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/uio.h>
20 #include <sys/stat.h>
21 #include <sys/time.h>
22 #include <sys/mman.h>
23 
24 #include <errno.h>
25 #include <stdint.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <time.h>
30 #include <unistd.h>
31 #include <limits.h>
32 #include <zlib.h>
33 
34 #if defined(__FreeBSD__)
35 #include <unistd.h>
36 #endif
37 
38 #include "got_error.h"
39 #include "got_cancel.h"
40 #include "got_object.h"
41 #include "got_path.h"
42 #include "got_reference.h"
43 #include "got_repository_admin.h"
44 #include "got_opentemp.h"
45 
46 #include "got_lib_deltify.h"
47 #include "got_lib_delta.h"
48 #include "got_lib_object.h"
49 #include "got_lib_object_idset.h"
50 #include "got_lib_object_cache.h"
51 #include "got_lib_deflate.h"
52 #include "got_lib_pack.h"
53 #include "got_lib_pack_create.h"
54 #include "got_lib_privsep.h"
55 #include "got_lib_repository.h"
56 #include "got_lib_ratelimit.h"
57 #include "got_lib_inflate.h"
58 
59 #include "murmurhash2.h"
60 
61 #ifndef MIN
62 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
63 #endif
64 
65 #ifndef MAX
66 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
67 #endif
68 
69 #ifndef nitems
70 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
71 #endif
72 
73 struct got_pack_meta {
74 	struct got_object_id id;
75 	uint32_t path_hash;
76 	int	obj_type;
77 	off_t	size;
78 	time_t	mtime;
79 
80 	/* The best delta we picked */
81 	struct got_pack_meta *head;
82 	struct got_pack_meta *prev;
83 	unsigned char *delta_buf; /* if encoded in memory (compressed) */
84 	off_t	delta_offset;	/* offset in delta cache file (compressed) */
85 	off_t	delta_len;	/* encoded delta length */
86 	off_t	delta_compressed_len; /* encoded+compressed delta length */
87 	int	nchain;
88 
89 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
90 	struct got_object_id *base_obj_id;
91 
92 	/* Only used for delta window */
93 	struct got_delta_table *dtab;
94 
95 	/* Only used for writing offset deltas */
96 	off_t	off;
97 };
98 
99 struct got_pack_metavec {
100 	struct got_pack_meta **meta;
101 	int nmeta;
102 	int metasz;
103 };
104 
105 static const struct got_error *
106 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
107     const char *path, int obj_type, time_t mtime, uint32_t seed)
108 {
109 	struct got_pack_meta *m;
110 
111 	*new = NULL;
112 
113 	m = calloc(1, sizeof(*m));
114 	if (m == NULL)
115 		return got_error_from_errno("calloc");
116 
117 	memcpy(&m->id, id, sizeof(m->id));
118 
119 	m->path_hash = murmurhash2(path, strlen(path), seed);
120 	m->obj_type = obj_type;
121 	m->mtime = mtime;
122 	*new = m;
123 	return NULL;
124 }
125 
126 static void
127 clear_meta(struct got_pack_meta *meta)
128 {
129 	if (meta == NULL)
130 		return;
131 	meta->path_hash = 0;
132 	free(meta->delta_buf);
133 	meta->delta_buf = NULL;
134 	free(meta->base_obj_id);
135 	meta->base_obj_id = NULL;
136 	meta->reused_delta_offset = 0;
137 }
138 
139 static void
140 free_nmeta(struct got_pack_meta **meta, int nmeta)
141 {
142 	int i;
143 
144 	for (i = 0; i < nmeta; i++)
145 		clear_meta(meta[i]);
146 	free(meta);
147 }
148 
149 static int
150 delta_order_cmp(const void *pa, const void *pb)
151 {
152 	struct got_pack_meta *a, *b;
153 
154 	a = *(struct got_pack_meta **)pa;
155 	b = *(struct got_pack_meta **)pb;
156 
157 	if (a->obj_type != b->obj_type)
158 		return a->obj_type - b->obj_type;
159 	if (a->path_hash < b->path_hash)
160 		return -1;
161 	if (a->path_hash > b->path_hash)
162 		return 1;
163 	if (a->mtime < b->mtime)
164 		return -1;
165 	if (a->mtime > b->mtime)
166 		return 1;
167 	return got_object_id_cmp(&a->id, &b->id);
168 }
169 
170 static off_t
171 delta_size(struct got_delta_instruction *deltas, int ndeltas)
172 {
173 	int i;
174 	off_t size = 32;
175 	for (i = 0; i < ndeltas; i++) {
176 		if (deltas[i].copy)
177 			size += GOT_DELTA_SIZE_SHIFT;
178 		else
179 			size += deltas[i].len + 1;
180 	}
181 	return size;
182 }
183 
184 static const struct got_error *
185 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
186 {
187 	char *n;
188 
189 	if (*len + nseg >= *sz) {
190 		while (*len + nseg >= *sz)
191 			*sz += *sz / 2;
192 		n = realloc(*p, *sz);
193 		if (n == NULL)
194 			return got_error_from_errno("realloc");
195 		*p = n;
196 	}
197 	memcpy(*p + *len, seg, nseg);
198 	*len += nseg;
199 	return NULL;
200 }
201 
202 static const struct got_error *
203 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
204     struct got_delta_instruction *deltas, int ndeltas,
205     off_t delta_size, off_t base_size)
206 {
207 	const struct got_error *err;
208 	unsigned char buf[16], *bp;
209 	int i, j;
210 	size_t len = 0, compressed_len;
211 	off_t bufsize = delta_size;
212 	off_t n;
213 	struct got_delta_instruction *d;
214 	uint8_t *delta_buf;
215 
216 	delta_buf = malloc(bufsize);
217 	if (delta_buf == NULL)
218 		return got_error_from_errno("malloc");
219 
220 	/* base object size */
221 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
222 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
223 	for (i = 1; n > 0; i++) {
224 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
225 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
226 		n >>= GOT_DELTA_SIZE_SHIFT;
227 	}
228 	err = append(&delta_buf, &len, &bufsize, buf, i);
229 	if (err)
230 		goto done;
231 
232 	/* target object size */
233 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
234 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
235 	for (i = 1; n > 0; i++) {
236 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
237 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
238 		n >>= GOT_DELTA_SIZE_SHIFT;
239 	}
240 	err = append(&delta_buf, &len, &bufsize, buf, i);
241 	if (err)
242 		goto done;
243 
244 	for (j = 0; j < ndeltas; j++) {
245 		d = &deltas[j];
246 		if (d->copy) {
247 			n = d->offset;
248 			bp = &buf[1];
249 			buf[0] = GOT_DELTA_BASE_COPY;
250 			for (i = 0; i < 4; i++) {
251 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
252 				buf[0] |= 1 << i;
253 				*bp++ = n & 0xff;
254 				n >>= 8;
255 				if (n == 0)
256 					break;
257 			}
258 
259 			n = d->len;
260 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
261 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
262 				for (i = 0; i < 3 && n > 0; i++) {
263 					buf[0] |= 1 << (i + 4);
264 					*bp++ = n & 0xff;
265 					n >>= 8;
266 				}
267 			}
268 			err = append(&delta_buf, &len, &bufsize,
269 			    buf, bp - buf);
270 			if (err)
271 				goto done;
272 		} else if (o->f == NULL) {
273 			n = 0;
274 			while (n != d->len) {
275 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
276 				err = append(&delta_buf, &len, &bufsize,
277 				    buf, 1);
278 				if (err)
279 					goto done;
280 				err = append(&delta_buf, &len, &bufsize,
281 				    o->data + o->hdrlen + d->offset + n,
282 				    buf[0]);
283 				if (err)
284 					goto done;
285 				n += buf[0];
286 			}
287 		} else {
288 			char content[128];
289 			size_t r;
290 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
291 				err = got_error_from_errno("fseeko");
292 				goto done;
293 			}
294 			n = 0;
295 			while (n != d->len) {
296 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
297 				err = append(&delta_buf, &len, &bufsize,
298 				    buf, 1);
299 				if (err)
300 					goto done;
301 				r = fread(content, 1, buf[0], o->f);
302 				if (r != buf[0]) {
303 					err = got_ferror(o->f, GOT_ERR_IO);
304 					goto done;
305 				}
306 				err = append(&delta_buf, &len, &bufsize,
307 				    content, buf[0]);
308 				if (err)
309 					goto done;
310 				n += buf[0];
311 			}
312 		}
313 	}
314 
315 	err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
316 	    NULL, NULL, delta_buf, 0, len);
317 	if (err)
318 		goto done;
319 
320 	m->delta_len = len;
321 	m->delta_compressed_len = compressed_len;
322 done:
323 	free(delta_buf);
324 	return err;
325 }
326 
327 static const struct got_error *
328 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
329     struct got_delta_instruction *deltas, int ndeltas,
330     off_t base_size, FILE *f)
331 {
332 	const struct got_error *err;
333 	unsigned char buf[16], *bp;
334 	int i, j;
335 	off_t n;
336 	struct got_deflate_buf zb;
337 	struct got_delta_instruction *d;
338 	off_t delta_len = 0, compressed_len = 0;
339 
340 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
341 	if (err)
342 		return err;
343 
344 	/* base object size */
345 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
346 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
347 	for (i = 1; n > 0; i++) {
348 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
349 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
350 		n >>= GOT_DELTA_SIZE_SHIFT;
351 	}
352 
353 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
354 	    buf, 0, i, f, NULL);
355 	if (err)
356 		goto done;
357 	delta_len += i;
358 
359 	/* target object size */
360 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
361 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
362 	for (i = 1; n > 0; i++) {
363 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
364 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
365 		n >>= GOT_DELTA_SIZE_SHIFT;
366 	}
367 
368 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
369 	    buf, 0, i, f, NULL);
370 	if (err)
371 		goto done;
372 	delta_len += i;
373 
374 	for (j = 0; j < ndeltas; j++) {
375 		d = &deltas[j];
376 		if (d->copy) {
377 			n = d->offset;
378 			bp = &buf[1];
379 			buf[0] = GOT_DELTA_BASE_COPY;
380 			for (i = 0; i < 4; i++) {
381 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
382 				buf[0] |= 1 << i;
383 				*bp++ = n & 0xff;
384 				n >>= 8;
385 				if (n == 0)
386 					break;
387 			}
388 			n = d->len;
389 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
390 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
391 				for (i = 0; i < 3 && n > 0; i++) {
392 					buf[0] |= 1 << (i + 4);
393 					*bp++ = n & 0xff;
394 					n >>= 8;
395 				}
396 			}
397 			err = got_deflate_append_to_file_mmap(&zb,
398 			    &compressed_len, buf, 0, bp - buf, f, NULL);
399 			if (err)
400 				goto done;
401 			delta_len += (bp - buf);
402 		} else if (o->f == NULL) {
403 			n = 0;
404 			while (n != d->len) {
405 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
406 				err = got_deflate_append_to_file_mmap(&zb,
407 				    &compressed_len, buf, 0, 1, f, NULL);
408 				if (err)
409 					goto done;
410 				delta_len++;
411 				err = got_deflate_append_to_file_mmap(&zb,
412 				    &compressed_len,
413 				    o->data + o->hdrlen + d->offset + n, 0,
414 				    buf[0], f, NULL);
415 				if (err)
416 					goto done;
417 				delta_len += buf[0];
418 				n += buf[0];
419 			}
420 		} else {
421 			char content[128];
422 			size_t r;
423 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
424 				err = got_error_from_errno("fseeko");
425 				goto done;
426 			}
427 			n = 0;
428 			while (n != d->len) {
429 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
430 				err = got_deflate_append_to_file_mmap(&zb,
431 				    &compressed_len, buf, 0, 1, f, NULL);
432 				if (err)
433 					goto done;
434 				delta_len++;
435 				r = fread(content, 1, buf[0], o->f);
436 				if (r != buf[0]) {
437 					err = got_ferror(o->f, GOT_ERR_IO);
438 					goto done;
439 				}
440 				err = got_deflate_append_to_file_mmap(&zb,
441 				    &compressed_len, content, 0, buf[0], f,
442 				    NULL);
443 				if (err)
444 					goto done;
445 				delta_len += buf[0];
446 				n += buf[0];
447 			}
448 		}
449 	}
450 
451 	err = got_deflate_flush(&zb, f, NULL, &compressed_len);
452 	if (err)
453 		goto done;
454 
455 	/* sanity check */
456 	if (compressed_len != ftello(f) - m->delta_offset) {
457 		err = got_error(GOT_ERR_COMPRESSION);
458 		goto done;
459 	}
460 
461 	m->delta_len = delta_len;
462 	m->delta_compressed_len = compressed_len;
463 done:
464 	got_deflate_end(&zb);
465 	return err;
466 }
467 
468 static const struct got_error *
469 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
470     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
471     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
472     int nobj_written)
473 {
474 	const struct got_error *err;
475 	int elapsed;
476 
477 	if (progress_cb == NULL)
478 		return NULL;
479 
480 	err = got_ratelimit_check(&elapsed, rl);
481 	if (err || !elapsed)
482 		return err;
483 
484 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
485 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
486 }
487 
488 static const struct got_error *
489 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
490 {
491 	if (v->nmeta == v->metasz){
492 		size_t newsize = 2 * v->metasz;
493 		struct got_pack_meta **new;
494 		new = reallocarray(v->meta, newsize, sizeof(*new));
495 		if (new == NULL)
496 			return got_error_from_errno("reallocarray");
497 		v->meta = new;
498 		v->metasz = newsize; 
499 	}
500 
501 	v->meta[v->nmeta++] = m;
502 	return NULL;
503 }
504 
505 static const struct got_error *
506 find_pack_for_reuse(struct got_packidx **best_packidx,
507     struct got_repository *repo)
508 {
509 	const struct got_error *err = NULL;
510 	struct got_pathlist_entry *pe;
511 	const char *best_packidx_path = NULL;
512 	int nobj_max = 0;
513 
514 	*best_packidx = NULL;
515 
516 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
517 		const char *path_packidx = pe->path;
518 		struct got_packidx *packidx;
519 		int nobj;
520 
521 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
522 		if (err)
523 			break;
524 
525 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
526 		if (nobj > nobj_max) {
527 			best_packidx_path = path_packidx;
528 			nobj_max = nobj;
529 		}
530 	}
531 
532 	if (best_packidx_path) {
533 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
534 		    repo);
535 	}
536 
537 	return err;
538 }
539 
540 struct send_id_arg {
541 	struct imsgbuf *ibuf;
542 	struct got_object_id *ids[GOT_IMSG_OBJ_ID_LIST_MAX_NIDS];
543 	size_t nids;
544 };
545 
546 static const struct got_error *
547 send_id(struct got_object_id *id, void *data, void *arg)
548 {
549 	const struct got_error *err = NULL;
550 	struct send_id_arg *a = arg;
551 
552 	a->ids[a->nids++] = id;
553 
554 	if (a->nids >= GOT_IMSG_OBJ_ID_LIST_MAX_NIDS) {
555 		err = got_privsep_send_object_idlist(a->ibuf, a->ids, a->nids);
556 		if (err)
557 			return err;
558 		a->nids = 0;
559 	}
560 
561 	return NULL;
562 }
563 
564 static const struct got_error *
565 recv_reused_delta(struct got_imsg_reused_delta *delta,
566     struct got_object_idset *idset, struct got_pack_metavec *v)
567 {
568 	struct got_pack_meta *m, *base;
569 
570 	if (delta->delta_offset + delta->delta_size < delta->delta_offset ||
571 	    delta->delta_offset +
572 	    delta->delta_compressed_size < delta->delta_offset)
573 		return got_error(GOT_ERR_BAD_PACKFILE);
574 
575 	m = got_object_idset_get(idset, &delta->id);
576 	if (m == NULL)
577 		return got_error(GOT_ERR_NO_OBJ);
578 
579 	base = got_object_idset_get(idset, &delta->base_id);
580 	if (base == NULL)
581 		return got_error(GOT_ERR_NO_OBJ);
582 
583 	m->delta_len = delta->delta_size;
584 	m->delta_compressed_len = delta->delta_compressed_size;
585 	m->delta_offset = delta->delta_out_offset;
586 	m->prev = base;
587 	m->size = delta->result_size;
588 	m->reused_delta_offset = delta->delta_offset;
589 	m->base_obj_id = got_object_id_dup(&delta->base_id);
590 	if (m->base_obj_id == NULL)
591 		return got_error_from_errno("got_object_id_dup");
592 
593 	return add_meta(m, v);
594 }
595 
596 static const struct got_error *
597 prepare_delta_reuse(struct got_pack **pack, struct got_packidx *packidx,
598     int delta_outfd, struct got_repository *repo)
599 {
600 	const struct got_error *err = NULL;
601 	char *path_packfile = NULL;
602 
603 	err = got_packidx_get_packfile_path(&path_packfile,
604 	    packidx->path_packidx);
605 	if (err)
606 		return err;
607 
608 	*pack = got_repo_get_cached_pack(repo, path_packfile);
609 	if (*pack == NULL) {
610 		err = got_repo_cache_pack(pack, repo, path_packfile, packidx);
611 		if (err)
612 			goto done;
613 	}
614 	if ((*pack)->privsep_child == NULL) {
615 		err = got_pack_start_privsep_child(*pack, packidx);
616 		if (err)
617 			goto done;
618 	}
619 
620 	if (!(*pack)->child_has_delta_outfd) {
621 		int outfd_child;
622 		outfd_child = dup(delta_outfd);
623 		if (outfd_child == -1) {
624 			err = got_error_from_errno("dup");
625 			goto done;
626 		}
627 		err = got_privsep_send_raw_delta_outfd(
628 		    (*pack)->privsep_child->ibuf, outfd_child);
629 		if (err)
630 			goto done;
631 		(*pack)->child_has_delta_outfd = 1;
632 	}
633 
634 	err = got_privsep_send_delta_reuse_req((*pack)->privsep_child->ibuf);
635 done:
636 	free(path_packfile);
637 	return err;
638 }
639 
640 
641 static const struct got_error *
642 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
643     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
644     struct got_repository *repo,
645     got_pack_progress_cb progress_cb, void *progress_arg,
646     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
647 {
648 	const struct got_error *err = NULL;
649 	struct got_packidx *packidx;
650 	struct got_pack *pack;
651 	struct send_id_arg sia;
652 	struct got_imsg_reused_delta deltas[GOT_IMSG_REUSED_DELTAS_MAX_NDELTAS];
653 	size_t ndeltas, i;
654 
655 	err = find_pack_for_reuse(&packidx, repo);
656 	if (err)
657 		return err;
658 
659 	if (packidx == NULL)
660 		return NULL;
661 
662 	err = prepare_delta_reuse(&pack, packidx, delta_cache_fd, repo);
663 	if (err)
664 		return err;
665 
666 	memset(&sia, 0, sizeof(sia));
667 	sia.ibuf = pack->privsep_child->ibuf;
668 	err = got_object_idset_for_each(idset, send_id, &sia);
669 	if (err)
670 		return err;
671 	if (sia.nids > 0) {
672 		err = got_privsep_send_object_idlist(pack->privsep_child->ibuf,
673 		    sia.ids, sia.nids);
674 		if (err)
675 			return err;
676 	}
677 	err = got_privsep_send_object_idlist_done(pack->privsep_child->ibuf);
678 	if (err)
679 		return err;
680 
681 	for (;;) {
682 		int done = 0;
683 
684 		if (cancel_cb) {
685 			err = (*cancel_cb)(cancel_arg);
686 			if (err)
687 				break;
688 		}
689 
690 		err = got_privsep_recv_reused_deltas(&done, deltas, &ndeltas,
691 		    pack->privsep_child->ibuf);
692 		if (err || done)
693 			break;
694 
695 		for (i = 0; i < ndeltas; i++) {
696 			struct got_imsg_reused_delta *delta = &deltas[i];
697 			err = recv_reused_delta(delta, idset, v);
698 			if (err)
699 				goto done;
700 		}
701 
702 		err = report_progress(progress_cb, progress_arg, rl,
703 		    ncolored, nfound, ntrees, 0L, ncommits,
704 		    got_object_idset_num_elements(idset), v->nmeta, 0);
705 		if (err)
706 			break;
707 	}
708 done:
709 	return err;
710 }
711 
712 static const struct got_error *
713 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
714     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
715     struct got_repository *repo,
716     got_pack_progress_cb progress_cb, void *progress_arg,
717     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
718 {
719 	const struct got_error *err = NULL;
720 	struct got_pack_meta *m = NULL, *base = NULL;
721 	struct got_raw_object *raw = NULL, *base_raw = NULL;
722 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
723 	int i, j, ndeltas, best_ndeltas;
724 	off_t size, best_size;
725 	const int max_base_candidates = 3;
726 	size_t delta_memsize = 0;
727 	const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
728 	int outfd = -1;
729 	uint32_t delta_seed;
730 
731 	delta_seed = arc4random();
732 
733 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
734 	for (i = 0; i < nmeta; i++) {
735 		if (cancel_cb) {
736 			err = (*cancel_cb)(cancel_arg);
737 			if (err)
738 				break;
739 		}
740 		err = report_progress(progress_cb, progress_arg, rl,
741 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
742 		    nreused + i, 0);
743 		if (err)
744 			goto done;
745 		m = meta[i];
746 
747 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
748 		    m->obj_type == GOT_OBJ_TYPE_TAG)
749 			continue;
750 
751 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
752 		if (err)
753 			goto done;
754 		m->size = raw->size;
755 
756 		if (raw->f == NULL) {
757 			err = got_deltify_init_mem(&m->dtab, raw->data,
758 			    raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
759 		} else {
760 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
761 			    raw->size + raw->hdrlen, delta_seed);
762 		}
763 		if (err)
764 			goto done;
765 
766 		if (i > max_base_candidates) {
767 			struct got_pack_meta *n = NULL;
768 			n = meta[i - (max_base_candidates + 1)];
769 			got_deltify_free(n->dtab);
770 			n->dtab = NULL;
771 		}
772 
773 		best_size = raw->size;
774 		best_ndeltas = 0;
775 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
776 			if (cancel_cb) {
777 				err = (*cancel_cb)(cancel_arg);
778 				if (err)
779 					goto done;
780 			}
781 			base = meta[j];
782 			/* long chains make unpacking slow, avoid such bases */
783 			if (base->nchain >= 128 ||
784 			    base->obj_type != m->obj_type)
785 				continue;
786 
787 			err = got_object_raw_open(&base_raw, &outfd, repo,
788 			    &base->id);
789 			if (err)
790 				goto done;
791 
792 			if (raw->f == NULL && base_raw->f == NULL) {
793 				err = got_deltify_mem_mem(&deltas, &ndeltas,
794 				    raw->data, raw->hdrlen,
795 				    raw->size + raw->hdrlen, delta_seed,
796 				    base->dtab, base_raw->data,
797 				    base_raw->hdrlen,
798 				    base_raw->size + base_raw->hdrlen);
799 			} else if (raw->f == NULL) {
800 				err = got_deltify_mem_file(&deltas, &ndeltas,
801 				    raw->data, raw->hdrlen,
802 				    raw->size + raw->hdrlen, delta_seed,
803 				    base->dtab, base_raw->f,
804 				    base_raw->hdrlen,
805 				    base_raw->size + base_raw->hdrlen);
806 			} else if (base_raw->f == NULL) {
807 				err = got_deltify_file_mem(&deltas, &ndeltas,
808 				    raw->f, raw->hdrlen,
809 				    raw->size + raw->hdrlen, delta_seed,
810 				    base->dtab, base_raw->data,
811 				    base_raw->hdrlen,
812 				    base_raw->size + base_raw->hdrlen);
813 			} else {
814 				err = got_deltify(&deltas, &ndeltas,
815 				    raw->f, raw->hdrlen,
816 				    raw->size + raw->hdrlen, delta_seed,
817 				    base->dtab, base_raw->f, base_raw->hdrlen,
818 				    base_raw->size + base_raw->hdrlen);
819 			}
820 			got_object_raw_close(base_raw);
821 			base_raw = NULL;
822 			if (err)
823 				goto done;
824 
825 			size = delta_size(deltas, ndeltas);
826 			if (size + 32 < best_size){
827 				/*
828 				 * if we already picked a best delta,
829 				 * replace it.
830 				 */
831 				best_size = size;
832 				free(best_deltas);
833 				best_deltas = deltas;
834 				best_ndeltas = ndeltas;
835 				deltas = NULL;
836 				m->nchain = base->nchain + 1;
837 				m->prev = base;
838 				m->head = base->head;
839 				if (m->head == NULL)
840 					m->head = base;
841 			} else {
842 				free(deltas);
843 				deltas = NULL;
844 				ndeltas = 0;
845 			}
846 		}
847 
848 		if (best_ndeltas > 0) {
849 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
850 			    delta_memsize + best_size <= max_delta_memsize) {
851 				delta_memsize += best_size;
852 				err = encode_delta_in_mem(m, raw, best_deltas,
853 				    best_ndeltas, best_size, m->prev->size);
854 			} else {
855 				m->delta_offset = ftello(delta_cache);
856 				err = encode_delta(m, raw, best_deltas,
857 				    best_ndeltas, m->prev->size, delta_cache);
858 			}
859 			free(best_deltas);
860 			best_deltas = NULL;
861 			best_ndeltas = 0;
862 			if (err)
863 				goto done;
864 		}
865 
866 		got_object_raw_close(raw);
867 		raw = NULL;
868 	}
869 done:
870 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
871 		got_deltify_free(meta[i]->dtab);
872 		meta[i]->dtab = NULL;
873 	}
874 	if (raw)
875 		got_object_raw_close(raw);
876 	if (base_raw)
877 		got_object_raw_close(base_raw);
878 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
879 		err = got_error_from_errno("close");
880 	free(deltas);
881 	free(best_deltas);
882 	return err;
883 }
884 
885 static const struct got_error *
886 search_packidx(int *found, struct got_object_id *id,
887     struct got_repository *repo)
888 {
889 	const struct got_error *err = NULL;
890 	struct got_packidx *packidx = NULL;
891 	int idx;
892 
893 	*found = 0;
894 
895 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
896 	if (err == NULL)
897 		*found = 1; /* object is already packed */
898 	else if (err->code == GOT_ERR_NO_OBJ)
899 		err = NULL;
900 	return err;
901 }
902 
903 static const struct got_error *
904 add_object(int want_meta, struct got_object_idset *idset,
905     struct got_object_id *id, const char *path, int obj_type,
906     time_t mtime, uint32_t seed, int loose_obj_only,
907     struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
908     got_pack_progress_cb progress_cb, void *progress_arg,
909     struct got_ratelimit *rl)
910 {
911 	const struct got_error *err;
912 	struct got_pack_meta *m = NULL;
913 
914 	if (loose_obj_only) {
915 		int is_packed;
916 		err = search_packidx(&is_packed, id, repo);
917 		if (err)
918 			return err;
919 		if (is_packed && want_meta)
920 			return NULL;
921 	}
922 
923 	if (want_meta) {
924 		err = alloc_meta(&m, id, path, obj_type, mtime, seed);
925 		if (err)
926 			return err;
927 
928 		(*nfound)++;
929 		err = report_progress(progress_cb, progress_arg, rl,
930 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
931 		if (err) {
932 			clear_meta(m);
933 			free(m);
934 			return err;
935 		}
936 	}
937 
938 	err = got_object_idset_add(idset, id, m);
939 	if (err) {
940 		clear_meta(m);
941 		free(m);
942 	}
943 	return err;
944 }
945 
946 static const struct got_error *
947 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
948     struct got_object_idset *idset, struct got_object_idset *idset_exclude,
949     struct got_tree_object *tree,
950     const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
951     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
952     got_pack_progress_cb progress_cb, void *progress_arg,
953     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
954 {
955 	const struct got_error *err;
956 	char *p = NULL;
957 	int i;
958 
959 	(*ntrees)++;
960 	err = report_progress(progress_cb, progress_arg, rl,
961 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
962 	if (err)
963 		return err;
964 
965 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
966 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
967 		struct got_object_id *id = got_tree_entry_get_id(e);
968 		mode_t mode = got_tree_entry_get_mode(e);
969 
970 		if (cancel_cb) {
971 			err = (*cancel_cb)(cancel_arg);
972 			if (err)
973 				break;
974 		}
975 
976 		if (got_object_tree_entry_is_submodule(e) ||
977 		    got_object_idset_contains(idset, id) ||
978 		    got_object_idset_contains(idset_exclude, id))
979 			continue;
980 
981 		/*
982 		 * If got-read-pack is crawling trees for us then
983 		 * we are only here to collect blob IDs.
984 		 */
985 		if (ids == NULL && S_ISDIR(mode))
986 			continue;
987 
988 		if (asprintf(&p, "%s%s%s", dpath,
989 		    got_path_is_root_dir(dpath) ? "" : "/",
990 		    got_tree_entry_get_name(e)) == -1) {
991 			err = got_error_from_errno("asprintf");
992 			break;
993 		}
994 
995 		if (S_ISDIR(mode)) {
996 			struct got_object_qid *qid;
997 			err = got_object_qid_alloc(&qid, id);
998 			if (err)
999 				break;
1000 			qid->data = p;
1001 			p = NULL;
1002 			STAILQ_INSERT_TAIL(ids, qid, entry);
1003 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
1004 			err = add_object(want_meta,
1005 			    want_meta ? idset : idset_exclude, id, p,
1006 			    GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
1007 			    repo, ncolored, nfound, ntrees,
1008 			    progress_cb, progress_arg, rl);
1009 			if (err)
1010 				break;
1011 			free(p);
1012 			p = NULL;
1013 		} else {
1014 			free(p);
1015 			p = NULL;
1016 		}
1017 	}
1018 
1019 	free(p);
1020 	return err;
1021 }
1022 
1023 static const struct got_error *
1024 load_tree(int want_meta, struct got_object_idset *idset,
1025     struct got_object_idset *idset_exclude,
1026     struct got_object_id *tree_id, const char *dpath, time_t mtime,
1027     uint32_t seed, struct got_repository *repo, int loose_obj_only,
1028     int *ncolored, int *nfound, int *ntrees,
1029     got_pack_progress_cb progress_cb, void *progress_arg,
1030     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1031 {
1032 	const struct got_error *err = NULL;
1033 	struct got_object_id_queue tree_ids;
1034 	struct got_object_qid *qid;
1035 	struct got_tree_object *tree = NULL;
1036 
1037 	if (got_object_idset_contains(idset, tree_id) ||
1038 	    got_object_idset_contains(idset_exclude, tree_id))
1039 		return NULL;
1040 
1041 	err = got_object_qid_alloc(&qid, tree_id);
1042 	if (err)
1043 		return err;
1044 	qid->data = strdup(dpath);
1045 	if (qid->data == NULL) {
1046 		err = got_error_from_errno("strdup");
1047 		got_object_qid_free(qid);
1048 		return err;
1049 	}
1050 
1051 	STAILQ_INIT(&tree_ids);
1052 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
1053 
1054 	while (!STAILQ_EMPTY(&tree_ids)) {
1055 		const char *path;
1056 		if (cancel_cb) {
1057 			err = (*cancel_cb)(cancel_arg);
1058 			if (err)
1059 				break;
1060 		}
1061 
1062 		qid = STAILQ_FIRST(&tree_ids);
1063 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
1064 		path = qid->data;
1065 
1066 		if (got_object_idset_contains(idset, &qid->id) ||
1067 		    got_object_idset_contains(idset_exclude, &qid->id)) {
1068 			free(qid->data);
1069 			got_object_qid_free(qid);
1070 			continue;
1071 		}
1072 
1073 		err = add_object(want_meta, want_meta ? idset : idset_exclude,
1074 		    &qid->id, path, GOT_OBJ_TYPE_TREE,
1075 		    mtime, seed, loose_obj_only, repo,
1076 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1077 		if (err) {
1078 			free(qid->data);
1079 			got_object_qid_free(qid);
1080 			break;
1081 		}
1082 
1083 		err = got_object_open_as_tree(&tree, repo, &qid->id);
1084 		if (err) {
1085 			free(qid->data);
1086 			got_object_qid_free(qid);
1087 			break;
1088 		}
1089 
1090 		err = load_tree_entries(&tree_ids, want_meta, idset,
1091 		    idset_exclude, tree, path, mtime, seed, repo,
1092 		    loose_obj_only, ncolored, nfound, ntrees,
1093 		    progress_cb, progress_arg, rl,
1094 		    cancel_cb, cancel_arg);
1095 		free(qid->data);
1096 		got_object_qid_free(qid);
1097 		if (err)
1098 			break;
1099 
1100 		got_object_tree_close(tree);
1101 		tree = NULL;
1102 	}
1103 
1104 	STAILQ_FOREACH(qid, &tree_ids, entry)
1105 		free(qid->data);
1106 	got_object_id_queue_free(&tree_ids);
1107 	if (tree)
1108 		got_object_tree_close(tree);
1109 	return err;
1110 }
1111 
1112 static const struct got_error *
1113 load_commit(int want_meta, struct got_object_idset *idset,
1114     struct got_object_idset *idset_exclude,
1115     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1116     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1117     got_pack_progress_cb progress_cb, void *progress_arg,
1118     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1119 {
1120 	const struct got_error *err;
1121 	struct got_commit_object *commit;
1122 
1123 	if (got_object_idset_contains(idset, id) ||
1124 	    got_object_idset_contains(idset_exclude, id))
1125 		return NULL;
1126 
1127 	if (loose_obj_only) {
1128 		int is_packed;
1129 		err = search_packidx(&is_packed, id, repo);
1130 		if (err)
1131 			return err;
1132 		if (is_packed && want_meta)
1133 			return NULL;
1134 	}
1135 
1136 	err = got_object_open_as_commit(&commit, repo, id);
1137 	if (err)
1138 		return err;
1139 
1140 	err = add_object(want_meta, want_meta ? idset : idset_exclude,
1141 	    id, "", GOT_OBJ_TYPE_COMMIT,
1142 	    got_object_commit_get_committer_time(commit), seed,
1143 	    loose_obj_only, repo,
1144 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1145 	if (err)
1146 		goto done;
1147 
1148 	err = load_tree(want_meta, idset, idset_exclude,
1149 	    got_object_commit_get_tree_id(commit),
1150 	    "", got_object_commit_get_committer_time(commit), seed,
1151 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1152 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1153 done:
1154 	got_object_commit_close(commit);
1155 	return err;
1156 }
1157 
1158 static const struct got_error *
1159 load_tag(int want_meta, struct got_object_idset *idset,
1160     struct got_object_idset *idset_exclude,
1161     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
1162     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
1163     got_pack_progress_cb progress_cb, void *progress_arg,
1164     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1165 {
1166 	const struct got_error *err;
1167 	struct got_tag_object *tag = NULL;
1168 
1169 	if (got_object_idset_contains(idset, id) ||
1170 	    got_object_idset_contains(idset_exclude, id))
1171 		return NULL;
1172 
1173 	if (loose_obj_only) {
1174 		int is_packed;
1175 		err = search_packidx(&is_packed, id, repo);
1176 		if (err)
1177 			return err;
1178 		if (is_packed && want_meta)
1179 			return NULL;
1180 	}
1181 
1182 	err = got_object_open_as_tag(&tag, repo, id);
1183 	if (err)
1184 		return err;
1185 
1186 	err = add_object(want_meta, want_meta ? idset : idset_exclude,
1187 	    id, "", GOT_OBJ_TYPE_TAG,
1188 	    got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1189 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1190 	if (err)
1191 		goto done;
1192 
1193 	switch (got_object_tag_get_object_type(tag)) {
1194 	case GOT_OBJ_TYPE_COMMIT:
1195 		err = load_commit(want_meta, idset, idset_exclude,
1196 		    got_object_tag_get_object_id(tag), repo, seed,
1197 		    loose_obj_only, ncolored, nfound, ntrees,
1198 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1199 		break;
1200 	case GOT_OBJ_TYPE_TREE:
1201 		err = load_tree(want_meta, idset, idset_exclude,
1202 		    got_object_tag_get_object_id(tag), "",
1203 		    got_object_tag_get_tagger_time(tag), seed, repo,
1204 		    loose_obj_only, ncolored, nfound, ntrees,
1205 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1206 		break;
1207 	default:
1208 		break;
1209 	}
1210 
1211 done:
1212 	got_object_tag_close(tag);
1213 	return err;
1214 }
1215 
1216 enum findtwixt_color {
1217 	COLOR_KEEP = 0,
1218 	COLOR_DROP,
1219 	COLOR_BLANK,
1220 	COLOR_SKIP,
1221 };
1222 
1223 static const int findtwixt_colors[] = {
1224 	COLOR_KEEP,
1225 	COLOR_DROP,
1226 	COLOR_BLANK,
1227 	COLOR_SKIP,
1228 };
1229 
1230 static const struct got_error *
1231 paint_commit(struct got_object_qid *qid, int color)
1232 {
1233 	if (color < 0 || color >= nitems(findtwixt_colors))
1234 		return got_error(GOT_ERR_RANGE);
1235 
1236 	qid->data = (void *)&findtwixt_colors[color];
1237 	return NULL;
1238 }
1239 
1240 static const struct got_error *
1241 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1242     int color, struct got_repository *repo)
1243 {
1244 	const struct got_error *err;
1245 	struct got_object_qid *qid;
1246 
1247 	err = got_object_qid_alloc(&qid, id);
1248 	if (err)
1249 		return err;
1250 
1251 	STAILQ_INSERT_TAIL(ids, qid, entry);
1252 	return paint_commit(qid, color);
1253 }
1254 
1255 struct append_id_arg {
1256 	struct got_object_id **array;
1257 	int idx;
1258 	struct got_object_idset *drop;
1259 	struct got_object_idset *skip;
1260 };
1261 
1262 static const struct got_error *
1263 append_id(struct got_object_id *id, void *data, void *arg)
1264 {
1265 	struct append_id_arg *a = arg;
1266 
1267 	if (got_object_idset_contains(a->skip, id) ||
1268 	    got_object_idset_contains(a->drop, id))
1269 		return NULL;
1270 
1271 	a->array[++a->idx] = got_object_id_dup(id);
1272 	if (a->array[a->idx] == NULL)
1273 		return got_error_from_errno("got_object_id_dup");
1274 
1275 	return NULL;
1276 }
1277 
1278 static const struct got_error *
1279 queue_commit_or_tag_id(struct got_object_id *id, int color,
1280     struct got_object_id_queue *ids, struct got_repository *repo)
1281 {
1282 	const struct got_error *err;
1283 	struct got_tag_object *tag = NULL;
1284 	int obj_type;
1285 
1286 	err = got_object_get_type(&obj_type, repo, id);
1287 	if (err)
1288 		return err;
1289 
1290 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1291 		err = got_object_open_as_tag(&tag, repo, id);
1292 		if (err)
1293 			return err;
1294 		obj_type = got_object_tag_get_object_type(tag);
1295 		id = got_object_tag_get_object_id(tag);
1296 	}
1297 
1298 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1299 		err = queue_commit_id(ids, id, color, repo);
1300 		if (err)
1301 			goto done;
1302 	}
1303 done:
1304 	if (tag)
1305 		got_object_tag_close(tag);
1306 	return err;
1307 }
1308 
1309 static const struct got_error *
1310 paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1311     struct got_object_idset *keep, struct got_object_idset *drop,
1312     struct got_object_idset *skip, struct got_repository *repo,
1313     got_pack_progress_cb progress_cb, void *progress_arg,
1314     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1315 {
1316 	const struct got_error *err = NULL;
1317 	struct got_commit_object *commit = NULL;
1318 	const struct got_object_id_queue *parents;
1319 	struct got_object_qid *qid;
1320 	int nqueued = nids, nskip = 0;
1321 
1322 	while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1323 		int color;
1324 
1325 		if (cancel_cb) {
1326 			err = cancel_cb(cancel_arg);
1327 			if (err)
1328 				break;
1329 		}
1330 
1331 		qid = STAILQ_FIRST(ids);
1332 		STAILQ_REMOVE_HEAD(ids, entry);
1333 		nqueued--;
1334 		color = *((int *)qid->data);
1335 		if (color == COLOR_SKIP)
1336 			nskip--;
1337 
1338 		if (got_object_idset_contains(skip, &qid->id)) {
1339 			got_object_qid_free(qid);
1340 			continue;
1341 		}
1342 
1343 		switch (color) {
1344 		case COLOR_KEEP:
1345 			if (got_object_idset_contains(keep, &qid->id)) {
1346 				got_object_qid_free(qid);
1347 				continue;
1348 			}
1349 			if (got_object_idset_contains(drop, &qid->id)) {
1350 				err = paint_commit(qid, COLOR_SKIP);
1351 				if (err)
1352 					goto done;
1353 			} else
1354 				(*ncolored)++;
1355 			err = got_object_idset_add(keep, &qid->id, NULL);
1356 			if (err)
1357 				goto done;
1358 			break;
1359 		case COLOR_DROP:
1360 			if (got_object_idset_contains(drop, &qid->id)) {
1361 				got_object_qid_free(qid);
1362 				continue;
1363 			}
1364 			if (got_object_idset_contains(keep, &qid->id)) {
1365 				err = paint_commit(qid, COLOR_SKIP);
1366 				if (err)
1367 					goto done;
1368 			} else
1369 				(*ncolored)++;
1370 			err = got_object_idset_add(drop, &qid->id, NULL);
1371 			if (err)
1372 				goto done;
1373 			break;
1374 		case COLOR_SKIP:
1375 			if (!got_object_idset_contains(skip, &qid->id)) {
1376 				err = got_object_idset_add(skip, &qid->id,
1377 				    NULL);
1378 				if (err)
1379 					goto done;
1380 			}
1381 			break;
1382 		default:
1383 			/* should not happen */
1384 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1385 			    "%s invalid commit color %d", __func__, color);
1386 			goto done;
1387 		}
1388 
1389 		err = report_progress(progress_cb, progress_arg, rl,
1390 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1391 		if (err)
1392 			break;
1393 
1394 
1395 		err = got_object_open_as_commit(&commit, repo, &qid->id);
1396 		if (err)
1397 			break;
1398 
1399 		parents = got_object_commit_get_parent_ids(commit);
1400 		if (parents) {
1401 			struct got_object_qid *pid;
1402 			color = *((int *)qid->data);
1403 			STAILQ_FOREACH(pid, parents, entry) {
1404 				err = queue_commit_id(ids, &pid->id, color,
1405 				    repo);
1406 				if (err)
1407 					break;
1408 				nqueued++;
1409 				if (color == COLOR_SKIP)
1410 					nskip++;
1411 			}
1412 		}
1413 
1414 		got_object_commit_close(commit);
1415 		commit = NULL;
1416 		got_object_qid_free(qid);
1417 	}
1418 done:
1419 	if (commit)
1420 		got_object_commit_close(commit);
1421 	return err;
1422 }
1423 
1424 static const struct got_error *
1425 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1426     struct got_object_id **head, int nhead,
1427     struct got_object_id **tail, int ntail,
1428     struct got_repository *repo,
1429     got_pack_progress_cb progress_cb, void *progress_arg,
1430     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1431 {
1432 	const struct got_error *err = NULL;
1433 	struct got_object_id_queue ids;
1434 	struct got_object_idset *keep, *drop, *skip = NULL;
1435 	int i, nkeep;
1436 
1437 	STAILQ_INIT(&ids);
1438 	*res = NULL;
1439 	*nres = 0;
1440 	*ncolored = 0;
1441 
1442 	keep = got_object_idset_alloc();
1443 	if (keep == NULL)
1444 		return got_error_from_errno("got_object_idset_alloc");
1445 
1446 	drop = got_object_idset_alloc();
1447 	if (drop == NULL) {
1448 		err = got_error_from_errno("got_object_idset_alloc");
1449 		goto done;
1450 	}
1451 
1452 	skip = got_object_idset_alloc();
1453 	if (skip == NULL) {
1454 		err = got_error_from_errno("got_object_idset_alloc");
1455 		goto done;
1456 	}
1457 
1458 	for (i = 0; i < nhead; i++) {
1459 		struct got_object_id *id = head[i];
1460 		if (id == NULL)
1461 			continue;
1462 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1463 		if (err)
1464 			goto done;
1465 	}		
1466 
1467 	for (i = 0; i < ntail; i++) {
1468 		struct got_object_id *id = tail[i];
1469 		if (id == NULL)
1470 			continue;
1471 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1472 		if (err)
1473 			goto done;
1474 	}
1475 
1476 	err = paint_commits(ncolored, &ids, nhead + ntail,
1477 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1478 	    cancel_cb, cancel_arg);
1479 	if (err)
1480 		goto done;
1481 
1482 	nkeep = got_object_idset_num_elements(keep);
1483 	if (nkeep > 0) {
1484 		struct append_id_arg arg;
1485 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1486 		if (arg.array == NULL) {
1487 			err = got_error_from_errno("calloc");
1488 			goto done;
1489 		}
1490 		arg.idx = -1;
1491 		arg.skip = skip;
1492 		arg.drop = drop;
1493 		err = got_object_idset_for_each(keep, append_id, &arg);
1494 		if (err) {
1495 			free(arg.array);
1496 			goto done;
1497 		}
1498 		*res = arg.array;
1499 		*nres = arg.idx + 1;
1500 	}
1501 done:
1502 	got_object_idset_free(keep);
1503 	got_object_idset_free(drop);
1504 	if (skip)
1505 		got_object_idset_free(skip);
1506 	got_object_id_queue_free(&ids);
1507 	return err;
1508 }
1509 
1510 struct load_packed_obj_arg {
1511 	/* output parameters: */
1512 	struct got_object_id *id;
1513 	char *dpath;
1514 	time_t mtime;
1515 
1516 	/* input parameters: */
1517 	uint32_t seed;
1518 	int want_meta;
1519 	struct got_object_idset *idset;
1520 	struct got_object_idset *idset_exclude;
1521 	int loose_obj_only;
1522 	int *ncolored;
1523 	int *nfound;
1524 	int *ntrees;
1525 	got_pack_progress_cb progress_cb;
1526 	void *progress_arg;
1527 	struct got_ratelimit *rl;
1528 	got_cancel_cb cancel_cb;
1529 	void *cancel_arg;
1530 };
1531 
1532 static const struct got_error *
1533 load_packed_commit_id(void *arg, time_t mtime, struct got_object_id *id,
1534     struct got_repository *repo)
1535 {
1536 	struct load_packed_obj_arg *a = arg;
1537 
1538 	if (got_object_idset_contains(a->idset, id) ||
1539 	    got_object_idset_contains(a->idset_exclude, id))
1540 		return NULL;
1541 
1542 	return add_object(a->want_meta,
1543 	    a->want_meta ? a->idset : a->idset_exclude,
1544 	    id, "", GOT_OBJ_TYPE_COMMIT, mtime, a->seed, a->loose_obj_only,
1545 	    repo, a->ncolored, a->nfound, a->ntrees,
1546 	    a->progress_cb, a->progress_arg, a->rl);
1547 }
1548 
1549 static const struct got_error *
1550 load_packed_tree_ids(void *arg, struct got_tree_object *tree, time_t mtime,
1551     struct got_object_id *id, const char *dpath, struct got_repository *repo)
1552 {
1553 	const struct got_error *err;
1554 	struct load_packed_obj_arg *a = arg;
1555 	const char *relpath;
1556 
1557 	/*
1558 	 * When we receive a tree's ID and path but not the tree itself,
1559 	 * this tree object was not found in the pack file. This is the
1560 	 * last time we are being called for this optimized traversal.
1561 	 * Return from here and switch to loading objects the slow way.
1562 	 */
1563 	if (tree == NULL) {
1564 		free(a->id);
1565 		a->id = got_object_id_dup(id);
1566 		if (a->id == NULL) {
1567 			err = got_error_from_errno("got_object_id_dup");
1568 			free(a->dpath);
1569 			a->dpath = NULL;
1570 			return err;
1571 		}
1572 
1573 		free(a->dpath);
1574 		a->dpath = strdup(dpath);
1575 		if (a->dpath == NULL) {
1576 			err = got_error_from_errno("strdup");
1577 			free(a->id);
1578 			a->id = NULL;
1579 			return err;
1580 		}
1581 
1582 		a->mtime = mtime;
1583 		return NULL;
1584 	}
1585 
1586 	if (got_object_idset_contains(a->idset, id) ||
1587 	    got_object_idset_contains(a->idset_exclude, id))
1588 		return NULL;
1589 
1590 	relpath = dpath;
1591 	while (relpath[0] == '/')
1592 		relpath++;
1593 
1594 	err = add_object(a->want_meta,
1595 	    a->want_meta ? a->idset : a->idset_exclude,
1596 	    id, relpath, GOT_OBJ_TYPE_TREE, mtime, a->seed,
1597 	    a->loose_obj_only, repo, a->ncolored, a->nfound, a->ntrees,
1598 	    a->progress_cb, a->progress_arg, a->rl);
1599 	if (err)
1600 		return err;
1601 
1602 	return load_tree_entries(NULL, a->want_meta, a->idset,
1603 	    a->idset_exclude, tree, dpath, mtime, a->seed, repo,
1604 	    a->loose_obj_only, a->ncolored, a->nfound, a->ntrees,
1605 	    a->progress_cb, a->progress_arg, a->rl,
1606 	    a->cancel_cb, a->cancel_arg);
1607 }
1608 
1609 static const struct got_error *
1610 load_packed_object_ids(int *found_all_objects,
1611     struct got_object_id **ours, int nours,
1612     struct got_object_id **theirs, int ntheirs,
1613     int want_meta, uint32_t seed, struct got_object_idset *idset,
1614     struct got_object_idset *idset_exclude, int loose_obj_only,
1615     struct got_repository *repo, struct got_packidx *packidx,
1616     int *ncolored, int *nfound, int *ntrees,
1617     got_pack_progress_cb progress_cb, void *progress_arg,
1618     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1619 {
1620 	const struct got_error *err = NULL;
1621 	struct load_packed_obj_arg lpa;
1622 
1623 	memset(&lpa, 0, sizeof(lpa));
1624 	lpa.seed = seed;
1625 	lpa.want_meta = want_meta;
1626 	lpa.idset = idset;
1627 	lpa.idset_exclude = idset_exclude;
1628 	lpa.loose_obj_only = loose_obj_only;
1629 	lpa.ncolored = ncolored;
1630 	lpa.nfound = nfound;
1631 	lpa.ntrees = ntrees;
1632 	lpa.progress_cb = progress_cb;
1633 	lpa.progress_arg = progress_arg;
1634 	lpa.rl = rl;
1635 	lpa.cancel_cb = cancel_cb;
1636 	lpa.cancel_arg = cancel_arg;
1637 
1638 	/* Attempt to load objects via got-read-pack, as far as possible. */
1639 	err = got_object_enumerate(found_all_objects, load_packed_commit_id,
1640 	   load_packed_tree_ids, &lpa, ours, nours, theirs, ntheirs,
1641 	   packidx, repo);
1642 	if (err)
1643 		return err;
1644 
1645 	if (lpa.id == NULL)
1646 		return NULL;
1647 
1648 	/*
1649 	 * An incomplete tree hierarchy was present in the pack file
1650 	 * and caused loading to be aborted.
1651 	 * Continue loading trees the slow way.
1652 	 */
1653 	err = load_tree(want_meta, idset, idset_exclude,
1654 	    lpa.id, lpa.dpath, lpa.mtime, seed, repo, loose_obj_only,
1655 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1656 	    cancel_cb, cancel_arg);
1657 	free(lpa.id);
1658 	free(lpa.dpath);
1659 	return err;
1660 }
1661 
1662 static const struct got_error *
1663 find_pack_for_enumeration(struct got_packidx **best_packidx,
1664     struct got_object_id **ids, int nids, struct got_repository *repo)
1665 {
1666 	const struct got_error *err = NULL;
1667 	struct got_pathlist_entry *pe;
1668 	const char *best_packidx_path = NULL;
1669 	int nobj_max = 0;
1670 	int ncommits_max = 0;
1671 
1672 	*best_packidx = NULL;
1673 
1674 	/*
1675 	 * Find the largest pack which contains at least some of the
1676 	 * commits and tags we are interested in.
1677 	 */
1678 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1679 		const char *path_packidx = pe->path;
1680 		struct got_packidx *packidx;
1681 		int nobj, i, idx, ncommits = 0;
1682 
1683 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1684 		if (err)
1685 			break;
1686 
1687 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1688 		if (nobj <= nobj_max)
1689 			continue;
1690 
1691 		for (i = 0; i < nids; i++) {
1692 			idx = got_packidx_get_object_idx(packidx, ids[i]);
1693 			if (idx != -1)
1694 				ncommits++;
1695 		}
1696 		if (ncommits > ncommits_max) {
1697 			best_packidx_path = path_packidx;
1698 			nobj_max = nobj;
1699 			ncommits_max = ncommits;
1700 		}
1701 	}
1702 
1703 	if (best_packidx_path && err == NULL) {
1704 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1705 		    repo);
1706 	}
1707 
1708 	return err;
1709 }
1710 
1711 static const struct got_error *
1712 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1713     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1714     struct got_object_id **ours, int nours, struct got_repository *repo,
1715     uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1716     void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1717     void *cancel_arg)
1718 {
1719 	const struct got_error *err = NULL;
1720 	struct got_object_id **ids = NULL;
1721 	struct got_packidx *packidx = NULL;
1722 	int i, nobj = 0, obj_type, found_all_objects = 0;
1723 	struct got_object_idset *idset_exclude;
1724 
1725 	idset_exclude = got_object_idset_alloc();
1726 	if (idset_exclude == NULL)
1727 		return got_error_from_errno("got_object_idset_alloc");
1728 
1729 	*ncolored = 0;
1730 	*nfound = 0;
1731 	*ntrees = 0;
1732 
1733 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1734 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1735 	if (err)
1736 		goto done;
1737 
1738 	err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
1739 	if (err)
1740 		goto done;
1741 	if (packidx) {
1742 		err = load_packed_object_ids(&found_all_objects,
1743 		    theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
1744 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1745 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1746 		if (err)
1747 			goto done;
1748 	}
1749 
1750 	for (i = 0; i < ntheirs; i++) {
1751 		struct got_object_id *id = theirs[i];
1752 		if (id == NULL)
1753 			continue;
1754 		err = got_object_get_type(&obj_type, repo, id);
1755 		if (err)
1756 			return err;
1757 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1758 			if (!found_all_objects) {
1759 				err = load_commit(0, idset, idset_exclude,
1760 				    id, repo, seed, loose_obj_only,
1761 				    ncolored, nfound, ntrees,
1762 				    progress_cb, progress_arg, rl,
1763 				    cancel_cb, cancel_arg);
1764 				if (err)
1765 					goto done;
1766 			}
1767 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1768 			err = load_tag(0, idset, idset_exclude, id, repo,
1769 			    seed, loose_obj_only, ncolored, nfound, ntrees,
1770 			    progress_cb, progress_arg, rl,
1771 			    cancel_cb, cancel_arg);
1772 			if (err)
1773 				goto done;
1774 		}
1775 	}
1776 
1777 	found_all_objects = 0;
1778 	err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
1779 	if (err)
1780 		goto done;
1781 	if (packidx) {
1782 		err = load_packed_object_ids(&found_all_objects, ids,
1783 		    nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
1784 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1785 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1786 		if (err)
1787 			goto done;
1788 	}
1789 
1790 	if (!found_all_objects) {
1791 		for (i = 0; i < nobj; i++) {
1792 			err = load_commit(1, idset, idset_exclude, ids[i],
1793 			    repo, seed, loose_obj_only, ncolored, nfound,
1794 			    ntrees, progress_cb, progress_arg, rl,
1795 			    cancel_cb, cancel_arg);
1796 			if (err)
1797 				goto done;
1798 		}
1799 	}
1800 
1801 	for (i = 0; i < nours; i++) {
1802 		struct got_object_id *id = ours[i];
1803 		struct got_pack_meta *m;
1804 		if (id == NULL)
1805 			continue;
1806 		m = got_object_idset_get(idset, id);
1807 		if (m == NULL) {
1808 			err = got_object_get_type(&obj_type, repo, id);
1809 			if (err)
1810 				goto done;
1811 		} else
1812 			obj_type = m->obj_type;
1813 		if (obj_type != GOT_OBJ_TYPE_TAG)
1814 			continue;
1815 		err = load_tag(1, idset, idset_exclude, id, repo,
1816 		    seed, loose_obj_only, ncolored, nfound, ntrees,
1817 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1818 		if (err)
1819 			goto done;
1820 	}
1821 done:
1822 	for (i = 0; i < nobj; i++) {
1823 		free(ids[i]);
1824 	}
1825 	free(ids);
1826 	got_object_idset_free(idset_exclude);
1827 	return err;
1828 }
1829 
1830 static const struct got_error *
1831 hwrite(FILE *f, void *buf, off_t len, SHA1_CTX *ctx)
1832 {
1833 	size_t n;
1834 
1835 	SHA1Update(ctx, buf, len);
1836 	n = fwrite(buf, 1, len, f);
1837 	if (n != len)
1838 		return got_ferror(f, GOT_ERR_IO);
1839 	return NULL;
1840 }
1841 
1842 static const struct got_error *
1843 hcopy(FILE *fsrc, FILE *fdst, off_t len, SHA1_CTX *ctx)
1844 {
1845 	unsigned char buf[65536];
1846 	off_t remain = len;
1847 	size_t n;
1848 
1849 	while (remain > 0) {
1850 		size_t copylen = MIN(sizeof(buf), remain);
1851 		n = fread(buf, 1, copylen, fsrc);
1852 		if (n != copylen)
1853 			return got_ferror(fsrc, GOT_ERR_IO);
1854 		SHA1Update(ctx, buf, copylen);
1855 		n = fwrite(buf, 1, copylen, fdst);
1856 		if (n != copylen)
1857 			return got_ferror(fdst, GOT_ERR_IO);
1858 		remain -= copylen;
1859 	}
1860 
1861 	return NULL;
1862 }
1863 
1864 static const struct got_error *
1865 hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
1866     FILE *fdst, off_t len, SHA1_CTX *ctx)
1867 {
1868 	size_t n;
1869 
1870 	if (src_offset + len > src_size)
1871 		return got_error(GOT_ERR_RANGE);
1872 
1873 	SHA1Update(ctx, src + src_offset, len);
1874 	n = fwrite(src + src_offset, 1, len, fdst);
1875 	if (n != len)
1876 		return got_ferror(fdst, GOT_ERR_IO);
1877 
1878 	return NULL;
1879 }
1880 
1881 static void
1882 putbe32(char *b, uint32_t n)
1883 {
1884 	b[0] = n >> 24;
1885 	b[1] = n >> 16;
1886 	b[2] = n >> 8;
1887 	b[3] = n >> 0;
1888 }
1889 
1890 static int
1891 write_order_cmp(const void *pa, const void *pb)
1892 {
1893 	struct got_pack_meta *a, *b, *ahd, *bhd;
1894 
1895 	a = *(struct got_pack_meta **)pa;
1896 	b = *(struct got_pack_meta **)pb;
1897 	ahd = (a->head == NULL) ? a : a->head;
1898 	bhd = (b->head == NULL) ? b : b->head;
1899 	if (bhd->mtime < ahd->mtime)
1900 		return -1;
1901 	if (bhd->mtime > ahd->mtime)
1902 		return 1;
1903 	if (bhd < ahd)
1904 		return -1;
1905 	if (bhd > ahd)
1906 		return 1;
1907 	if (a->nchain != b->nchain)
1908 		return a->nchain - b->nchain;
1909 	if (a->mtime < b->mtime)
1910 		return -1;
1911 	if (a->mtime > b->mtime)
1912 		return 1;
1913 	return got_object_id_cmp(&a->id, &b->id);
1914 }
1915 
1916 static int
1917 reuse_write_order_cmp(const void *pa, const void *pb)
1918 {
1919 	struct got_pack_meta *a, *b;
1920 
1921 	a = *(struct got_pack_meta **)pa;
1922 	b = *(struct got_pack_meta **)pb;
1923 
1924 	if (a->reused_delta_offset < b->reused_delta_offset)
1925 		return -1;
1926 	if (a->reused_delta_offset > b->reused_delta_offset)
1927 		return 1;
1928 	return 0;
1929 }
1930 
1931 static const struct got_error *
1932 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1933 {
1934 	size_t i;
1935 
1936 	*hdrlen = 0;
1937 
1938 	hdr[0] = obj_type << 4;
1939 	hdr[0] |= len & 0xf;
1940 	len >>= 4;
1941 	for (i = 1; len != 0; i++){
1942 		if (i >= bufsize)
1943 			return got_error(GOT_ERR_NO_SPACE);
1944 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1945 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1946 		len >>= GOT_DELTA_SIZE_SHIFT;
1947 	}
1948 
1949 	*hdrlen = i;
1950 	return NULL;
1951 }
1952 
1953 static int
1954 packoff(char *hdr, off_t off)
1955 {
1956 	int i, j;
1957 	char rbuf[8];
1958 
1959 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1960 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1961 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1962 		    GOT_DELTA_SIZE_MORE;
1963 	}
1964 
1965 	j = 0;
1966 	while (i > 0)
1967 		hdr[j++] = rbuf[--i];
1968 	return j;
1969 }
1970 
1971 static const struct got_error *
1972 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1973     struct got_pack_meta *m)
1974 {
1975 	const struct got_error *err;
1976 	char buf[32];
1977 	int nh;
1978 
1979 	if (m->prev->off != 0) {
1980 		err = packhdr(&nh, buf, sizeof(buf),
1981 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1982 		if (err)
1983 			return err;
1984 		nh += packoff(buf + nh, m->off - m->prev->off);
1985 		err = hwrite(packfile, buf, nh, ctx);
1986 		if (err)
1987 			return err;
1988 		*packfile_size += nh;
1989 	} else {
1990 		err = packhdr(&nh, buf, sizeof(buf),
1991 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1992 		if (err)
1993 			return err;
1994 		err = hwrite(packfile, buf, nh, ctx);
1995 		if (err)
1996 			return err;
1997 		*packfile_size += nh;
1998 		err = hwrite(packfile, m->prev->id.sha1,
1999 		    sizeof(m->prev->id.sha1), ctx);
2000 		if (err)
2001 			return err;
2002 		*packfile_size += sizeof(m->prev->id.sha1);
2003 	}
2004 
2005 	return NULL;
2006 }
2007 
2008 static const struct got_error *
2009 write_packed_object(off_t *packfile_size, FILE *packfile,
2010     FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
2011     struct got_pack_meta *m, int *outfd, SHA1_CTX *ctx,
2012     struct got_repository *repo)
2013 {
2014 	const struct got_error *err = NULL;
2015 	struct got_deflate_checksum csum;
2016 	char buf[32];
2017 	int nh;
2018 	struct got_raw_object *raw = NULL;
2019 	off_t outlen;
2020 
2021 	csum.output_sha1 = ctx;
2022 	csum.output_crc = NULL;
2023 
2024 	m->off = ftello(packfile);
2025 	if (m->delta_len == 0) {
2026 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
2027 		if (err)
2028 			goto done;
2029 		err = packhdr(&nh, buf, sizeof(buf),
2030 		    m->obj_type, raw->size);
2031 		if (err)
2032 			goto done;
2033 		err = hwrite(packfile, buf, nh, ctx);
2034 		if (err)
2035 			goto done;
2036 		*packfile_size += nh;
2037 		if (raw->f == NULL) {
2038 			err = got_deflate_to_file_mmap(&outlen,
2039 			    raw->data + raw->hdrlen, 0, raw->size,
2040 			    packfile, &csum);
2041 			if (err)
2042 				goto done;
2043 		} else {
2044 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
2045 			    == -1) {
2046 				err = got_error_from_errno("fseeko");
2047 				goto done;
2048 			}
2049 			err = got_deflate_to_file(&outlen, raw->f,
2050 			    raw->size, packfile, &csum);
2051 			if (err)
2052 				goto done;
2053 		}
2054 		*packfile_size += outlen;
2055 		got_object_raw_close(raw);
2056 		raw = NULL;
2057 	} else if (m->delta_buf) {
2058 		err = deltahdr(packfile_size, ctx, packfile, m);
2059 		if (err)
2060 			goto done;
2061 		err = hwrite(packfile, m->delta_buf,
2062 		    m->delta_compressed_len, ctx);
2063 		if (err)
2064 			goto done;
2065 		*packfile_size += m->delta_compressed_len;
2066 		free(m->delta_buf);
2067 		m->delta_buf = NULL;
2068 	} else if (delta_cache_map) {
2069 		err = deltahdr(packfile_size, ctx, packfile, m);
2070 		if (err)
2071 			goto done;
2072 		err = hcopy_mmap(delta_cache_map, m->delta_offset,
2073 		    delta_cache_size, packfile, m->delta_compressed_len,
2074 		    ctx);
2075 		if (err)
2076 			goto done;
2077 		*packfile_size += m->delta_compressed_len;
2078 	} else {
2079 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
2080 		    == -1) {
2081 			err = got_error_from_errno("fseeko");
2082 			goto done;
2083 		}
2084 		err = deltahdr(packfile_size, ctx, packfile, m);
2085 		if (err)
2086 			goto done;
2087 		err = hcopy(delta_cache, packfile,
2088 		    m->delta_compressed_len, ctx);
2089 		if (err)
2090 			goto done;
2091 		*packfile_size += m->delta_compressed_len;
2092 	}
2093 done:
2094 	if (raw)
2095 		got_object_raw_close(raw);
2096 	return err;
2097 }
2098 
2099 static const struct got_error *
2100 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
2101     struct got_pack_meta **deltify, int ndeltify,
2102     struct got_pack_meta **reuse, int nreuse,
2103     int ncolored, int nfound, int ntrees, int nours,
2104     struct got_repository *repo,
2105     got_pack_progress_cb progress_cb, void *progress_arg,
2106     struct got_ratelimit *rl,
2107     got_cancel_cb cancel_cb, void *cancel_arg)
2108 {
2109 	const struct got_error *err = NULL;
2110 	int i;
2111 	SHA1_CTX ctx;
2112 	struct got_pack_meta *m;
2113 	char buf[32];
2114 	size_t n;
2115 	off_t packfile_size = 0;
2116 	int outfd = -1;
2117 	int delta_cache_fd = -1;
2118 	uint8_t *delta_cache_map = NULL;
2119 	size_t delta_cache_size = 0;
2120 
2121 	SHA1Init(&ctx);
2122 
2123 #ifndef GOT_PACK_NO_MMAP
2124 	delta_cache_fd = dup(fileno(delta_cache));
2125 	if (delta_cache_fd != -1) {
2126 		struct stat sb;
2127 		if (fstat(delta_cache_fd, &sb) == -1) {
2128 			err = got_error_from_errno("fstat");
2129 			goto done;
2130 		}
2131 		if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
2132 			delta_cache_map = mmap(NULL, sb.st_size,
2133 			    PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
2134 			if (delta_cache_map == MAP_FAILED) {
2135 				if (errno != ENOMEM) {
2136 					err = got_error_from_errno("mmap");
2137 					goto done;
2138 				}
2139 				delta_cache_map = NULL; /* fallback on stdio */
2140 			} else
2141 				delta_cache_size = (size_t)sb.st_size;
2142 		}
2143 	}
2144 #endif
2145 	err = hwrite(packfile, "PACK", 4, &ctx);
2146 	if (err)
2147 		goto done;
2148 	putbe32(buf, GOT_PACKFILE_VERSION);
2149 	err = hwrite(packfile, buf, 4, &ctx);
2150 	if (err)
2151 		goto done;
2152 	putbe32(buf, ndeltify + nreuse);
2153 	err = hwrite(packfile, buf, 4, &ctx);
2154 	if (err)
2155 		goto done;
2156 
2157 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
2158 	    write_order_cmp);
2159 	for (i = 0; i < ndeltify; i++) {
2160 		err = report_progress(progress_cb, progress_arg, rl,
2161 		    ncolored, nfound, ntrees, packfile_size, nours,
2162 		    ndeltify + nreuse, ndeltify + nreuse, i);
2163 		if (err)
2164 			goto done;
2165 		m = deltify[i];
2166 		err = write_packed_object(&packfile_size, packfile,
2167 		    delta_cache, delta_cache_map, delta_cache_size,
2168 		    m, &outfd, &ctx, repo);
2169 		if (err)
2170 			goto done;
2171 	}
2172 
2173 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
2174 	    reuse_write_order_cmp);
2175 	for (i = 0; i < nreuse; i++) {
2176 		err = report_progress(progress_cb, progress_arg, rl,
2177 		    ncolored, nfound, ntrees, packfile_size, nours,
2178 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
2179 		if (err)
2180 			goto done;
2181 		m = reuse[i];
2182 		err = write_packed_object(&packfile_size, packfile,
2183 		    delta_cache, delta_cache_map, delta_cache_size,
2184 		    m, &outfd, &ctx, repo);
2185 		if (err)
2186 			goto done;
2187 	}
2188 
2189 	SHA1Final(pack_sha1, &ctx);
2190 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
2191 	if (n != SHA1_DIGEST_LENGTH)
2192 		err = got_ferror(packfile, GOT_ERR_IO);
2193 	packfile_size += SHA1_DIGEST_LENGTH;
2194 	packfile_size += sizeof(struct got_packfile_hdr);
2195 	if (progress_cb) {
2196 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2197 		    packfile_size, nours, ndeltify + nreuse,
2198 		    ndeltify + nreuse, ndeltify + nreuse);
2199 		if (err)
2200 			goto done;
2201 	}
2202 done:
2203 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
2204 		err = got_error_from_errno("close");
2205 	if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
2206 		err = got_error_from_errno("munmap");
2207 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2208 		err = got_error_from_errno("close");
2209 	return err;
2210 }
2211 
2212 static const struct got_error *
2213 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
2214 {
2215 	struct got_pack_meta *m = data;
2216 	struct got_pack_metavec *v = arg;
2217 
2218 	if (m->reused_delta_offset != 0)
2219 		return NULL;
2220 
2221 	return add_meta(m, v);
2222 }
2223 
2224 const struct got_error *
2225 got_pack_create(uint8_t *packsha1, FILE *packfile,
2226     struct got_object_id **theirs, int ntheirs,
2227     struct got_object_id **ours, int nours,
2228     struct got_repository *repo, int loose_obj_only, int allow_empty,
2229     got_pack_progress_cb progress_cb, void *progress_arg,
2230     got_cancel_cb cancel_cb, void *cancel_arg)
2231 {
2232 	const struct got_error *err;
2233 	int delta_cache_fd = -1;
2234 	FILE *delta_cache = NULL;
2235 	struct got_object_idset *idset;
2236 	struct got_ratelimit rl;
2237 	struct got_pack_metavec deltify, reuse;
2238 	int ncolored = 0, nfound = 0, ntrees = 0;
2239 	size_t ndeltify;
2240 	uint32_t seed;
2241 
2242 	seed = arc4random();
2243 
2244 	memset(&deltify, 0, sizeof(deltify));
2245 	memset(&reuse, 0, sizeof(reuse));
2246 
2247 	got_ratelimit_init(&rl, 0, 500);
2248 
2249 	idset = got_object_idset_alloc();
2250 	if (idset == NULL)
2251 		return got_error_from_errno("got_object_idset_alloc");
2252 
2253 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
2254 	    ntheirs, ours, nours, repo, seed, loose_obj_only,
2255 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
2256 	if (err)
2257 		goto done;
2258 
2259 	if (progress_cb) {
2260 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
2261 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
2262 		if (err)
2263 			goto done;
2264 	}
2265 
2266 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
2267 		err = got_error(GOT_ERR_CANNOT_PACK);
2268 		goto done;
2269 	}
2270 
2271 	delta_cache_fd = got_opentempfd();
2272 	if (delta_cache_fd == -1) {
2273 		err = got_error_from_errno("got_opentemp");
2274 		goto done;
2275 	}
2276 
2277 	reuse.metasz = 64;
2278 	reuse.meta = calloc(reuse.metasz,
2279 	    sizeof(struct got_pack_meta *));
2280 	if (reuse.meta == NULL) {
2281 		err = got_error_from_errno("calloc");
2282 		goto done;
2283 	}
2284 
2285 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
2286 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
2287 	    cancel_cb, cancel_arg);
2288 	if (err)
2289 		goto done;
2290 
2291 	delta_cache = fdopen(delta_cache_fd, "a+");
2292 	if (delta_cache == NULL) {
2293 		err = got_error_from_errno("fdopen");
2294 		goto done;
2295 	}
2296 	delta_cache_fd = -1;
2297 
2298 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
2299 		err = got_error_from_errno("fseeko");
2300 		goto done;
2301 	}
2302 
2303 	ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
2304 	if (ndeltify > 0) {
2305 		deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
2306 		if (deltify.meta == NULL) {
2307 			err = got_error_from_errno("calloc");
2308 			goto done;
2309 		}
2310 		deltify.metasz = ndeltify;
2311 
2312 		err = got_object_idset_for_each(idset, add_meta_idset_cb,
2313 		    &deltify);
2314 		if (err)
2315 			goto done;
2316 		if (deltify.nmeta > 0) {
2317 			err = pick_deltas(deltify.meta, deltify.nmeta,
2318 			    ncolored, nfound, ntrees, nours, reuse.nmeta,
2319 			    delta_cache, repo, progress_cb, progress_arg, &rl,
2320 			    cancel_cb, cancel_arg);
2321 			if (err)
2322 				goto done;
2323 		}
2324 	}
2325 
2326 	if (fflush(delta_cache) == EOF) {
2327 		err = got_error_from_errno("fflush");
2328 		goto done;
2329 	}
2330 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
2331 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
2332 	    nours, repo, progress_cb, progress_arg, &rl,
2333 	    cancel_cb, cancel_arg);
2334 	if (err)
2335 		goto done;
2336 done:
2337 	free_nmeta(deltify.meta, deltify.nmeta);
2338 	free_nmeta(reuse.meta, reuse.nmeta);
2339 	got_object_idset_free(idset);
2340 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
2341 		err = got_error_from_errno("close");
2342 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
2343 		err = got_error_from_errno("fclose");
2344 	return err;
2345 }