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