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