1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 #include <sys/queue.h>
20 #include <sys/tree.h>
21 #include <sys/uio.h>
22 #include <sys/stat.h>
23 #include <sys/time.h>
24 
25 #include <endian.h>
26 #include <stdint.h>
27 #include <imsg.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <sha1.h>
32 #include <time.h>
33 #include <limits.h>
34 #include <zlib.h>
35 
36 #include "got_error.h"
37 #include "got_cancel.h"
38 #include "got_object.h"
39 #include "got_path.h"
40 #include "got_reference.h"
41 #include "got_repository_admin.h"
42 #include "got_opentemp.h"
43 
44 #include "got_lib_deltify.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_idset.h"
48 #include "got_lib_object_cache.h"
49 #include "got_lib_deflate.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_privsep.h"
52 #include "got_lib_repository.h"
53 #include "got_lib_ratelimit.h"
54 
55 #ifndef MIN
56 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 #endif
58 
59 #ifndef MAX
60 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
61 #endif
62 
63 struct got_pack_meta {
64 	struct got_object_id id;
65 	char	*path;
66 	int	obj_type;
67 	off_t	size;
68 	time_t	mtime;
69 
70 	/* The best delta we picked */
71 	struct got_pack_meta *head;
72 	struct got_pack_meta *prev;
73 	unsigned char *delta_buf; /* if not encoded in delta cache file */
74 	off_t	delta_offset;	/* offset in delta cache file */
75 	off_t	delta_len;	/* encoded delta length */
76 	int	nchain;
77 
78 	int	have_reused_delta;
79 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
80 	struct got_object_id *base_obj_id;
81 
82 	/* Only used for delta window */
83 	struct got_delta_table *dtab;
84 
85 	/* Only used for writing offset deltas */
86 	off_t	off;
87 };
88 
89 struct got_pack_metavec {
90 	struct got_pack_meta **meta;
91 	int nmeta;
92 	int metasz;
93 };
94 
95 static const struct got_error *
96 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
97     const char *path, int obj_type, time_t mtime)
98 {
99 	const struct got_error *err = NULL;
100 	struct got_pack_meta *m;
101 
102 	*new = NULL;
103 
104 	m = calloc(1, sizeof(*m));
105 	if (m == NULL)
106 		return got_error_from_errno("calloc");
107 
108 	memcpy(&m->id, id, sizeof(m->id));
109 
110 	m->path = strdup(path);
111 	if (m->path == NULL) {
112 		err = got_error_from_errno("strdup");
113 		free(m);
114 		return err;
115 	}
116 
117 	m->obj_type = obj_type;
118 	m->mtime = mtime;
119 	*new = m;
120 	return NULL;
121 }
122 
123 static void
124 clear_meta(struct got_pack_meta *meta)
125 {
126 	if (meta == NULL)
127 		return;
128 	free(meta->path);
129 	meta->path = NULL;
130 	free(meta->delta_buf);
131 	meta->delta_buf = NULL;
132 	free(meta->base_obj_id);
133 	meta->base_obj_id = NULL;
134 }
135 
136 static void
137 free_nmeta(struct got_pack_meta **meta, int nmeta)
138 {
139 	int i;
140 
141 	for (i = 0; i < nmeta; i++)
142 		clear_meta(meta[i]);
143 	free(meta);
144 }
145 
146 static int
147 delta_order_cmp(const void *pa, const void *pb)
148 {
149 	struct got_pack_meta *a, *b;
150 	int cmp;
151 
152 	a = *(struct got_pack_meta **)pa;
153 	b = *(struct got_pack_meta **)pb;
154 
155 	if (a->obj_type != b->obj_type)
156 		return a->obj_type - b->obj_type;
157 	cmp = strcmp(a->path, b->path);
158 	if (cmp != 0)
159 		return cmp;
160 	if (a->mtime != b->mtime)
161 		return a->mtime - b->mtime;
162 	return got_object_id_cmp(&a->id, &b->id);
163 }
164 
165 static off_t
166 delta_size(struct got_delta_instruction *deltas, int ndeltas)
167 {
168 	int i;
169 	off_t size = 32;
170 	for (i = 0; i < ndeltas; i++) {
171 		if (deltas[i].copy)
172 			size += GOT_DELTA_SIZE_SHIFT;
173 		else
174 			size += deltas[i].len + 1;
175 	}
176 	return size;
177 }
178 
179 static const struct got_error *
180 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
181 {
182 	char *n;
183 
184 	if (*len + nseg >= *sz) {
185 		while (*len + nseg >= *sz)
186 			*sz += *sz / 2;
187 		n = realloc(*p, *sz);
188 		if (n == NULL)
189 			return got_error_from_errno("realloc");
190 		*p = n;
191 	}
192 	memcpy(*p + *len, seg, nseg);
193 	*len += nseg;
194 	return NULL;
195 }
196 
197 static const struct got_error *
198 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
199     struct got_delta_instruction *deltas, int ndeltas,
200     off_t delta_size, off_t base_size)
201 {
202 	const struct got_error *err;
203 	unsigned char buf[16], *bp;
204 	int i, j;
205 	size_t len = 0;
206 	off_t n;
207 	struct got_delta_instruction *d;
208 
209 	m->delta_buf = malloc(delta_size);
210 	if (m->delta_buf == NULL)
211 		return got_error_from_errno("calloc");
212 
213 	/* base object size */
214 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
215 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
216 	for (i = 1; n > 0; i++) {
217 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
218 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
219 		n >>= GOT_DELTA_SIZE_SHIFT;
220 	}
221 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
222 	if (err)
223 		return err;
224 
225 	/* target object size */
226 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
227 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
228 	for (i = 1; n > 0; i++) {
229 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
230 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
231 		n >>= GOT_DELTA_SIZE_SHIFT;
232 	}
233 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
234 	if (err)
235 		return err;
236 
237 	for (j = 0; j < ndeltas; j++) {
238 		d = &deltas[j];
239 		if (d->copy) {
240 			n = d->offset;
241 			bp = &buf[1];
242 			buf[0] = GOT_DELTA_BASE_COPY;
243 			for (i = 0; i < 4; i++) {
244 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
245 				buf[0] |= 1 << i;
246 				*bp++ = n & 0xff;
247 				n >>= 8;
248 				if (n == 0)
249 					break;
250 			}
251 
252 			n = d->len;
253 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
254 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
255 				for (i = 0; i < 3 && n > 0; i++) {
256 					buf[0] |= 1 << (i + 4);
257 					*bp++ = n & 0xff;
258 					n >>= 8;
259 				}
260 			}
261 			err = append(&m->delta_buf, &len, &delta_size,
262 			    buf, bp - buf);
263 			if (err)
264 				return err;
265 		} else if (o->f == NULL) {
266 			n = 0;
267 			while (n != d->len) {
268 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
269 				err = append(&m->delta_buf, &len, &delta_size,
270 				    buf, 1);
271 				if (err)
272 					return err;
273 				err = append(&m->delta_buf, &len, &delta_size,
274 				    o->data + o->hdrlen + d->offset + n,
275 				    buf[0]);
276 				if (err)
277 					return err;
278 				n += buf[0];
279 			}
280 		} else {
281 			char content[128];
282 			size_t r;
283 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
284 				return got_error_from_errno("fseeko");
285 			n = 0;
286 			while (n != d->len) {
287 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
288 				err = append(&m->delta_buf, &len, &delta_size,
289 				    buf, 1);
290 				if (err)
291 					return err;
292 				r = fread(content, 1, buf[0], o->f);
293 				if (r != buf[0])
294 					return got_ferror(o->f, GOT_ERR_IO);
295 				err = append(&m->delta_buf, &len, &delta_size,
296 				    content, buf[0]);
297 				if (err)
298 					return err;
299 				n += buf[0];
300 			}
301 		}
302 	}
303 
304 	m->delta_len = len;
305 	return NULL;
306 }
307 
308 static const struct got_error *
309 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
310     struct got_delta_instruction *deltas, int ndeltas,
311     off_t base_size, FILE *f)
312 {
313 	unsigned char buf[16], *bp;
314 	int i, j;
315 	off_t n;
316 	size_t w;
317 	struct got_delta_instruction *d;
318 
319 	/* base object size */
320 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
321 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
322 	for (i = 1; n > 0; i++) {
323 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
324 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
325 		n >>= GOT_DELTA_SIZE_SHIFT;
326 	}
327 	w = fwrite(buf, 1, i, f);
328 	if (w != i)
329 		return got_ferror(f, GOT_ERR_IO);
330 
331 	/* target object size */
332 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
333 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
334 	for (i = 1; n > 0; i++) {
335 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
336 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
337 		n >>= GOT_DELTA_SIZE_SHIFT;
338 	}
339 	w = fwrite(buf, 1, i, f);
340 	if (w != i)
341 		return got_ferror(f, GOT_ERR_IO);
342 
343 	for (j = 0; j < ndeltas; j++) {
344 		d = &deltas[j];
345 		if (d->copy) {
346 			n = d->offset;
347 			bp = &buf[1];
348 			buf[0] = GOT_DELTA_BASE_COPY;
349 			for (i = 0; i < 4; i++) {
350 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
351 				buf[0] |= 1 << i;
352 				*bp++ = n & 0xff;
353 				n >>= 8;
354 				if (n == 0)
355 					break;
356 			}
357 
358 			n = d->len;
359 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
360 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
361 				for (i = 0; i < 3 && n > 0; i++) {
362 					buf[0] |= 1 << (i + 4);
363 					*bp++ = n & 0xff;
364 					n >>= 8;
365 				}
366 			}
367 			w = fwrite(buf, 1, bp - buf, f);
368 			if (w != bp - buf)
369 				return got_ferror(f, GOT_ERR_IO);
370 		} else if (o->f == NULL) {
371 			n = 0;
372 			while (n != d->len) {
373 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
374 				w = fwrite(buf, 1, 1, f);
375 				if (w != 1)
376 					return got_ferror(f, GOT_ERR_IO);
377 				w = fwrite(o->data + o->hdrlen + d->offset + n,
378 				    1, buf[0], f);
379 				if (w != buf[0])
380 					return got_ferror(f, GOT_ERR_IO);
381 				n += buf[0];
382 			}
383 		} else {
384 			char content[128];
385 			size_t r;
386 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
387 				return got_error_from_errno("fseeko");
388 			n = 0;
389 			while (n != d->len) {
390 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
391 				w = fwrite(buf, 1, 1, f);
392 				if (w != 1)
393 					return got_ferror(f, GOT_ERR_IO);
394 				r = fread(content, 1, buf[0], o->f);
395 				if (r != buf[0])
396 					return got_ferror(o->f, GOT_ERR_IO);
397 				w = fwrite(content, 1, buf[0], f);
398 				if (w != buf[0])
399 					return got_ferror(f, GOT_ERR_IO);
400 				n += buf[0];
401 			}
402 		}
403 	}
404 
405 	m->delta_len = ftello(f) - m->delta_offset;
406 	return NULL;
407 }
408 
409 static const struct got_error *
410 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
411     struct got_ratelimit *rl, off_t packfile_size, int ncommits,
412     int nobj_total, int obj_deltify, int nobj_written)
413 {
414 	const struct got_error *err;
415 	int elapsed;
416 
417 	if (progress_cb == NULL)
418 		return NULL;
419 
420 	err = got_ratelimit_check(&elapsed, rl);
421 	if (err || !elapsed)
422 		return err;
423 
424 	return progress_cb(progress_arg, packfile_size, ncommits,
425 	    nobj_total, obj_deltify, nobj_written);
426 }
427 
428 static const struct got_error *
429 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
430 {
431 	if (v->nmeta == v->metasz){
432 		size_t newsize = 2 * v->metasz;
433 		struct got_pack_meta **new;
434 		new = reallocarray(v->meta, newsize, sizeof(*new));
435 		if (new == NULL)
436 			return got_error_from_errno("reallocarray");
437 		v->meta = new;
438 		v->metasz = newsize; 
439 	}
440 
441 	v->meta[v->nmeta++] = m;
442 	return NULL;
443 }
444 
445 static const struct got_error *
446 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
447     struct got_object_idset *idset, struct got_pack *pack,
448     struct got_packidx *packidx, int delta_cache_fd,
449     struct got_repository *repo)
450 {
451 	const struct got_error *err = NULL;
452 	struct got_pack_meta *base = NULL;
453 	struct got_object_id *base_obj_id = NULL;
454 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
455 	uint64_t base_size, result_size;
456 
457 	if (m->have_reused_delta)
458 		return NULL;
459 
460 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
461 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
462 	    packidx, idx, &m->id, repo);
463 	if (err)
464 		return err;
465 
466 	if (delta_offset + delta_len < delta_offset)
467 		return got_error(GOT_ERR_BAD_PACKFILE);
468 
469 	base = got_object_idset_get(idset, base_obj_id);
470 	if (base == NULL)
471 		goto done;
472 
473 	m->delta_len = delta_len;
474 	m->delta_offset = delta_cache_offset;
475 	m->prev = base;
476 	m->size = result_size;
477 	m->have_reused_delta = 1;
478 	m->reused_delta_offset = delta_offset;
479 	m->base_obj_id = base_obj_id;
480 	base_obj_id = NULL;
481 	err = add_meta(m, v);
482 done:
483 	free(base_obj_id);
484 	return err;
485 }
486 
487 static const struct got_error *
488 find_pack_for_reuse(struct got_packidx **best_packidx,
489     struct got_repository *repo)
490 {
491 	const struct got_error *err;
492 	struct got_pathlist_head packidx_paths;
493 	struct got_pathlist_entry *pe;
494 	const char *best_packidx_path = NULL;
495 	int nobj_max = 0;
496 
497 	TAILQ_INIT(&packidx_paths);
498 	*best_packidx = NULL;
499 
500 	err = got_repo_list_packidx(&packidx_paths, repo);
501 	if (err)
502 		return err;
503 
504 	TAILQ_FOREACH(pe, &packidx_paths, entry) {
505 		const char *path_packidx = pe->path;
506 		struct got_packidx *packidx;
507 		int nobj;
508 
509 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
510 		if (err)
511 			break;
512 
513 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
514 		if (nobj > nobj_max) {
515 			best_packidx_path = path_packidx;
516 			nobj_max = nobj;
517 		}
518 	}
519 
520 	if (best_packidx_path) {
521 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
522 		    repo);
523 	}
524 
525 	TAILQ_FOREACH(pe, &packidx_paths, entry)
526 		free((void *)pe->path);
527 	got_pathlist_free(&packidx_paths);
528 	return err;
529 }
530 
531 struct search_deltas_arg {
532 	struct got_packidx *packidx;
533 	struct got_pack *pack;
534 	struct got_object_idset *idset;
535 	struct got_pack_metavec *v;
536 	int delta_cache_fd;
537 	struct got_repository *repo;
538 	got_pack_progress_cb progress_cb;
539 	void *progress_arg;
540 	struct got_ratelimit *rl;
541 	got_cancel_cb cancel_cb;
542 	void *cancel_arg;
543 	int ncommits;
544 };
545 
546 static const struct got_error *
547 search_delta_for_object(struct got_object_id *id, void *data, void *arg)
548 {
549 	const struct got_error *err;
550 	struct got_pack_meta *m = data;
551 	struct search_deltas_arg *a = arg;
552 	int obj_idx;
553 	struct got_object *obj = NULL;
554 
555 	if (a->cancel_cb) {
556 		err = (*a->cancel_cb)(a->cancel_arg);
557 		if (err)
558 			return err;
559 	}
560 
561 	if (!got_repo_check_packidx_bloom_filter(a->repo,
562 	    a->packidx->path_packidx, id))
563 		return NULL;
564 
565 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
566 	if (obj_idx == -1)
567 		return NULL;
568 
569 	/* TODO:
570 	 * Opening and closing an object just to check its flags
571 	 * is a bit expensive. We could have an imsg which requests
572 	 * plain type/size information for an object without doing
573 	 * work such as traversing the object's entire delta chain
574 	 * to find the base object type, and other such info which
575 	 * we don't really need here.
576 	 */
577 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
578 	    a->packidx, obj_idx, a->repo);
579 	if (err)
580 		return err;
581 
582 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
583 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
584 		    a->delta_cache_fd, a->repo);
585 		if (err)
586 			goto done;
587 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
588 		    0L, a->ncommits, got_object_idset_num_elements(a->idset),
589 		    a->v->nmeta, 0);
590 	}
591 done:
592 	got_object_close(obj);
593 	return err;
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 ncommits, struct got_repository *repo,
599     got_pack_progress_cb progress_cb, void *progress_arg,
600     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
601 {
602 	const struct got_error *err = NULL;
603 	char *path_packfile = NULL;
604 	struct got_packidx *packidx;
605 	struct got_pack *pack;
606 	struct search_deltas_arg sda;
607 
608 	err = find_pack_for_reuse(&packidx, repo);
609 	if (err)
610 		return err;
611 
612 	if (packidx == NULL)
613 		return NULL;
614 
615 	err = got_packidx_get_packfile_path(&path_packfile,
616 	    packidx->path_packidx);
617 	if (err)
618 		return err;
619 
620 	pack = got_repo_get_cached_pack(repo, path_packfile);
621 	if (pack == NULL) {
622 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
623 		if (err)
624 			goto done;
625 	}
626 
627 	sda.packidx = packidx;
628 	sda.pack = pack;
629 	sda.idset = idset;
630 	sda.v = v;
631 	sda.delta_cache_fd = delta_cache_fd;
632 	sda.repo = repo;
633 	sda.progress_cb = progress_cb;
634 	sda.progress_arg = progress_arg;
635 	sda.rl = rl;
636 	sda.cancel_cb = cancel_cb;
637 	sda.cancel_arg = cancel_arg;
638 	sda.ncommits = ncommits;
639 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
640 done:
641 	free(path_packfile);
642 	return err;
643 }
644 
645 static const struct got_error *
646 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncommits,
647     int nreused, FILE *delta_cache, struct got_repository *repo,
648     got_pack_progress_cb progress_cb, void *progress_arg,
649     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
650 {
651 	const struct got_error *err = NULL;
652 	struct got_pack_meta *m = NULL, *base = NULL;
653 	struct got_raw_object *raw = NULL, *base_raw = NULL;
654 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
655 	int i, j, ndeltas, best_ndeltas;
656 	off_t size, best_size;
657 	const int max_base_candidates = 3;
658 	size_t delta_memsize = 0;
659 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
660 	int outfd = -1;
661 
662 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
663 	for (i = 0; i < nmeta; i++) {
664 		if (cancel_cb) {
665 			err = (*cancel_cb)(cancel_arg);
666 			if (err)
667 				break;
668 		}
669 		err = report_progress(progress_cb, progress_arg, rl,
670 		    0L, ncommits, nreused + nmeta, nreused + i, 0);
671 		if (err)
672 			goto done;
673 		m = meta[i];
674 
675 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
676 		    m->obj_type == GOT_OBJ_TYPE_TAG)
677 			continue;
678 
679 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
680 		if (err)
681 			goto done;
682 		m->size = raw->size;
683 
684 		if (raw->f == NULL) {
685 			err = got_deltify_init_mem(&m->dtab, raw->data,
686 			    raw->hdrlen, raw->size + raw->hdrlen);
687 		} else {
688 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
689 			    raw->size + raw->hdrlen);
690 		}
691 		if (err)
692 			goto done;
693 
694 		if (i > max_base_candidates) {
695 			struct got_pack_meta *n = NULL;
696 			n = meta[i - (max_base_candidates + 1)];
697 			got_deltify_free(n->dtab);
698 			n->dtab = NULL;
699 		}
700 
701 		best_size = raw->size;
702 		best_ndeltas = 0;
703 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
704 			if (cancel_cb) {
705 				err = (*cancel_cb)(cancel_arg);
706 				if (err)
707 					goto done;
708 			}
709 			base = meta[j];
710 			/* long chains make unpacking slow, avoid such bases */
711 			if (base->nchain >= 128 ||
712 			    base->obj_type != m->obj_type)
713 				continue;
714 
715 			err = got_object_raw_open(&base_raw, &outfd, repo,
716 			    &base->id);
717 			if (err)
718 				goto done;
719 
720 			if (raw->f == NULL && base_raw->f == NULL) {
721 				err = got_deltify_mem_mem(&deltas, &ndeltas,
722 				    raw->data, raw->hdrlen,
723 				    raw->size + raw->hdrlen,
724 				    base->dtab, base_raw->data,
725 				    base_raw->hdrlen,
726 				    base_raw->size + base_raw->hdrlen);
727 			} else if (raw->f == NULL) {
728 				err = got_deltify_mem_file(&deltas, &ndeltas,
729 				    raw->data, raw->hdrlen,
730 				    raw->size + raw->hdrlen,
731 				    base->dtab, base_raw->f,
732 				    base_raw->hdrlen,
733 				    base_raw->size + base_raw->hdrlen);
734 			} else if (base_raw->f == NULL) {
735 				err = got_deltify_file_mem(&deltas, &ndeltas,
736 				    raw->f, raw->hdrlen,
737 				    raw->size + raw->hdrlen,
738 				    base->dtab, base_raw->data,
739 				    base_raw->hdrlen,
740 				    base_raw->size + base_raw->hdrlen);
741 			} else {
742 				err = got_deltify(&deltas, &ndeltas,
743 				    raw->f, raw->hdrlen,
744 				    raw->size + raw->hdrlen,
745 				    base->dtab, base_raw->f, base_raw->hdrlen,
746 				    base_raw->size + base_raw->hdrlen);
747 			}
748 			got_object_raw_close(base_raw);
749 			base_raw = NULL;
750 			if (err)
751 				goto done;
752 
753 			size = delta_size(deltas, ndeltas);
754 			if (size + 32 < best_size){
755 				/*
756 				 * if we already picked a best delta,
757 				 * replace it.
758 				 */
759 				best_size = size;
760 				free(best_deltas);
761 				best_deltas = deltas;
762 				best_ndeltas = ndeltas;
763 				deltas = NULL;
764 				m->nchain = base->nchain + 1;
765 				m->prev = base;
766 				m->head = base->head;
767 				if (m->head == NULL)
768 					m->head = base;
769 			} else {
770 				free(deltas);
771 				deltas = NULL;
772 				ndeltas = 0;
773 			}
774 		}
775 
776 		if (best_ndeltas > 0) {
777 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
778 			    delta_memsize + best_size <= max_delta_memsize) {
779 				delta_memsize += best_size;
780 				err = encode_delta_in_mem(m, raw, best_deltas,
781 				    best_ndeltas, best_size, m->prev->size);
782 			} else {
783 				m->delta_offset = ftello(delta_cache);
784 				/*
785 				 * TODO:
786 				 * Storing compressed delta data in the delta
787 				 * cache file would probably be more efficient
788 				 * than writing uncompressed delta data here
789 				 * and compressing it while writing the pack
790 				 * file. This would also allow for reusing
791 				 * deltas in their compressed form.
792 				 */
793 				err = encode_delta(m, raw, best_deltas,
794 				    best_ndeltas, m->prev->size, delta_cache);
795 			}
796 			free(best_deltas);
797 			best_deltas = NULL;
798 			best_ndeltas = 0;
799 			if (err)
800 				goto done;
801 		}
802 
803 		got_object_raw_close(raw);
804 		raw = NULL;
805 	}
806 done:
807 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
808 		got_deltify_free(meta[i]->dtab);
809 		meta[i]->dtab = NULL;
810 	}
811 	if (raw)
812 		got_object_raw_close(raw);
813 	if (base_raw)
814 		got_object_raw_close(base_raw);
815 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
816 		err = got_error_from_errno("close");
817 	free(deltas);
818 	free(best_deltas);
819 	return err;
820 }
821 
822 static const struct got_error *
823 search_packidx(int *found, struct got_object_id *id,
824     struct got_repository *repo)
825 {
826 	const struct got_error *err = NULL;
827 	struct got_packidx *packidx = NULL;
828 	int idx;
829 
830 	*found = 0;
831 
832 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
833 	if (err == NULL)
834 		*found = 1; /* object is already packed */
835 	else if (err->code == GOT_ERR_NO_OBJ)
836 		err = NULL;
837 	return err;
838 }
839 
840 static const int obj_types[] = {
841 	GOT_OBJ_TYPE_ANY,
842 	GOT_OBJ_TYPE_COMMIT,
843 	GOT_OBJ_TYPE_TREE,
844 	GOT_OBJ_TYPE_BLOB,
845 	GOT_OBJ_TYPE_TAG,
846 	GOT_OBJ_TYPE_OFFSET_DELTA,
847 	GOT_OBJ_TYPE_REF_DELTA
848 };
849 
850 static const struct got_error *
851 add_object(int want_meta, struct got_object_idset *idset,
852     struct got_object_id *id, const char *path, int obj_type,
853     time_t mtime, int loose_obj_only, struct got_repository *repo)
854 {
855 	const struct got_error *err;
856 	struct got_pack_meta *m = NULL;
857 
858 	if (loose_obj_only) {
859 		int is_packed;
860 		err = search_packidx(&is_packed, id, repo);
861 		if (err)
862 			return err;
863 		if (is_packed)
864 			return NULL;
865 	}
866 
867 	if (want_meta) {
868 		err = alloc_meta(&m, id, path, obj_type, mtime);
869 		if (err)
870 			return err;
871 	}
872 
873 	return got_object_idset_add(idset, id, m);
874 }
875 
876 static const struct got_error *
877 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
878     struct got_object_idset *idset, struct got_object_id *tree_id,
879     const char *dpath, time_t mtime, struct got_repository *repo,
880     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
881 {
882 	const struct got_error *err;
883 	struct got_tree_object *tree;
884 	char *p = NULL;
885 	int i;
886 
887 	err = got_object_open_as_tree(&tree, repo, tree_id);
888 	if (err)
889 		return err;
890 
891 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
892 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
893 		struct got_object_id *id = got_tree_entry_get_id(e);
894 		mode_t mode = got_tree_entry_get_mode(e);
895 
896 		if (cancel_cb) {
897 			err = (*cancel_cb)(cancel_arg);
898 			if (err)
899 				break;
900 		}
901 
902 		if (got_object_tree_entry_is_submodule(e) ||
903 		    got_object_idset_contains(idset, id))
904 			continue;
905 		
906 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
907 		    got_tree_entry_get_name(e)) == -1) {
908 			err = got_error_from_errno("asprintf");
909 			break;
910 		}
911 
912 		if (S_ISDIR(mode)) {
913 			struct got_object_qid *qid;
914 			err = got_object_qid_alloc(&qid, id);
915 			if (err)
916 				break;
917 			STAILQ_INSERT_TAIL(ids, qid, entry);
918 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
919 			err = add_object(want_meta, idset, id, p,
920 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo);
921 			if (err)
922 				break;
923 		}
924 		free(p);
925 		p = NULL;
926 	}
927 
928 	got_object_tree_close(tree);
929 	free(p);
930 	return err;
931 }
932 
933 static const struct got_error *
934 load_tree(int want_meta, struct got_object_idset *idset,
935     struct got_object_id *tree_id, const char *dpath, time_t mtime,
936     int loose_obj_only, struct got_repository *repo,
937     got_cancel_cb cancel_cb, void *cancel_arg)
938 {
939 	const struct got_error *err = NULL;
940 	struct got_object_id_queue tree_ids;
941 	struct got_object_qid *qid;
942 
943 	if (got_object_idset_contains(idset, tree_id))
944 		return NULL;
945 
946 	err = got_object_qid_alloc(&qid, tree_id);
947 	if (err)
948 		return err;
949 
950 	STAILQ_INIT(&tree_ids);
951 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
952 
953 	while (!STAILQ_EMPTY(&tree_ids)) {
954 		if (cancel_cb) {
955 			err = (*cancel_cb)(cancel_arg);
956 			if (err)
957 				break;
958 		}
959 
960 		qid = STAILQ_FIRST(&tree_ids);
961 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
962 
963 		if (got_object_idset_contains(idset, qid->id)) {
964 			got_object_qid_free(qid);
965 			continue;
966 		}
967 
968 		err = add_object(want_meta, idset, qid->id, dpath,
969 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo);
970 		if (err) {
971 			got_object_qid_free(qid);
972 			break;
973 		}
974 
975 		err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
976 		    dpath, mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
977 		got_object_qid_free(qid);
978 		if (err)
979 			break;
980 	}
981 
982 	got_object_id_queue_free(&tree_ids);
983 	return err;
984 }
985 
986 static const struct got_error *
987 load_commit(int want_meta, struct got_object_idset *idset,
988     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
989     got_cancel_cb cancel_cb, void *cancel_arg)
990 {
991 	const struct got_error *err;
992 	struct got_commit_object *commit;
993 
994 	if (got_object_idset_contains(idset, id))
995 		return NULL;
996 
997 	if (loose_obj_only) {
998 		int is_packed;
999 		err = search_packidx(&is_packed, id, repo);
1000 		if (err)
1001 			return err;
1002 		if (is_packed)
1003 			return NULL;
1004 	}
1005 
1006 	err = got_object_open_as_commit(&commit, repo, id);
1007 	if (err)
1008 		return err;
1009 
1010 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1011 	    got_object_commit_get_committer_time(commit),
1012 	    loose_obj_only, repo);
1013 	if (err)
1014 		goto done;
1015 
1016 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1017 	    "", got_object_commit_get_committer_time(commit),
1018 	    loose_obj_only, repo, cancel_cb, cancel_arg);
1019 done:
1020 	got_object_commit_close(commit);
1021 	return err;
1022 }
1023 
1024 static const struct got_error *
1025 load_tag(int want_meta, struct got_object_idset *idset,
1026     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1027     got_cancel_cb cancel_cb, void *cancel_arg)
1028 {
1029 	const struct got_error *err;
1030 	struct got_tag_object *tag = NULL;
1031 
1032 	if (got_object_idset_contains(idset, id))
1033 		return NULL;
1034 
1035 	if (loose_obj_only) {
1036 		int is_packed;
1037 		err = search_packidx(&is_packed, id, repo);
1038 		if (err)
1039 			return err;
1040 		if (is_packed)
1041 			return NULL;
1042 	}
1043 
1044 	err = got_object_open_as_tag(&tag, repo, id);
1045 	if (err)
1046 		return err;
1047 
1048 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1049 	    got_object_tag_get_tagger_time(tag),
1050 	    loose_obj_only, repo);
1051 	if (err)
1052 		goto done;
1053 
1054 	switch (got_object_tag_get_object_type(tag)) {
1055 	case GOT_OBJ_TYPE_COMMIT:
1056 		err = load_commit(want_meta, idset,
1057 		    got_object_tag_get_object_id(tag), repo,
1058 		    loose_obj_only, cancel_cb, cancel_arg);
1059 		break;
1060 	case GOT_OBJ_TYPE_TREE:
1061 		err = load_tree(want_meta, idset,
1062 		    got_object_tag_get_object_id(tag), "",
1063 		    got_object_tag_get_tagger_time(tag),
1064 		    loose_obj_only, repo, cancel_cb, cancel_arg);
1065 		break;
1066 	default:
1067 		break;
1068 	}
1069 
1070 done:
1071 	got_object_tag_close(tag);
1072 	return err;
1073 }
1074 
1075 enum findtwixt_color {
1076 	COLOR_KEEP = 0,
1077 	COLOR_DROP,
1078 	COLOR_BLANK,
1079 };
1080 static const int findtwixt_colors[] = {
1081 	COLOR_KEEP,
1082 	COLOR_DROP,
1083 	COLOR_BLANK
1084 };
1085 
1086 static const struct got_error *
1087 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1088     int color, struct got_repository *repo)
1089 {
1090 	const struct got_error *err;
1091 	struct got_object_qid *qid;
1092 
1093 	err = got_object_qid_alloc(&qid, id);
1094 	if (err)
1095 		return err;
1096 
1097 	STAILQ_INSERT_TAIL(ids, qid, entry);
1098 	qid->data = (void *)&findtwixt_colors[color];
1099 	return NULL;
1100 }
1101 
1102 static const struct got_error *
1103 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1104     struct got_object_id *id, struct got_repository *repo,
1105     got_cancel_cb cancel_cb, void *cancel_arg)
1106 {
1107 	const struct got_error *err = NULL;
1108 	struct got_commit_object *commit;
1109 	const struct got_object_id_queue *parents;
1110 	struct got_object_id_queue ids;
1111 	struct got_object_qid *qid;
1112 
1113 	STAILQ_INIT(&ids);
1114 
1115 	err = got_object_qid_alloc(&qid, id);
1116 	if (err)
1117 		return err;
1118 	STAILQ_INSERT_HEAD(&ids, qid, entry);
1119 
1120 	while (!STAILQ_EMPTY(&ids)) {
1121 		if (cancel_cb) {
1122 			err = (*cancel_cb)(cancel_arg);
1123 			if (err)
1124 				break;
1125 		}
1126 
1127 		qid = STAILQ_FIRST(&ids);
1128 		STAILQ_REMOVE_HEAD(&ids, entry);
1129 
1130 		if (got_object_idset_contains(drop, qid->id)) {
1131 			got_object_qid_free(qid);
1132 			continue;
1133 		}
1134 
1135 		err = got_object_idset_add(drop, qid->id,
1136 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1137 		if (err) {
1138 			got_object_qid_free(qid);
1139 			break;
1140 		}
1141 
1142 		if (!got_object_idset_contains(keep, qid->id)) {
1143 			got_object_qid_free(qid);
1144 			continue;
1145 		}
1146 
1147 		err = got_object_open_as_commit(&commit, repo, qid->id);
1148 		got_object_qid_free(qid);
1149 		if (err)
1150 			break;
1151 
1152 		parents = got_object_commit_get_parent_ids(commit);
1153 		if (parents) {
1154 			err = got_object_id_queue_copy(parents, &ids);
1155 			if (err) {
1156 				got_object_commit_close(commit);
1157 				break;
1158 			}
1159 		}
1160 		got_object_commit_close(commit);
1161 	}
1162 
1163 	got_object_id_queue_free(&ids);
1164 	return err;
1165 }
1166 
1167 struct append_id_arg {
1168 	struct got_object_id **array;
1169 	int idx;
1170 };
1171 
1172 static const struct got_error *
1173 append_id(struct got_object_id *id, void *data, void *arg)
1174 {
1175 	struct append_id_arg *a = arg;
1176 
1177 	a->array[a->idx] = got_object_id_dup(id);
1178 	if (a->array[a->idx] == NULL)
1179 		return got_error_from_errno("got_object_id_dup");
1180 
1181 	a->idx++;
1182 	return NULL;
1183 }
1184 
1185 static const struct got_error *
1186 findtwixt(struct got_object_id ***res, int *nres,
1187     struct got_object_id **head, int nhead,
1188     struct got_object_id **tail, int ntail,
1189     struct got_repository *repo,
1190     got_cancel_cb cancel_cb, void *cancel_arg)
1191 {
1192 	const struct got_error *err = NULL;
1193 	struct got_object_id_queue ids;
1194 	struct got_object_idset *keep, *drop;
1195 	struct got_object_qid *qid;
1196 	int i, ncolor, nkeep, obj_type;
1197 
1198 	STAILQ_INIT(&ids);
1199 	*res = NULL;
1200 	*nres = 0;
1201 
1202 	keep = got_object_idset_alloc();
1203 	if (keep == NULL)
1204 		return got_error_from_errno("got_object_idset_alloc");
1205 
1206 	drop = got_object_idset_alloc();
1207 	if (drop == NULL) {
1208 		err = got_error_from_errno("got_object_idset_alloc");
1209 		goto done;
1210 	}
1211 
1212 	for (i = 0; i < nhead; i++) {
1213 		struct got_object_id *id = head[i];
1214 		if (id == NULL)
1215 			continue;
1216 		err = got_object_get_type(&obj_type, repo, id);
1217 		if (err)
1218 			return err;
1219 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1220 			continue;
1221 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1222 		if (err)
1223 			goto done;
1224 	}		
1225 	for (i = 0; i < ntail; i++) {
1226 		struct got_object_id *id = tail[i];
1227 		if (id == NULL)
1228 			continue;
1229 		err = got_object_get_type(&obj_type, repo, id);
1230 		if (err)
1231 			return err;
1232 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1233 			continue;
1234 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1235 		if (err)
1236 			goto done;
1237 	}
1238 
1239 	while (!STAILQ_EMPTY(&ids)) {
1240 		int qcolor;
1241 		qid = STAILQ_FIRST(&ids);
1242 		qcolor = *((int *)qid->data);
1243 
1244 		if (got_object_idset_contains(drop, qid->id))
1245 			ncolor = COLOR_DROP;
1246 		else if (got_object_idset_contains(keep, qid->id))
1247 			ncolor = COLOR_KEEP;
1248 		else
1249 			ncolor = COLOR_BLANK;
1250 
1251 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1252 		    qcolor == COLOR_KEEP)) {
1253 			STAILQ_REMOVE_HEAD(&ids, entry);
1254 			got_object_qid_free(qid);
1255 			continue;
1256 		}
1257 
1258 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1259 			err = drop_commit(keep, drop, qid->id, repo,
1260 			    cancel_cb, cancel_arg);
1261 			if (err)
1262 				goto done;
1263 		} else if (ncolor == COLOR_BLANK) {
1264 			struct got_commit_object *commit;
1265 			struct got_object_id *id;
1266 			const struct got_object_id_queue *parents;
1267 			struct got_object_qid *pid;
1268 
1269 			id = got_object_id_dup(qid->id);
1270 			if (id == NULL) {
1271 				err = got_error_from_errno("got_object_id_dup");
1272 				goto done;
1273 			}
1274 			if (qcolor == COLOR_KEEP)
1275 				err = got_object_idset_add(keep, id,
1276 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1277 			else
1278 				err = got_object_idset_add(drop, id,
1279 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1280 			if (err) {
1281 				free(id);
1282 				goto done;
1283 			}
1284 
1285 			err = got_object_open_as_commit(&commit, repo, id);
1286 			if (err) {
1287 				free(id);
1288 				goto done;
1289 			}
1290 			parents = got_object_commit_get_parent_ids(commit);
1291 			if (parents) {
1292 				STAILQ_FOREACH(pid, parents, entry) {
1293 					err = queue_commit_id(&ids, pid->id,
1294 					    qcolor, repo);
1295 					if (err) {
1296 						free(id);
1297 						goto done;
1298 					}
1299 				}
1300 			}
1301 			got_object_commit_close(commit);
1302 			commit = NULL;
1303 		} else {
1304 			/* should not happen */
1305 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1306 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1307 			goto done;
1308 		}
1309 
1310 		STAILQ_REMOVE_HEAD(&ids, entry);
1311 		got_object_qid_free(qid);
1312 	}
1313 
1314 	nkeep = got_object_idset_num_elements(keep);
1315 	if (nkeep > 0) {
1316 		struct append_id_arg arg;
1317 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1318 		if (arg.array == NULL) {
1319 			err = got_error_from_errno("calloc");
1320 			goto done;
1321 		}
1322 		arg.idx = 0;
1323 		err = got_object_idset_for_each(keep, append_id, &arg);
1324 		if (err) {
1325 			free(arg.array);
1326 			goto done;
1327 		}
1328 		*res = arg.array;
1329 		*nres = nkeep;
1330 	}
1331 done:
1332 	got_object_idset_free(keep);
1333 	got_object_idset_free(drop);
1334 	got_object_id_queue_free(&ids);
1335 	return err;
1336 }
1337 
1338 static const struct got_error *
1339 load_object_ids(struct got_object_idset *idset,
1340     struct got_object_id **theirs, int ntheirs,
1341     struct got_object_id **ours, int nours, struct got_repository *repo,
1342     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
1343 {
1344 	const struct got_error *err = NULL;
1345 	struct got_object_id **ids = NULL;
1346 	int i, nobj = 0, obj_type;
1347 
1348 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
1349 	    cancel_cb, cancel_arg);
1350 	if (err || nobj == 0)
1351 		goto done;
1352 
1353 	for (i = 0; i < ntheirs; i++) {
1354 		struct got_object_id *id = theirs[i];
1355 		if (id == NULL)
1356 			continue;
1357 		err = got_object_get_type(&obj_type, repo, id);
1358 		if (err)
1359 			return err;
1360 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1361 			continue;
1362 		err = load_commit(0, idset, id, repo,
1363 		    loose_obj_only, cancel_cb, cancel_arg);
1364 		if (err)
1365 			goto done;
1366 	}
1367 
1368 	for (i = 0; i < ntheirs; i++) {
1369 		struct got_object_id *id = theirs[i];
1370 		struct got_pack_meta *m;
1371 		if (id == NULL)
1372 			continue;
1373 		m = got_object_idset_get(idset, id);
1374 		if (m == NULL) {
1375 			err = got_object_get_type(&obj_type, repo, id);
1376 			if (err)
1377 				goto done;
1378 		} else
1379 			obj_type = m->obj_type;
1380 		if (obj_type != GOT_OBJ_TYPE_TAG)
1381 			continue;
1382 		err = load_tag(0, idset, id, repo,
1383 		    loose_obj_only, cancel_cb, cancel_arg);
1384 		if (err)
1385 			goto done;
1386 	}
1387 
1388 	for (i = 0; i < nobj; i++) {
1389 		err = load_commit(1, idset, ids[i], repo,
1390 		    loose_obj_only, cancel_cb, cancel_arg);
1391 		if (err)
1392 			goto done;
1393 	}
1394 
1395 	for (i = 0; i < nours; i++) {
1396 		struct got_object_id *id = ours[i];
1397 		struct got_pack_meta *m;
1398 		if (id == NULL)
1399 			continue;
1400 		m = got_object_idset_get(idset, id);
1401 		if (m == NULL) {
1402 			err = got_object_get_type(&obj_type, repo, id);
1403 			if (err)
1404 				goto done;
1405 		} else
1406 			obj_type = m->obj_type;
1407 		if (obj_type != GOT_OBJ_TYPE_TAG)
1408 			continue;
1409 		err = load_tag(1, idset, id, repo,
1410 		    loose_obj_only, cancel_cb, cancel_arg);
1411 		if (err)
1412 			goto done;
1413 	}
1414 done:
1415 	for (i = 0; i < nobj; i++) {
1416 		free(ids[i]);
1417 	}
1418 	free(ids);
1419 	return err;
1420 }
1421 
1422 const struct got_error *
1423 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1424 {
1425 	size_t n;
1426 
1427 	SHA1Update(ctx, buf, len);
1428 	n = fwrite(buf, 1, len, f);
1429 	if (n != len)
1430 		return got_ferror(f, GOT_ERR_IO);
1431 	return NULL;
1432 }
1433 
1434 static void
1435 putbe32(char *b, uint32_t n)
1436 {
1437 	b[0] = n >> 24;
1438 	b[1] = n >> 16;
1439 	b[2] = n >> 8;
1440 	b[3] = n >> 0;
1441 }
1442 
1443 static int
1444 write_order_cmp(const void *pa, const void *pb)
1445 {
1446 	struct got_pack_meta *a, *b, *ahd, *bhd;
1447 
1448 	a = *(struct got_pack_meta **)pa;
1449 	b = *(struct got_pack_meta **)pb;
1450 	ahd = (a->head == NULL) ? a : a->head;
1451 	bhd = (b->head == NULL) ? b : b->head;
1452 	if (ahd->mtime != bhd->mtime)
1453 		return bhd->mtime - ahd->mtime;
1454 	if (ahd != bhd)
1455 		return (uintptr_t)bhd - (uintptr_t)ahd;
1456 	if (a->nchain != b->nchain)
1457 		return a->nchain - b->nchain;
1458 	return a->mtime - b->mtime;
1459 }
1460 
1461 static int
1462 reuse_write_order_cmp(const void *pa, const void *pb)
1463 {
1464 	struct got_pack_meta *a, *b;
1465 
1466 	a = *(struct got_pack_meta **)pa;
1467 	b = *(struct got_pack_meta **)pb;
1468 
1469 	if (a->reused_delta_offset < b->reused_delta_offset)
1470 		return -1;
1471 	if (a->reused_delta_offset > b->reused_delta_offset)
1472 		return 1;
1473 	return 0;
1474 }
1475 
1476 static const struct got_error *
1477 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1478 {
1479 	size_t i;
1480 
1481 	*hdrlen = 0;
1482 
1483 	hdr[0] = obj_type << 4;
1484 	hdr[0] |= len & 0xf;
1485 	len >>= 4;
1486 	for (i = 1; len != 0; i++){
1487 		if (i >= bufsize)
1488 			return got_error(GOT_ERR_NO_SPACE);
1489 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1490 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1491 		len >>= GOT_DELTA_SIZE_SHIFT;
1492 	}
1493 
1494 	*hdrlen = i;
1495 	return NULL;
1496 }
1497 
1498 static int
1499 packoff(char *hdr, off_t off)
1500 {
1501 	int i, j;
1502 	char rbuf[8];
1503 
1504 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1505 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1506 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1507 		    GOT_DELTA_SIZE_MORE;
1508 	}
1509 
1510 	j = 0;
1511 	while (i > 0)
1512 		hdr[j++] = rbuf[--i];
1513 	return j;
1514 }
1515 
1516 static const struct got_error *
1517 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1518     struct got_pack_meta *m)
1519 {
1520 	const struct got_error *err;
1521 	char buf[32];
1522 	int nh;
1523 
1524 	if (m->prev->off != 0) {
1525 		err = packhdr(&nh, buf, sizeof(buf),
1526 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1527 		if (err)
1528 			return err;
1529 		nh += packoff(buf + nh, m->off - m->prev->off);
1530 		err = hwrite(packfile, buf, nh, ctx);
1531 		if (err)
1532 			return err;
1533 		*packfile_size += nh;
1534 	} else {
1535 		err = packhdr(&nh, buf, sizeof(buf),
1536 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1537 		if (err)
1538 			return err;
1539 		err = hwrite(packfile, buf, nh, ctx);
1540 		if (err)
1541 			return err;
1542 		*packfile_size += nh;
1543 		err = hwrite(packfile, m->prev->id.sha1,
1544 		    sizeof(m->prev->id.sha1), ctx);
1545 		if (err)
1546 			return err;
1547 		*packfile_size += sizeof(m->prev->id.sha1);
1548 	}
1549 
1550 	return NULL;
1551 }
1552 
1553 static const struct got_error *
1554 write_packed_object(off_t *packfile_size, FILE *packfile,
1555     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1556     SHA1_CTX *ctx, struct got_repository *repo)
1557 {
1558 	const struct got_error *err = NULL;
1559 	struct got_deflate_checksum csum;
1560 	char buf[32];
1561 	int nh;
1562 	struct got_raw_object *raw = NULL;
1563 	off_t outlen;
1564 
1565 	csum.output_sha1 = ctx;
1566 	csum.output_crc = NULL;
1567 
1568 	m->off = ftello(packfile);
1569 	if (m->delta_len == 0) {
1570 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1571 		if (err)
1572 			goto done;
1573 		err = packhdr(&nh, buf, sizeof(buf),
1574 		    m->obj_type, raw->size);
1575 		if (err)
1576 			goto done;
1577 		err = hwrite(packfile, buf, nh, ctx);
1578 		if (err)
1579 			goto done;
1580 		*packfile_size += nh;
1581 		if (raw->f == NULL) {
1582 			err = got_deflate_to_file_mmap(&outlen,
1583 			    raw->data + raw->hdrlen, 0, raw->size,
1584 			    packfile, &csum);
1585 			if (err)
1586 				goto done;
1587 		} else {
1588 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1589 			    == -1) {
1590 				err = got_error_from_errno("fseeko");
1591 				goto done;
1592 			}
1593 			err = got_deflate_to_file(&outlen, raw->f,
1594 			    raw->size, packfile, &csum);
1595 			if (err)
1596 				goto done;
1597 		}
1598 		*packfile_size += outlen;
1599 		got_object_raw_close(raw);
1600 		raw = NULL;
1601 	} else if (m->delta_buf) {
1602 		err = deltahdr(packfile_size, ctx, packfile, m);
1603 		if (err)
1604 			goto done;
1605 		err = got_deflate_to_file_mmap(&outlen,
1606 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1607 		if (err)
1608 			goto done;
1609 		*packfile_size += outlen;
1610 		free(m->delta_buf);
1611 		m->delta_buf = NULL;
1612 	} else {
1613 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1614 		    == -1) {
1615 			err = got_error_from_errno("fseeko");
1616 			goto done;
1617 		}
1618 		err = deltahdr(packfile_size, ctx, packfile, m);
1619 		if (err)
1620 			goto done;
1621 		err = got_deflate_to_file(&outlen, delta_cache,
1622 		    m->delta_len, packfile, &csum);
1623 		if (err)
1624 			goto done;
1625 		*packfile_size += outlen;
1626 	}
1627 done:
1628 	if (raw)
1629 		got_object_raw_close(raw);
1630 	return err;
1631 }
1632 
1633 static const struct got_error *
1634 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1635     struct got_pack_meta **deltify, int ndeltify,
1636     struct got_pack_meta **reuse, int nreuse,
1637     int nours, struct got_repository *repo,
1638     got_pack_progress_cb progress_cb, void *progress_arg,
1639     struct got_ratelimit *rl,
1640     got_cancel_cb cancel_cb, void *cancel_arg)
1641 {
1642 	const struct got_error *err = NULL;
1643 	int i;
1644 	SHA1_CTX ctx;
1645 	struct got_pack_meta *m;
1646 	char buf[32];
1647 	size_t n;
1648 	off_t packfile_size = 0;
1649 	int outfd = -1;
1650 
1651 	SHA1Init(&ctx);
1652 
1653 	err = hwrite(packfile, "PACK", 4, &ctx);
1654 	if (err)
1655 		return err;
1656 	putbe32(buf, GOT_PACKFILE_VERSION);
1657 	err = hwrite(packfile, buf, 4, &ctx);
1658 	if (err)
1659 		goto done;
1660 	putbe32(buf, ndeltify + nreuse);
1661 	err = hwrite(packfile, buf, 4, &ctx);
1662 	if (err)
1663 		goto done;
1664 
1665 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1666 	    write_order_cmp);
1667 	for (i = 0; i < ndeltify; i++) {
1668 		err = report_progress(progress_cb, progress_arg, rl,
1669 		    packfile_size, nours, ndeltify + nreuse,
1670 		    ndeltify + nreuse, i);
1671 		if (err)
1672 			goto done;
1673 		m = deltify[i];
1674 		err = write_packed_object(&packfile_size, packfile,
1675 		    delta_cache, m, &outfd, &ctx, repo);
1676 		if (err)
1677 			goto done;
1678 	}
1679 
1680 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1681 	    reuse_write_order_cmp);
1682 	for (i = 0; i < nreuse; i++) {
1683 		err = report_progress(progress_cb, progress_arg, rl,
1684 		    packfile_size, nours, ndeltify + nreuse,
1685 		    ndeltify + nreuse, ndeltify + i);
1686 		if (err)
1687 			goto done;
1688 		m = reuse[i];
1689 		err = write_packed_object(&packfile_size, packfile,
1690 		    delta_cache, m, &outfd, &ctx, repo);
1691 		if (err)
1692 			goto done;
1693 	}
1694 
1695 	SHA1Final(pack_sha1, &ctx);
1696 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1697 	if (n != SHA1_DIGEST_LENGTH)
1698 		err = got_ferror(packfile, GOT_ERR_IO);
1699 	packfile_size += SHA1_DIGEST_LENGTH;
1700 	packfile_size += sizeof(struct got_packfile_hdr);
1701 	if (progress_cb) {
1702 		err = progress_cb(progress_arg, packfile_size, nours,
1703 		    ndeltify + nreuse, ndeltify + nreuse,
1704 		    ndeltify + nreuse);
1705 		if (err)
1706 			goto done;
1707 	}
1708 done:
1709 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1710 		err = got_error_from_errno("close");
1711 	return err;
1712 }
1713 
1714 static const struct got_error *
1715 remove_unused_object(struct got_object_idset_element *entry, void *arg)
1716 {
1717 	struct got_object_idset *idset = arg;
1718 
1719 	if (got_object_idset_get_element_data(entry) == NULL)
1720 		got_object_idset_remove_element(idset, entry);
1721 
1722 	return NULL;
1723 }
1724 
1725 static const struct got_error *
1726 remove_reused_object(struct got_object_idset_element *entry, void *arg)
1727 {
1728 	struct got_object_idset *idset = arg;
1729 	struct got_pack_meta *m;
1730 
1731 	m = got_object_idset_get_element_data(entry);
1732 	if (m->have_reused_delta)
1733 		got_object_idset_remove_element(idset, entry);
1734 
1735 	return NULL;
1736 }
1737 
1738 static const struct got_error *
1739 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1740 {
1741 	struct got_pack_meta *m = data;
1742 	struct got_pack_metavec *v = arg;
1743 
1744 	return add_meta(m, v);
1745 }
1746 
1747 const struct got_error *
1748 got_pack_create(uint8_t *packsha1, FILE *packfile,
1749     struct got_object_id **theirs, int ntheirs,
1750     struct got_object_id **ours, int nours,
1751     struct got_repository *repo, int loose_obj_only, int allow_empty,
1752     got_pack_progress_cb progress_cb, void *progress_arg,
1753     got_cancel_cb cancel_cb, void *cancel_arg)
1754 {
1755 	const struct got_error *err;
1756 	int delta_cache_fd = -1;
1757 	FILE *delta_cache = NULL;
1758 	struct got_object_idset *idset;
1759 	struct got_ratelimit rl;
1760 	struct got_pack_metavec deltify, reuse;
1761 
1762 	memset(&deltify, 0, sizeof(deltify));
1763 	memset(&reuse, 0, sizeof(reuse));
1764 
1765 	got_ratelimit_init(&rl, 0, 500);
1766 
1767 	idset = got_object_idset_alloc();
1768 	if (idset == NULL)
1769 		return got_error_from_errno("got_object_idset_alloc");
1770 
1771 	err = load_object_ids(idset, theirs, ntheirs, ours, nours,
1772 	    repo, loose_obj_only, cancel_cb, cancel_arg);
1773 	if (err)
1774 		return err;
1775 
1776 	err = got_object_idset_for_each_element(idset,
1777 	    remove_unused_object, idset);
1778 	if (err)
1779 		goto done;
1780 
1781 	if (progress_cb) {
1782 		err = progress_cb(progress_arg, 0L, nours,
1783 		    got_object_idset_num_elements(idset), 0, 0);
1784 		if (err)
1785 			goto done;
1786 	}
1787 
1788 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1789 		err = got_error(GOT_ERR_CANNOT_PACK);
1790 		goto done;
1791 	}
1792 
1793 	delta_cache_fd = got_opentempfd();
1794 	if (delta_cache_fd == -1) {
1795 		err = got_error_from_errno("got_opentemp");
1796 		goto done;
1797 	}
1798 
1799 	reuse.metasz = 64;
1800 	reuse.meta = calloc(reuse.metasz,
1801 	    sizeof(struct got_pack_meta *));
1802 	if (reuse.meta == NULL) {
1803 		err = got_error_from_errno("calloc");
1804 		goto done;
1805 	}
1806 
1807 	err = search_deltas(&reuse, idset, delta_cache_fd, nours, repo,
1808 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1809 	if (err)
1810 		goto done;
1811 	if (reuse.nmeta > 0) {
1812 		err = got_object_idset_for_each_element(idset,
1813 		    remove_reused_object, idset);
1814 		if (err)
1815 			goto done;
1816 	}
1817 
1818 	delta_cache = fdopen(delta_cache_fd, "a+");
1819 	if (delta_cache == NULL) {
1820 		err = got_error_from_errno("fdopen");
1821 		goto done;
1822 	}
1823 	delta_cache_fd = -1;
1824 
1825 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1826 		err = got_error_from_errno("fseeko");
1827 		goto done;
1828 	}
1829 
1830 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1831 	    sizeof(struct got_pack_meta *));
1832 	if (deltify.meta == NULL) {
1833 		err = got_error_from_errno("calloc");
1834 		goto done;
1835 	}
1836 	deltify.metasz = got_object_idset_num_elements(idset);
1837 
1838 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1839 	if (err)
1840 		goto done;
1841 	if (deltify.nmeta > 0) {
1842 		err = pick_deltas(deltify.meta, deltify.nmeta, nours,
1843 		    reuse.nmeta, delta_cache, repo,
1844 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1845 		if (err)
1846 			goto done;
1847 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1848 			err = got_error_from_errno("fseeko");
1849 			goto done;
1850 		}
1851 	}
1852 
1853 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1854 	    deltify.nmeta, reuse.meta, reuse.nmeta, nours, repo,
1855 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1856 	if (err)
1857 		goto done;
1858 done:
1859 	free_nmeta(deltify.meta, deltify.nmeta);
1860 	free_nmeta(reuse.meta, reuse.nmeta);
1861 	got_object_idset_free(idset);
1862 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1863 		err = got_error_from_errno("close");
1864 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1865 		err = got_error_from_errno("fclose");
1866 	return err;
1867 }