelimit_init(&rl, 0, 500);
2526 
2527 	idset = got_object_idset_alloc();
2528 	if (idset == NULL)
2529 		return got_error_from_errno("got_object_idset_alloc");
2530 
2531 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
2532 	    ntheirs, ours, nours, repo, seed, loose_obj_only,
2533 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
2534 	if (err)
2535 		goto done;
2536 
2537 	if (progress_cb) {
2538 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2539 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
2540 		if (err)
2541 			goto done;
2542 	}
2543 
2544 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
2545 		err = got_error(GOT_ERR_CANNOT_PACK);
2546 		goto done;
2547 	}
2548 
2549 	delta_cache_fd = got_opentempfd();
2550 	if (delta_cache_fd == -1) {
2551 		err = got_error_from_errno("got_opentemp");
2552 		goto done;
2553 	}
2554 
2555 	reuse.metasz = 64;
2556 	reuse.meta = calloc(reuse.metasz,
2557 	    sizeof(struct got_pack_meta *));
2558 	if (reuse.meta == NULL) {
2559 		err = got_error_from_errno("calloc");
2560 		goto done;
2561 	}
2562 
2563 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
2564 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
2565 	    cancel_cb, cancel_arg);
2566 	if (err)
2567 		goto done;
2568 
2569 	delta_cache = fdopen(delta_cache_fd, "a+");
2570 	if (delta_cache == NULL) {
2571 		err = got_error_from_errno("fdopen");
2572 		goto done;
2573 	}
2574 	delta_cache_fd = -1;
2575 
2576 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
2577 		err = got_error_from_errno("fseeko");
2578 		goto done;
2579 	}
2580 
2581 	ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
2582 	if (ndeltify > 0) {
2583 		deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
2584 		if (deltify.meta == NULL) {
2585 			err = got_error_from_errno("calloc");
2586 			goto done;
2587 		}
2588 		deltify.metasz = ndeltify;
2589 
2590 		err = got_object_idset_for_each(idset, add_meta_idset_cb,
2591 		    &deltify);
2592 		if (err)
2593 			goto done;
2594 		if (deltify.nmeta > 0) {
2595 			err = pick_deltas(deltify.meta, deltify.nmeta,
2596 			    ncolored, nfound, ntrees, nours, reuse.nmeta,
2597 			    delta_cache, repo, progress_cb, progress_arg, &rl,
2598 			    cancel_cb, cancel_arg);
2599 			if (err)
2600 				goto done;
2601 		}
2602 	}
2603 
2604 	if (fflush(delta_cache) == EOF) {
2605 		err = got_error_from_errno("fflush");
2606 		goto done;
2607 	}
2608 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
2609 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
2610 	    nours, repo, progress_cb, progress_arg, &rl,
2611 	    cancel_cb, cancel_arg);
2612 	if (err)
2613 		goto done;
2614 done:
2615 	free_nmeta(deltify.meta, deltify.nmeta);
2616 	free_nmeta(reuse.meta, reuse.nmeta);
2617 	got_object_idset_free(idset);
2618 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2619 		err = got_error_from_errno("close");
2620 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
2621 		err = got_error_from_errno("fclose");
2622 	return err;
2623 }