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 
23 #include <stdint.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <time.h>
28 #include <unistd.h>
29 #include <limits.h>
30 #include <zlib.h>
31 
32 #if defined(__FreeBSD__)
33 #include <unistd.h>
34 #endif
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, int ncolored, int nfound, int ntrees,
412     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
413     int nobj_written)
414 {
415 	const struct got_error *err;
416 	int elapsed;
417 
418 	if (progress_cb == NULL)
419 		return NULL;
420 
421 	err = got_ratelimit_check(&elapsed, rl);
422 	if (err || !elapsed)
423 		return err;
424 
425 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
426 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
427 }
428 
429 static const struct got_error *
430 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
431 {
432 	if (v->nmeta == v->metasz){
433 		size_t newsize = 2 * v->metasz;
434 		struct got_pack_meta **new;
435 		new = reallocarray(v->meta, newsize, sizeof(*new));
436 		if (new == NULL)
437 			return got_error_from_errno("reallocarray");
438 		v->meta = new;
439 		v->metasz = newsize; 
440 	}
441 
442 	v->meta[v->nmeta++] = m;
443 	return NULL;
444 }
445 
446 static const struct got_error *
447 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
448     struct got_object_idset *idset, struct got_pack *pack,
449     struct got_packidx *packidx, int delta_cache_fd,
450     struct got_repository *repo)
451 {
452 	const struct got_error *err = NULL;
453 	struct got_pack_meta *base = NULL;
454 	struct got_object_id *base_obj_id = NULL;
455 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
456 	uint64_t base_size, result_size;
457 
458 	if (m->have_reused_delta)
459 		return NULL;
460 
461 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
462 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
463 	    packidx, idx, &m->id, repo);
464 	if (err)
465 		return err;
466 
467 	if (delta_offset + delta_len < delta_offset)
468 		return got_error(GOT_ERR_BAD_PACKFILE);
469 
470 	base = got_object_idset_get(idset, base_obj_id);
471 	if (base == NULL)
472 		goto done;
473 
474 	m->delta_len = delta_len;
475 	m->delta_offset = delta_cache_offset;
476 	m->prev = base;
477 	m->size = result_size;
478 	m->have_reused_delta = 1;
479 	m->reused_delta_offset = delta_offset;
480 	m->base_obj_id = base_obj_id;
481 	base_obj_id = NULL;
482 	err = add_meta(m, v);
483 done:
484 	free(base_obj_id);
485 	return err;
486 }
487 
488 static const struct got_error *
489 find_pack_for_reuse(struct got_packidx **best_packidx,
490     struct got_repository *repo)
491 {
492 	const struct got_error *err = NULL;
493 	struct got_pathlist_entry *pe;
494 	const char *best_packidx_path = NULL;
495 	int nobj_max = 0;
496 
497 	*best_packidx = NULL;
498 
499 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
500 		const char *path_packidx = pe->path;
501 		struct got_packidx *packidx;
502 		int nobj;
503 
504 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
505 		if (err)
506 			break;
507 
508 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
509 		if (nobj > nobj_max) {
510 			best_packidx_path = path_packidx;
511 			nobj_max = nobj;
512 		}
513 	}
514 
515 	if (best_packidx_path) {
516 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
517 		    repo);
518 	}
519 
520 	return err;
521 }
522 
523 struct search_deltas_arg {
524 	struct got_packidx *packidx;
525 	struct got_pack *pack;
526 	struct got_object_idset *idset;
527 	struct got_pack_metavec *v;
528 	int delta_cache_fd;
529 	struct got_repository *repo;
530 	got_pack_progress_cb progress_cb;
531 	void *progress_arg;
532 	struct got_ratelimit *rl;
533 	got_cancel_cb cancel_cb;
534 	void *cancel_arg;
535 	int ncolored;
536 	int nfound;
537 	int ntrees;
538 	int ncommits;
539 };
540 
541 static const struct got_error *
542 search_delta_for_object(struct got_object_id *id, void *data, void *arg)
543 {
544 	const struct got_error *err;
545 	struct got_pack_meta *m = data;
546 	struct search_deltas_arg *a = arg;
547 	int obj_idx;
548 	struct got_object *obj = NULL;
549 
550 	if (a->cancel_cb) {
551 		err = (*a->cancel_cb)(a->cancel_arg);
552 		if (err)
553 			return err;
554 	}
555 
556 	if (!got_repo_check_packidx_bloom_filter(a->repo,
557 	    a->packidx->path_packidx, id))
558 		return NULL;
559 
560 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
561 	if (obj_idx == -1)
562 		return NULL;
563 
564 	/* TODO:
565 	 * Opening and closing an object just to check its flags
566 	 * is a bit expensive. We could have an imsg which requests
567 	 * plain type/size information for an object without doing
568 	 * work such as traversing the object's entire delta chain
569 	 * to find the base object type, and other such info which
570 	 * we don't really need here.
571 	 */
572 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
573 	    a->packidx, obj_idx, a->repo);
574 	if (err)
575 		return err;
576 
577 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
578 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
579 		    a->delta_cache_fd, a->repo);
580 		if (err)
581 			goto done;
582 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
583 		    a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
584 		    got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
585 	}
586 done:
587 	got_object_close(obj);
588 	return err;
589 }
590 
591 static const struct got_error *
592 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
593     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
594     struct got_repository *repo,
595     got_pack_progress_cb progress_cb, void *progress_arg,
596     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
597 {
598 	const struct got_error *err = NULL;
599 	char *path_packfile = NULL;
600 	struct got_packidx *packidx;
601 	struct got_pack *pack;
602 	struct search_deltas_arg sda;
603 
604 	err = find_pack_for_reuse(&packidx, repo);
605 	if (err)
606 		return err;
607 
608 	if (packidx == NULL)
609 		return NULL;
610 
611 	err = got_packidx_get_packfile_path(&path_packfile,
612 	    packidx->path_packidx);
613 	if (err)
614 		return err;
615 
616 	pack = got_repo_get_cached_pack(repo, path_packfile);
617 	if (pack == NULL) {
618 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
619 		if (err)
620 			goto done;
621 	}
622 
623 	sda.packidx = packidx;
624 	sda.pack = pack;
625 	sda.idset = idset;
626 	sda.v = v;
627 	sda.delta_cache_fd = delta_cache_fd;
628 	sda.repo = repo;
629 	sda.progress_cb = progress_cb;
630 	sda.progress_arg = progress_arg;
631 	sda.rl = rl;
632 	sda.cancel_cb = cancel_cb;
633 	sda.cancel_arg = cancel_arg;
634 	sda.ncolored = ncolored;
635 	sda.nfound = nfound;
636 	sda.ntrees = ntrees;
637 	sda.ncommits = ncommits;
638 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
639 done:
640 	free(path_packfile);
641 	return err;
642 }
643 
644 static const struct got_error *
645 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
646     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
647     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 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
671 		    nreused + i, 0);
672 		if (err)
673 			goto done;
674 		m = meta[i];
675 
676 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
677 		    m->obj_type == GOT_OBJ_TYPE_TAG)
678 			continue;
679 
680 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
681 		if (err)
682 			goto done;
683 		m->size = raw->size;
684 
685 		if (raw->f == NULL) {
686 			err = got_deltify_init_mem(&m->dtab, raw->data,
687 			    raw->hdrlen, raw->size + raw->hdrlen);
688 		} else {
689 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
690 			    raw->size + raw->hdrlen);
691 		}
692 		if (err)
693 			goto done;
694 
695 		if (i > max_base_candidates) {
696 			struct got_pack_meta *n = NULL;
697 			n = meta[i - (max_base_candidates + 1)];
698 			got_deltify_free(n->dtab);
699 			n->dtab = NULL;
700 		}
701 
702 		best_size = raw->size;
703 		best_ndeltas = 0;
704 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
705 			if (cancel_cb) {
706 				err = (*cancel_cb)(cancel_arg);
707 				if (err)
708 					goto done;
709 			}
710 			base = meta[j];
711 			/* long chains make unpacking slow, avoid such bases */
712 			if (base->nchain >= 128 ||
713 			    base->obj_type != m->obj_type)
714 				continue;
715 
716 			err = got_object_raw_open(&base_raw, &outfd, repo,
717 			    &base->id);
718 			if (err)
719 				goto done;
720 
721 			if (raw->f == NULL && base_raw->f == NULL) {
722 				err = got_deltify_mem_mem(&deltas, &ndeltas,
723 				    raw->data, raw->hdrlen,
724 				    raw->size + raw->hdrlen,
725 				    base->dtab, base_raw->data,
726 				    base_raw->hdrlen,
727 				    base_raw->size + base_raw->hdrlen);
728 			} else if (raw->f == NULL) {
729 				err = got_deltify_mem_file(&deltas, &ndeltas,
730 				    raw->data, raw->hdrlen,
731 				    raw->size + raw->hdrlen,
732 				    base->dtab, base_raw->f,
733 				    base_raw->hdrlen,
734 				    base_raw->size + base_raw->hdrlen);
735 			} else if (base_raw->f == NULL) {
736 				err = got_deltify_file_mem(&deltas, &ndeltas,
737 				    raw->f, raw->hdrlen,
738 				    raw->size + raw->hdrlen,
739 				    base->dtab, base_raw->data,
740 				    base_raw->hdrlen,
741 				    base_raw->size + base_raw->hdrlen);
742 			} else {
743 				err = got_deltify(&deltas, &ndeltas,
744 				    raw->f, raw->hdrlen,
745 				    raw->size + raw->hdrlen,
746 				    base->dtab, base_raw->f, base_raw->hdrlen,
747 				    base_raw->size + base_raw->hdrlen);
748 			}
749 			got_object_raw_close(base_raw);
750 			base_raw = NULL;
751 			if (err)
752 				goto done;
753 
754 			size = delta_size(deltas, ndeltas);
755 			if (size + 32 < best_size){
756 				/*
757 				 * if we already picked a best delta,
758 				 * replace it.
759 				 */
760 				best_size = size;
761 				free(best_deltas);
762 				best_deltas = deltas;
763 				best_ndeltas = ndeltas;
764 				deltas = NULL;
765 				m->nchain = base->nchain + 1;
766 				m->prev = base;
767 				m->head = base->head;
768 				if (m->head == NULL)
769 					m->head = base;
770 			} else {
771 				free(deltas);
772 				deltas = NULL;
773 				ndeltas = 0;
774 			}
775 		}
776 
777 		if (best_ndeltas > 0) {
778 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
779 			    delta_memsize + best_size <= max_delta_memsize) {
780 				delta_memsize += best_size;
781 				err = encode_delta_in_mem(m, raw, best_deltas,
782 				    best_ndeltas, best_size, m->prev->size);
783 			} else {
784 				m->delta_offset = ftello(delta_cache);
785 				/*
786 				 * TODO:
787 				 * Storing compressed delta data in the delta
788 				 * cache file would probably be more efficient
789 				 * than writing uncompressed delta data here
790 				 * and compressing it while writing the pack
791 				 * file. This would also allow for reusing
792 				 * deltas in their compressed form.
793 				 */
794 				err = encode_delta(m, raw, best_deltas,
795 				    best_ndeltas, m->prev->size, delta_cache);
796 			}
797 			free(best_deltas);
798 			best_deltas = NULL;
799 			best_ndeltas = 0;
800 			if (err)
801 				goto done;
802 		}
803 
804 		got_object_raw_close(raw);
805 		raw = NULL;
806 	}
807 done:
808 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
809 		got_deltify_free(meta[i]->dtab);
810 		meta[i]->dtab = NULL;
811 	}
812 	if (raw)
813 		got_object_raw_close(raw);
814 	if (base_raw)
815 		got_object_raw_close(base_raw);
816 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
817 		err = got_error_from_errno("close");
818 	free(deltas);
819 	free(best_deltas);
820 	return err;
821 }
822 
823 static const struct got_error *
824 search_packidx(int *found, struct got_object_id *id,
825     struct got_repository *repo)
826 {
827 	const struct got_error *err = NULL;
828 	struct got_packidx *packidx = NULL;
829 	int idx;
830 
831 	*found = 0;
832 
833 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
834 	if (err == NULL)
835 		*found = 1; /* object is already packed */
836 	else if (err->code == GOT_ERR_NO_OBJ)
837 		err = NULL;
838 	return err;
839 }
840 
841 static const int obj_types[] = {
842 	GOT_OBJ_TYPE_ANY,
843 	GOT_OBJ_TYPE_COMMIT,
844 	GOT_OBJ_TYPE_TREE,
845 	GOT_OBJ_TYPE_BLOB,
846 	GOT_OBJ_TYPE_TAG,
847 	GOT_OBJ_TYPE_OFFSET_DELTA,
848 	GOT_OBJ_TYPE_REF_DELTA
849 };
850 
851 static const struct got_error *
852 add_object(int want_meta, struct got_object_idset *idset,
853     struct got_object_id *id, const char *path, int obj_type,
854     time_t mtime, int loose_obj_only, struct got_repository *repo,
855     int *ncolored, int *nfound, int *ntrees,
856     got_pack_progress_cb progress_cb, void *progress_arg,
857     struct got_ratelimit *rl)
858 {
859 	const struct got_error *err;
860 	struct got_pack_meta *m = NULL;
861 
862 	if (loose_obj_only) {
863 		int is_packed;
864 		err = search_packidx(&is_packed, id, repo);
865 		if (err)
866 			return err;
867 		if (is_packed && want_meta)
868 			return NULL;
869 	}
870 
871 	if (want_meta) {
872 		err = alloc_meta(&m, id, path, obj_type, mtime);
873 		if (err)
874 			return err;
875 
876 		(*nfound)++;
877 		err = report_progress(progress_cb, progress_arg, rl,
878 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
879 		if (err)
880 			return err;
881 	}
882 
883 	return got_object_idset_add(idset, id, m);
884 }
885 
886 static const struct got_error *
887 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
888     struct got_object_idset *idset, struct got_object_id *tree_id,
889     const char *dpath, time_t mtime, struct got_repository *repo,
890     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
891     got_pack_progress_cb progress_cb, void *progress_arg,
892     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
893 {
894 	const struct got_error *err;
895 	struct got_tree_object *tree;
896 	char *p = NULL;
897 	int i;
898 
899 	err = got_object_open_as_tree(&tree, repo, tree_id);
900 	if (err)
901 		return err;
902 
903 	(*ntrees)++;
904 	err = report_progress(progress_cb, progress_arg, rl,
905 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
906 	if (err)
907 		return err;
908 
909 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
910 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
911 		struct got_object_id *id = got_tree_entry_get_id(e);
912 		mode_t mode = got_tree_entry_get_mode(e);
913 
914 		if (cancel_cb) {
915 			err = (*cancel_cb)(cancel_arg);
916 			if (err)
917 				break;
918 		}
919 
920 		if (got_object_tree_entry_is_submodule(e) ||
921 		    got_object_idset_contains(idset, id))
922 			continue;
923 		
924 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
925 		    got_tree_entry_get_name(e)) == -1) {
926 			err = got_error_from_errno("asprintf");
927 			break;
928 		}
929 
930 		if (S_ISDIR(mode)) {
931 			struct got_object_qid *qid;
932 			err = got_object_qid_alloc(&qid, id);
933 			if (err)
934 				break;
935 			STAILQ_INSERT_TAIL(ids, qid, entry);
936 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
937 			err = add_object(want_meta, idset, id, p,
938 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo,
939 			    ncolored, nfound, ntrees,
940 			    progress_cb, progress_arg, rl);
941 			if (err)
942 				break;
943 		}
944 		free(p);
945 		p = NULL;
946 	}
947 
948 	got_object_tree_close(tree);
949 	free(p);
950 	return err;
951 }
952 
953 static const struct got_error *
954 load_tree(int want_meta, struct got_object_idset *idset,
955     struct got_object_id *tree_id, const char *dpath, time_t mtime,
956     struct got_repository *repo, int loose_obj_only,
957     int *ncolored, int *nfound, int *ntrees,
958     got_pack_progress_cb progress_cb, void *progress_arg,
959     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
960 {
961 	const struct got_error *err = NULL;
962 	struct got_object_id_queue tree_ids;
963 	struct got_object_qid *qid;
964 
965 	if (got_object_idset_contains(idset, tree_id))
966 		return NULL;
967 
968 	err = got_object_qid_alloc(&qid, tree_id);
969 	if (err)
970 		return err;
971 
972 	STAILQ_INIT(&tree_ids);
973 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
974 
975 	while (!STAILQ_EMPTY(&tree_ids)) {
976 		if (cancel_cb) {
977 			err = (*cancel_cb)(cancel_arg);
978 			if (err)
979 				break;
980 		}
981 
982 		qid = STAILQ_FIRST(&tree_ids);
983 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
984 
985 		if (got_object_idset_contains(idset, qid->id)) {
986 			got_object_qid_free(qid);
987 			continue;
988 		}
989 
990 		err = add_object(want_meta, idset, qid->id, dpath,
991 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo,
992 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
993 		if (err) {
994 			got_object_qid_free(qid);
995 			break;
996 		}
997 
998 		err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
999 		    dpath, mtime, repo, loose_obj_only, ncolored, nfound,
1000 		    ntrees, progress_cb, progress_arg, rl,
1001 		    cancel_cb, cancel_arg);
1002 		got_object_qid_free(qid);
1003 		if (err)
1004 			break;
1005 	}
1006 
1007 	got_object_id_queue_free(&tree_ids);
1008 	return err;
1009 }
1010 
1011 static const struct got_error *
1012 load_commit(int want_meta, struct got_object_idset *idset,
1013     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1014     int *ncolored, int *nfound, int *ntrees,
1015     got_pack_progress_cb progress_cb, void *progress_arg,
1016     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1017 {
1018 	const struct got_error *err;
1019 	struct got_commit_object *commit;
1020 
1021 	if (got_object_idset_contains(idset, id))
1022 		return NULL;
1023 
1024 	if (loose_obj_only) {
1025 		int is_packed;
1026 		err = search_packidx(&is_packed, id, repo);
1027 		if (err)
1028 			return err;
1029 		if (is_packed && want_meta)
1030 			return NULL;
1031 	}
1032 
1033 	err = got_object_open_as_commit(&commit, repo, id);
1034 	if (err)
1035 		return err;
1036 
1037 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1038 	    got_object_commit_get_committer_time(commit),
1039 	    loose_obj_only, repo,
1040 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1041 	if (err)
1042 		goto done;
1043 
1044 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1045 	    "", got_object_commit_get_committer_time(commit),
1046 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1047 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1048 done:
1049 	got_object_commit_close(commit);
1050 	return err;
1051 }
1052 
1053 static const struct got_error *
1054 load_tag(int want_meta, struct got_object_idset *idset,
1055     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1056     int *ncolored, int *nfound, int *ntrees,
1057     got_pack_progress_cb progress_cb, void *progress_arg,
1058     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1059 {
1060 	const struct got_error *err;
1061 	struct got_tag_object *tag = NULL;
1062 
1063 	if (got_object_idset_contains(idset, id))
1064 		return NULL;
1065 
1066 	if (loose_obj_only) {
1067 		int is_packed;
1068 		err = search_packidx(&is_packed, id, repo);
1069 		if (err)
1070 			return err;
1071 		if (is_packed && want_meta)
1072 			return NULL;
1073 	}
1074 
1075 	err = got_object_open_as_tag(&tag, repo, id);
1076 	if (err)
1077 		return err;
1078 
1079 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1080 	    got_object_tag_get_tagger_time(tag), loose_obj_only, repo,
1081 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1082 	if (err)
1083 		goto done;
1084 
1085 	switch (got_object_tag_get_object_type(tag)) {
1086 	case GOT_OBJ_TYPE_COMMIT:
1087 		err = load_commit(want_meta, idset,
1088 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1089 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1090 		    cancel_cb, cancel_arg);
1091 		break;
1092 	case GOT_OBJ_TYPE_TREE:
1093 		err = load_tree(want_meta, idset,
1094 		    got_object_tag_get_object_id(tag), "",
1095 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1096 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1097 		    cancel_cb, cancel_arg);
1098 		break;
1099 	default:
1100 		break;
1101 	}
1102 
1103 done:
1104 	got_object_tag_close(tag);
1105 	return err;
1106 }
1107 
1108 enum findtwixt_color {
1109 	COLOR_KEEP = 0,
1110 	COLOR_DROP,
1111 	COLOR_BLANK,
1112 };
1113 static const int findtwixt_colors[] = {
1114 	COLOR_KEEP,
1115 	COLOR_DROP,
1116 	COLOR_BLANK
1117 };
1118 
1119 static const struct got_error *
1120 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1121     int color, struct got_repository *repo)
1122 {
1123 	const struct got_error *err;
1124 	struct got_object_qid *qid;
1125 
1126 	err = got_object_qid_alloc(&qid, id);
1127 	if (err)
1128 		return err;
1129 
1130 	STAILQ_INSERT_TAIL(ids, qid, entry);
1131 	qid->data = (void *)&findtwixt_colors[color];
1132 	return NULL;
1133 }
1134 
1135 static const struct got_error *
1136 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1137     struct got_object_id *id, struct got_repository *repo,
1138     got_cancel_cb cancel_cb, void *cancel_arg)
1139 {
1140 	const struct got_error *err = NULL;
1141 	struct got_commit_object *commit;
1142 	const struct got_object_id_queue *parents;
1143 	struct got_object_id_queue ids;
1144 	struct got_object_qid *qid;
1145 
1146 	STAILQ_INIT(&ids);
1147 
1148 	err = got_object_qid_alloc(&qid, id);
1149 	if (err)
1150 		return err;
1151 	STAILQ_INSERT_HEAD(&ids, qid, entry);
1152 
1153 	while (!STAILQ_EMPTY(&ids)) {
1154 		if (cancel_cb) {
1155 			err = (*cancel_cb)(cancel_arg);
1156 			if (err)
1157 				break;
1158 		}
1159 
1160 		qid = STAILQ_FIRST(&ids);
1161 		STAILQ_REMOVE_HEAD(&ids, entry);
1162 
1163 		if (got_object_idset_contains(drop, qid->id)) {
1164 			got_object_qid_free(qid);
1165 			continue;
1166 		}
1167 
1168 		err = got_object_idset_add(drop, qid->id,
1169 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1170 		if (err) {
1171 			got_object_qid_free(qid);
1172 			break;
1173 		}
1174 
1175 		if (!got_object_idset_contains(keep, qid->id)) {
1176 			got_object_qid_free(qid);
1177 			continue;
1178 		}
1179 
1180 		err = got_object_open_as_commit(&commit, repo, qid->id);
1181 		got_object_qid_free(qid);
1182 		if (err)
1183 			break;
1184 
1185 		parents = got_object_commit_get_parent_ids(commit);
1186 		if (parents) {
1187 			err = got_object_id_queue_copy(parents, &ids);
1188 			if (err) {
1189 				got_object_commit_close(commit);
1190 				break;
1191 			}
1192 		}
1193 		got_object_commit_close(commit);
1194 	}
1195 
1196 	got_object_id_queue_free(&ids);
1197 	return err;
1198 }
1199 
1200 struct append_id_arg {
1201 	struct got_object_id **array;
1202 	int idx;
1203 };
1204 
1205 static const struct got_error *
1206 append_id(struct got_object_id *id, void *data, void *arg)
1207 {
1208 	struct append_id_arg *a = arg;
1209 
1210 	a->array[a->idx] = got_object_id_dup(id);
1211 	if (a->array[a->idx] == NULL)
1212 		return got_error_from_errno("got_object_id_dup");
1213 
1214 	a->idx++;
1215 	return NULL;
1216 }
1217 
1218 static const struct got_error *
1219 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1220     struct got_object_id **head, int nhead,
1221     struct got_object_id **tail, int ntail,
1222     struct got_repository *repo,
1223     got_pack_progress_cb progress_cb, void *progress_arg,
1224     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1225 {
1226 	const struct got_error *err = NULL;
1227 	struct got_object_id_queue ids;
1228 	struct got_object_idset *keep, *drop;
1229 	struct got_object_qid *qid;
1230 	int i, ncolor, nkeep, obj_type;
1231 
1232 	STAILQ_INIT(&ids);
1233 	*res = NULL;
1234 	*nres = 0;
1235 	*ncolored = 0;
1236 
1237 	keep = got_object_idset_alloc();
1238 	if (keep == NULL)
1239 		return got_error_from_errno("got_object_idset_alloc");
1240 
1241 	drop = got_object_idset_alloc();
1242 	if (drop == NULL) {
1243 		err = got_error_from_errno("got_object_idset_alloc");
1244 		goto done;
1245 	}
1246 
1247 	for (i = 0; i < nhead; i++) {
1248 		struct got_object_id *id = head[i];
1249 		if (id == NULL)
1250 			continue;
1251 		err = got_object_get_type(&obj_type, repo, id);
1252 		if (err)
1253 			return err;
1254 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1255 			continue;
1256 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1257 		if (err)
1258 			goto done;
1259 	}		
1260 	for (i = 0; i < ntail; i++) {
1261 		struct got_object_id *id = tail[i];
1262 		if (id == NULL)
1263 			continue;
1264 		err = got_object_get_type(&obj_type, repo, id);
1265 		if (err)
1266 			return err;
1267 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1268 			continue;
1269 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1270 		if (err)
1271 			goto done;
1272 	}
1273 
1274 	while (!STAILQ_EMPTY(&ids)) {
1275 		int qcolor;
1276 		qid = STAILQ_FIRST(&ids);
1277 		qcolor = *((int *)qid->data);
1278 
1279 		if (got_object_idset_contains(drop, qid->id))
1280 			ncolor = COLOR_DROP;
1281 		else if (got_object_idset_contains(keep, qid->id))
1282 			ncolor = COLOR_KEEP;
1283 		else
1284 			ncolor = COLOR_BLANK;
1285 
1286 		(*ncolored)++;
1287 		err = report_progress(progress_cb, progress_arg, rl,
1288 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1289 		if (err)
1290 			goto done;
1291 
1292 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1293 		    qcolor == COLOR_KEEP)) {
1294 			STAILQ_REMOVE_HEAD(&ids, entry);
1295 			got_object_qid_free(qid);
1296 			continue;
1297 		}
1298 
1299 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1300 			err = drop_commit(keep, drop, qid->id, repo,
1301 			    cancel_cb, cancel_arg);
1302 			if (err)
1303 				goto done;
1304 		} else if (ncolor == COLOR_BLANK) {
1305 			struct got_commit_object *commit;
1306 			struct got_object_id *id;
1307 			const struct got_object_id_queue *parents;
1308 			struct got_object_qid *pid;
1309 
1310 			id = got_object_id_dup(qid->id);
1311 			if (id == NULL) {
1312 				err = got_error_from_errno("got_object_id_dup");
1313 				goto done;
1314 			}
1315 			if (qcolor == COLOR_KEEP)
1316 				err = got_object_idset_add(keep, id,
1317 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1318 			else
1319 				err = got_object_idset_add(drop, id,
1320 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1321 			if (err) {
1322 				free(id);
1323 				goto done;
1324 			}
1325 
1326 			err = got_object_open_as_commit(&commit, repo, id);
1327 			if (err) {
1328 				free(id);
1329 				goto done;
1330 			}
1331 			parents = got_object_commit_get_parent_ids(commit);
1332 			if (parents) {
1333 				STAILQ_FOREACH(pid, parents, entry) {
1334 					err = queue_commit_id(&ids, pid->id,
1335 					    qcolor, repo);
1336 					if (err) {
1337 						free(id);
1338 						goto done;
1339 					}
1340 				}
1341 			}
1342 			got_object_commit_close(commit);
1343 			commit = NULL;
1344 		} else {
1345 			/* should not happen */
1346 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1347 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1348 			goto done;
1349 		}
1350 
1351 		STAILQ_REMOVE_HEAD(&ids, entry);
1352 		got_object_qid_free(qid);
1353 	}
1354 
1355 	nkeep = got_object_idset_num_elements(keep);
1356 	if (nkeep > 0) {
1357 		struct append_id_arg arg;
1358 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1359 		if (arg.array == NULL) {
1360 			err = got_error_from_errno("calloc");
1361 			goto done;
1362 		}
1363 		arg.idx = 0;
1364 		err = got_object_idset_for_each(keep, append_id, &arg);
1365 		if (err) {
1366 			free(arg.array);
1367 			goto done;
1368 		}
1369 		*res = arg.array;
1370 		*nres = nkeep;
1371 	}
1372 done:
1373 	got_object_idset_free(keep);
1374 	got_object_idset_free(drop);
1375 	got_object_id_queue_free(&ids);
1376 	return err;
1377 }
1378 
1379 static const struct got_error *
1380 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1381     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1382     struct got_object_id **ours, int nours, struct got_repository *repo,
1383     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1384     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1385 {
1386 	const struct got_error *err = NULL;
1387 	struct got_object_id **ids = NULL;
1388 	int i, nobj = 0, obj_type;
1389 
1390 	*ncolored = 0;
1391 	*nfound = 0;
1392 	*ntrees = 0;
1393 
1394 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1395 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1396 	if (err || nobj == 0)
1397 		goto done;
1398 
1399 	for (i = 0; i < ntheirs; i++) {
1400 		struct got_object_id *id = theirs[i];
1401 		if (id == NULL)
1402 			continue;
1403 		err = got_object_get_type(&obj_type, repo, id);
1404 		if (err)
1405 			return err;
1406 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1407 			continue;
1408 		err = load_commit(0, idset, id, repo, loose_obj_only,
1409 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1410 		    cancel_cb, cancel_arg);
1411 		if (err)
1412 			goto done;
1413 	}
1414 
1415 	for (i = 0; i < ntheirs; i++) {
1416 		struct got_object_id *id = theirs[i];
1417 		struct got_pack_meta *m;
1418 		if (id == NULL)
1419 			continue;
1420 		m = got_object_idset_get(idset, id);
1421 		if (m == NULL) {
1422 			err = got_object_get_type(&obj_type, repo, id);
1423 			if (err)
1424 				goto done;
1425 		} else
1426 			obj_type = m->obj_type;
1427 		if (obj_type != GOT_OBJ_TYPE_TAG)
1428 			continue;
1429 		err = load_tag(0, idset, id, repo, loose_obj_only,
1430 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1431 		    cancel_cb, cancel_arg);
1432 		if (err)
1433 			goto done;
1434 	}
1435 
1436 	for (i = 0; i < nobj; i++) {
1437 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1438 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1439 		    cancel_cb, cancel_arg);
1440 		if (err)
1441 			goto done;
1442 	}
1443 
1444 	for (i = 0; i < nours; i++) {
1445 		struct got_object_id *id = ours[i];
1446 		struct got_pack_meta *m;
1447 		if (id == NULL)
1448 			continue;
1449 		m = got_object_idset_get(idset, id);
1450 		if (m == NULL) {
1451 			err = got_object_get_type(&obj_type, repo, id);
1452 			if (err)
1453 				goto done;
1454 		} else
1455 			obj_type = m->obj_type;
1456 		if (obj_type != GOT_OBJ_TYPE_TAG)
1457 			continue;
1458 		err = load_tag(1, idset, id, repo, loose_obj_only,
1459 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1460 		    cancel_cb, cancel_arg);
1461 		if (err)
1462 			goto done;
1463 	}
1464 done:
1465 	for (i = 0; i < nobj; i++) {
1466 		free(ids[i]);
1467 	}
1468 	free(ids);
1469 	return err;
1470 }
1471 
1472 const struct got_error *
1473 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1474 {
1475 	size_t n;
1476 
1477 	SHA1Update(ctx, buf, len);
1478 	n = fwrite(buf, 1, len, f);
1479 	if (n != len)
1480 		return got_ferror(f, GOT_ERR_IO);
1481 	return NULL;
1482 }
1483 
1484 static void
1485 putbe32(char *b, uint32_t n)
1486 {
1487 	b[0] = n >> 24;
1488 	b[1] = n >> 16;
1489 	b[2] = n >> 8;
1490 	b[3] = n >> 0;
1491 }
1492 
1493 static int
1494 write_order_cmp(const void *pa, const void *pb)
1495 {
1496 	struct got_pack_meta *a, *b, *ahd, *bhd;
1497 
1498 	a = *(struct got_pack_meta **)pa;
1499 	b = *(struct got_pack_meta **)pb;
1500 	ahd = (a->head == NULL) ? a : a->head;
1501 	bhd = (b->head == NULL) ? b : b->head;
1502 	if (ahd->mtime != bhd->mtime)
1503 		return bhd->mtime - ahd->mtime;
1504 	if (ahd != bhd)
1505 		return (uintptr_t)bhd - (uintptr_t)ahd;
1506 	if (a->nchain != b->nchain)
1507 		return a->nchain - b->nchain;
1508 	return a->mtime - b->mtime;
1509 }
1510 
1511 static int
1512 reuse_write_order_cmp(const void *pa, const void *pb)
1513 {
1514 	struct got_pack_meta *a, *b;
1515 
1516 	a = *(struct got_pack_meta **)pa;
1517 	b = *(struct got_pack_meta **)pb;
1518 
1519 	if (a->reused_delta_offset < b->reused_delta_offset)
1520 		return -1;
1521 	if (a->reused_delta_offset > b->reused_delta_offset)
1522 		return 1;
1523 	return 0;
1524 }
1525 
1526 static const struct got_error *
1527 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1528 {
1529 	size_t i;
1530 
1531 	*hdrlen = 0;
1532 
1533 	hdr[0] = obj_type << 4;
1534 	hdr[0] |= len & 0xf;
1535 	len >>= 4;
1536 	for (i = 1; len != 0; i++){
1537 		if (i >= bufsize)
1538 			return got_error(GOT_ERR_NO_SPACE);
1539 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1540 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1541 		len >>= GOT_DELTA_SIZE_SHIFT;
1542 	}
1543 
1544 	*hdrlen = i;
1545 	return NULL;
1546 }
1547 
1548 static int
1549 packoff(char *hdr, off_t off)
1550 {
1551 	int i, j;
1552 	char rbuf[8];
1553 
1554 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1555 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1556 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1557 		    GOT_DELTA_SIZE_MORE;
1558 	}
1559 
1560 	j = 0;
1561 	while (i > 0)
1562 		hdr[j++] = rbuf[--i];
1563 	return j;
1564 }
1565 
1566 static const struct got_error *
1567 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1568     struct got_pack_meta *m)
1569 {
1570 	const struct got_error *err;
1571 	char buf[32];
1572 	int nh;
1573 
1574 	if (m->prev->off != 0) {
1575 		err = packhdr(&nh, buf, sizeof(buf),
1576 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1577 		if (err)
1578 			return err;
1579 		nh += packoff(buf + nh, m->off - m->prev->off);
1580 		err = hwrite(packfile, buf, nh, ctx);
1581 		if (err)
1582 			return err;
1583 		*packfile_size += nh;
1584 	} else {
1585 		err = packhdr(&nh, buf, sizeof(buf),
1586 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1587 		if (err)
1588 			return err;
1589 		err = hwrite(packfile, buf, nh, ctx);
1590 		if (err)
1591 			return err;
1592 		*packfile_size += nh;
1593 		err = hwrite(packfile, m->prev->id.sha1,
1594 		    sizeof(m->prev->id.sha1), ctx);
1595 		if (err)
1596 			return err;
1597 		*packfile_size += sizeof(m->prev->id.sha1);
1598 	}
1599 
1600 	return NULL;
1601 }
1602 
1603 static const struct got_error *
1604 write_packed_object(off_t *packfile_size, FILE *packfile,
1605     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1606     SHA1_CTX *ctx, struct got_repository *repo)
1607 {
1608 	const struct got_error *err = NULL;
1609 	struct got_deflate_checksum csum;
1610 	char buf[32];
1611 	int nh;
1612 	struct got_raw_object *raw = NULL;
1613 	off_t outlen;
1614 
1615 	csum.output_sha1 = ctx;
1616 	csum.output_crc = NULL;
1617 
1618 	m->off = ftello(packfile);
1619 	if (m->delta_len == 0) {
1620 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1621 		if (err)
1622 			goto done;
1623 		err = packhdr(&nh, buf, sizeof(buf),
1624 		    m->obj_type, raw->size);
1625 		if (err)
1626 			goto done;
1627 		err = hwrite(packfile, buf, nh, ctx);
1628 		if (err)
1629 			goto done;
1630 		*packfile_size += nh;
1631 		if (raw->f == NULL) {
1632 			err = got_deflate_to_file_mmap(&outlen,
1633 			    raw->data + raw->hdrlen, 0, raw->size,
1634 			    packfile, &csum);
1635 			if (err)
1636 				goto done;
1637 		} else {
1638 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1639 			    == -1) {
1640 				err = got_error_from_errno("fseeko");
1641 				goto done;
1642 			}
1643 			err = got_deflate_to_file(&outlen, raw->f,
1644 			    raw->size, packfile, &csum);
1645 			if (err)
1646 				goto done;
1647 		}
1648 		*packfile_size += outlen;
1649 		got_object_raw_close(raw);
1650 		raw = NULL;
1651 	} else if (m->delta_buf) {
1652 		err = deltahdr(packfile_size, ctx, packfile, m);
1653 		if (err)
1654 			goto done;
1655 		err = got_deflate_to_file_mmap(&outlen,
1656 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1657 		if (err)
1658 			goto done;
1659 		*packfile_size += outlen;
1660 		free(m->delta_buf);
1661 		m->delta_buf = NULL;
1662 	} else {
1663 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1664 		    == -1) {
1665 			err = got_error_from_errno("fseeko");
1666 			goto done;
1667 		}
1668 		err = deltahdr(packfile_size, ctx, packfile, m);
1669 		if (err)
1670 			goto done;
1671 		err = got_deflate_to_file(&outlen, delta_cache,
1672 		    m->delta_len, packfile, &csum);
1673 		if (err)
1674 			goto done;
1675 		*packfile_size += outlen;
1676 	}
1677 done:
1678 	if (raw)
1679 		got_object_raw_close(raw);
1680 	return err;
1681 }
1682 
1683 static const struct got_error *
1684 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1685     struct got_pack_meta **deltify, int ndeltify,
1686     struct got_pack_meta **reuse, int nreuse,
1687     int ncolored, int nfound, int ntrees, int nours,
1688     struct got_repository *repo,
1689     got_pack_progress_cb progress_cb, void *progress_arg,
1690     struct got_ratelimit *rl,
1691     got_cancel_cb cancel_cb, void *cancel_arg)
1692 {
1693 	const struct got_error *err = NULL;
1694 	int i;
1695 	SHA1_CTX ctx;
1696 	struct got_pack_meta *m;
1697 	char buf[32];
1698 	size_t n;
1699 	off_t packfile_size = 0;
1700 	int outfd = -1;
1701 
1702 	SHA1Init(&ctx);
1703 
1704 	err = hwrite(packfile, "PACK", 4, &ctx);
1705 	if (err)
1706 		return err;
1707 	putbe32(buf, GOT_PACKFILE_VERSION);
1708 	err = hwrite(packfile, buf, 4, &ctx);
1709 	if (err)
1710 		goto done;
1711 	putbe32(buf, ndeltify + nreuse);
1712 	err = hwrite(packfile, buf, 4, &ctx);
1713 	if (err)
1714 		goto done;
1715 
1716 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1717 	    write_order_cmp);
1718 	for (i = 0; i < ndeltify; i++) {
1719 		err = report_progress(progress_cb, progress_arg, rl,
1720 		    ncolored, nfound, ntrees, packfile_size, nours,
1721 		    ndeltify + nreuse, ndeltify + nreuse, i);
1722 		if (err)
1723 			goto done;
1724 		m = deltify[i];
1725 		err = write_packed_object(&packfile_size, packfile,
1726 		    delta_cache, m, &outfd, &ctx, repo);
1727 		if (err)
1728 			goto done;
1729 	}
1730 
1731 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1732 	    reuse_write_order_cmp);
1733 	for (i = 0; i < nreuse; i++) {
1734 		err = report_progress(progress_cb, progress_arg, rl,
1735 		    ncolored, nfound, ntrees, packfile_size, nours,
1736 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1737 		if (err)
1738 			goto done;
1739 		m = reuse[i];
1740 		err = write_packed_object(&packfile_size, packfile,
1741 		    delta_cache, m, &outfd, &ctx, repo);
1742 		if (err)
1743 			goto done;
1744 	}
1745 
1746 	SHA1Final(pack_sha1, &ctx);
1747 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1748 	if (n != SHA1_DIGEST_LENGTH)
1749 		err = got_ferror(packfile, GOT_ERR_IO);
1750 	packfile_size += SHA1_DIGEST_LENGTH;
1751 	packfile_size += sizeof(struct got_packfile_hdr);
1752 	if (progress_cb) {
1753 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1754 		    packfile_size, nours, ndeltify + nreuse,
1755 		    ndeltify + nreuse, ndeltify + nreuse);
1756 		if (err)
1757 			goto done;
1758 	}
1759 done:
1760 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1761 		err = got_error_from_errno("close");
1762 	return err;
1763 }
1764 
1765 static const struct got_error *
1766 remove_unused_object(struct got_object_idset_element *entry, void *arg)
1767 {
1768 	struct got_object_idset *idset = arg;
1769 
1770 	if (got_object_idset_get_element_data(entry) == NULL)
1771 		got_object_idset_remove_element(idset, entry);
1772 
1773 	return NULL;
1774 }
1775 
1776 static const struct got_error *
1777 remove_reused_object(struct got_object_idset_element *entry, void *arg)
1778 {
1779 	struct got_object_idset *idset = arg;
1780 	struct got_pack_meta *m;
1781 
1782 	m = got_object_idset_get_element_data(entry);
1783 	if (m->have_reused_delta)
1784 		got_object_idset_remove_element(idset, entry);
1785 
1786 	return NULL;
1787 }
1788 
1789 static const struct got_error *
1790 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1791 {
1792 	struct got_pack_meta *m = data;
1793 	struct got_pack_metavec *v = arg;
1794 
1795 	return add_meta(m, v);
1796 }
1797 
1798 const struct got_error *
1799 got_pack_create(uint8_t *packsha1, FILE *packfile,
1800     struct got_object_id **theirs, int ntheirs,
1801     struct got_object_id **ours, int nours,
1802     struct got_repository *repo, int loose_obj_only, int allow_empty,
1803     got_pack_progress_cb progress_cb, void *progress_arg,
1804     got_cancel_cb cancel_cb, void *cancel_arg)
1805 {
1806 	const struct got_error *err;
1807 	int delta_cache_fd = -1;
1808 	FILE *delta_cache = NULL;
1809 	struct got_object_idset *idset;
1810 	struct got_ratelimit rl;
1811 	struct got_pack_metavec deltify, reuse;
1812 	int ncolored = 0, nfound = 0, ntrees = 0;
1813 
1814 	memset(&deltify, 0, sizeof(deltify));
1815 	memset(&reuse, 0, sizeof(reuse));
1816 
1817 	got_ratelimit_init(&rl, 0, 500);
1818 
1819 	idset = got_object_idset_alloc();
1820 	if (idset == NULL)
1821 		return got_error_from_errno("got_object_idset_alloc");
1822 
1823 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1824 	    ntheirs, ours, nours, repo, loose_obj_only,
1825 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1826 	if (err)
1827 		return err;
1828 
1829 	err = got_object_idset_for_each_element(idset,
1830 	    remove_unused_object, idset);
1831 	if (err)
1832 		goto done;
1833 
1834 	if (progress_cb) {
1835 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1836 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1837 		if (err)
1838 			goto done;
1839 	}
1840 
1841 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1842 		err = got_error(GOT_ERR_CANNOT_PACK);
1843 		goto done;
1844 	}
1845 
1846 	delta_cache_fd = got_opentempfd();
1847 	if (delta_cache_fd == -1) {
1848 		err = got_error_from_errno("got_opentemp");
1849 		goto done;
1850 	}
1851 
1852 	reuse.metasz = 64;
1853 	reuse.meta = calloc(reuse.metasz,
1854 	    sizeof(struct got_pack_meta *));
1855 	if (reuse.meta == NULL) {
1856 		err = got_error_from_errno("calloc");
1857 		goto done;
1858 	}
1859 
1860 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1861 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1862 	    cancel_cb, cancel_arg);
1863 	if (err)
1864 		goto done;
1865 	if (reuse.nmeta > 0) {
1866 		err = got_object_idset_for_each_element(idset,
1867 		    remove_reused_object, idset);
1868 		if (err)
1869 			goto done;
1870 	}
1871 
1872 	delta_cache = fdopen(delta_cache_fd, "a+");
1873 	if (delta_cache == NULL) {
1874 		err = got_error_from_errno("fdopen");
1875 		goto done;
1876 	}
1877 	delta_cache_fd = -1;
1878 
1879 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1880 		err = got_error_from_errno("fseeko");
1881 		goto done;
1882 	}
1883 
1884 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1885 	    sizeof(struct got_pack_meta *));
1886 	if (deltify.meta == NULL) {
1887 		err = got_error_from_errno("calloc");
1888 		goto done;
1889 	}
1890 	deltify.metasz = got_object_idset_num_elements(idset);
1891 
1892 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1893 	if (err)
1894 		goto done;
1895 	if (deltify.nmeta > 0) {
1896 		err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1897 		    nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1898 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1899 		if (err)
1900 			goto done;
1901 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1902 			err = got_error_from_errno("fseeko");
1903 			goto done;
1904 		}
1905 	}
1906 
1907 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1908 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1909 	    nours, repo, progress_cb, progress_arg, &rl,
1910 	    cancel_cb, cancel_arg);
1911 	if (err)
1912 		goto done;
1913 done:
1914 	free_nmeta(deltify.meta, deltify.nmeta);
1915 	free_nmeta(reuse.meta, reuse.nmeta);
1916 	got_object_idset_free(idset);
1917 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1918 		err = got_error_from_errno("close");
1919 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1920 		err = got_error_from_errno("fclose");
1921 	return err;
1922 }