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 struct got_error *
842 add_object(int want_meta, struct got_object_idset *idset,
843     struct got_object_id *id, const char *path, int obj_type,
844     time_t mtime, int loose_obj_only, struct got_repository *repo,
845     int *ncolored, int *nfound, int *ntrees,
846     got_pack_progress_cb progress_cb, void *progress_arg,
847     struct got_ratelimit *rl)
848 {
849 	const struct got_error *err;
850 	struct got_pack_meta *m = NULL;
851 
852 	if (loose_obj_only) {
853 		int is_packed;
854 		err = search_packidx(&is_packed, id, repo);
855 		if (err)
856 			return err;
857 		if (is_packed && want_meta)
858 			return NULL;
859 	}
860 
861 	if (want_meta) {
862 		err = alloc_meta(&m, id, path, obj_type, mtime);
863 		if (err)
864 			return err;
865 
866 		(*nfound)++;
867 		err = report_progress(progress_cb, progress_arg, rl,
868 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
869 		if (err) {
870 			clear_meta(m);
871 			free(m);
872 			return err;
873 		}
874 	}
875 
876 	err = got_object_idset_add(idset, id, m);
877 	if (err) {
878 		clear_meta(m);
879 		free(m);
880 	}
881 	return err;
882 }
883 
884 static const struct got_error *
885 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
886     struct got_object_idset *idset, struct got_object_id *tree_id,
887     const char *dpath, time_t mtime, struct got_repository *repo,
888     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
889     got_pack_progress_cb progress_cb, void *progress_arg,
890     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
891 {
892 	const struct got_error *err;
893 	struct got_tree_object *tree;
894 	char *p = NULL;
895 	int i;
896 
897 	err = got_object_open_as_tree(&tree, repo, tree_id);
898 	if (err)
899 		return err;
900 
901 	(*ntrees)++;
902 	err = report_progress(progress_cb, progress_arg, rl,
903 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
904 	if (err)
905 		return err;
906 
907 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
908 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
909 		struct got_object_id *id = got_tree_entry_get_id(e);
910 		mode_t mode = got_tree_entry_get_mode(e);
911 
912 		if (cancel_cb) {
913 			err = (*cancel_cb)(cancel_arg);
914 			if (err)
915 				break;
916 		}
917 
918 		if (got_object_tree_entry_is_submodule(e) ||
919 		    got_object_idset_contains(idset, id))
920 			continue;
921 		
922 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
923 		    got_tree_entry_get_name(e)) == -1) {
924 			err = got_error_from_errno("asprintf");
925 			break;
926 		}
927 
928 		if (S_ISDIR(mode)) {
929 			struct got_object_qid *qid;
930 			err = got_object_qid_alloc(&qid, id);
931 			if (err)
932 				break;
933 			STAILQ_INSERT_TAIL(ids, qid, entry);
934 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
935 			err = add_object(want_meta, idset, id, p,
936 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo,
937 			    ncolored, nfound, ntrees,
938 			    progress_cb, progress_arg, rl);
939 			if (err)
940 				break;
941 		}
942 		free(p);
943 		p = NULL;
944 	}
945 
946 	got_object_tree_close(tree);
947 	free(p);
948 	return err;
949 }
950 
951 static const struct got_error *
952 load_tree(int want_meta, struct got_object_idset *idset,
953     struct got_object_id *tree_id, const char *dpath, time_t mtime,
954     struct got_repository *repo, int loose_obj_only,
955     int *ncolored, int *nfound, int *ntrees,
956     got_pack_progress_cb progress_cb, void *progress_arg,
957     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
958 {
959 	const struct got_error *err = NULL;
960 	struct got_object_id_queue tree_ids;
961 	struct got_object_qid *qid;
962 
963 	if (got_object_idset_contains(idset, tree_id))
964 		return NULL;
965 
966 	err = got_object_qid_alloc(&qid, tree_id);
967 	if (err)
968 		return err;
969 
970 	STAILQ_INIT(&tree_ids);
971 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
972 
973 	while (!STAILQ_EMPTY(&tree_ids)) {
974 		if (cancel_cb) {
975 			err = (*cancel_cb)(cancel_arg);
976 			if (err)
977 				break;
978 		}
979 
980 		qid = STAILQ_FIRST(&tree_ids);
981 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
982 
983 		if (got_object_idset_contains(idset, qid->id)) {
984 			got_object_qid_free(qid);
985 			continue;
986 		}
987 
988 		err = add_object(want_meta, idset, qid->id, dpath,
989 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo,
990 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
991 		if (err) {
992 			got_object_qid_free(qid);
993 			break;
994 		}
995 
996 		err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
997 		    dpath, mtime, repo, loose_obj_only, ncolored, nfound,
998 		    ntrees, progress_cb, progress_arg, rl,
999 		    cancel_cb, cancel_arg);
1000 		got_object_qid_free(qid);
1001 		if (err)
1002 			break;
1003 	}
1004 
1005 	got_object_id_queue_free(&tree_ids);
1006 	return err;
1007 }
1008 
1009 static const struct got_error *
1010 load_commit(int want_meta, struct got_object_idset *idset,
1011     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1012     int *ncolored, int *nfound, int *ntrees,
1013     got_pack_progress_cb progress_cb, void *progress_arg,
1014     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1015 {
1016 	const struct got_error *err;
1017 	struct got_commit_object *commit;
1018 
1019 	if (got_object_idset_contains(idset, id))
1020 		return NULL;
1021 
1022 	if (loose_obj_only) {
1023 		int is_packed;
1024 		err = search_packidx(&is_packed, id, repo);
1025 		if (err)
1026 			return err;
1027 		if (is_packed && want_meta)
1028 			return NULL;
1029 	}
1030 
1031 	err = got_object_open_as_commit(&commit, repo, id);
1032 	if (err)
1033 		return err;
1034 
1035 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1036 	    got_object_commit_get_committer_time(commit),
1037 	    loose_obj_only, repo,
1038 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1039 	if (err)
1040 		goto done;
1041 
1042 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1043 	    "", got_object_commit_get_committer_time(commit),
1044 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1045 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1046 done:
1047 	got_object_commit_close(commit);
1048 	return err;
1049 }
1050 
1051 static const struct got_error *
1052 load_tag(int want_meta, struct got_object_idset *idset,
1053     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1054     int *ncolored, int *nfound, int *ntrees,
1055     got_pack_progress_cb progress_cb, void *progress_arg,
1056     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1057 {
1058 	const struct got_error *err;
1059 	struct got_tag_object *tag = NULL;
1060 
1061 	if (got_object_idset_contains(idset, id))
1062 		return NULL;
1063 
1064 	if (loose_obj_only) {
1065 		int is_packed;
1066 		err = search_packidx(&is_packed, id, repo);
1067 		if (err)
1068 			return err;
1069 		if (is_packed && want_meta)
1070 			return NULL;
1071 	}
1072 
1073 	err = got_object_open_as_tag(&tag, repo, id);
1074 	if (err)
1075 		return err;
1076 
1077 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1078 	    got_object_tag_get_tagger_time(tag), loose_obj_only, repo,
1079 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1080 	if (err)
1081 		goto done;
1082 
1083 	switch (got_object_tag_get_object_type(tag)) {
1084 	case GOT_OBJ_TYPE_COMMIT:
1085 		err = load_commit(want_meta, idset,
1086 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1087 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1088 		    cancel_cb, cancel_arg);
1089 		break;
1090 	case GOT_OBJ_TYPE_TREE:
1091 		err = load_tree(want_meta, idset,
1092 		    got_object_tag_get_object_id(tag), "",
1093 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1094 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1095 		    cancel_cb, cancel_arg);
1096 		break;
1097 	default:
1098 		break;
1099 	}
1100 
1101 done:
1102 	got_object_tag_close(tag);
1103 	return err;
1104 }
1105 
1106 enum findtwixt_color {
1107 	COLOR_KEEP = 0,
1108 	COLOR_DROP,
1109 	COLOR_BLANK,
1110 };
1111 static const int findtwixt_colors[] = {
1112 	COLOR_KEEP,
1113 	COLOR_DROP,
1114 	COLOR_BLANK
1115 };
1116 
1117 static const struct got_error *
1118 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1119     int color, struct got_repository *repo)
1120 {
1121 	const struct got_error *err;
1122 	struct got_object_qid *qid;
1123 
1124 	err = got_object_qid_alloc(&qid, id);
1125 	if (err)
1126 		return err;
1127 
1128 	STAILQ_INSERT_TAIL(ids, qid, entry);
1129 	qid->data = (void *)&findtwixt_colors[color];
1130 	return NULL;
1131 }
1132 
1133 static const struct got_error *
1134 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1135     struct got_object_id *id, struct got_repository *repo,
1136     got_cancel_cb cancel_cb, void *cancel_arg)
1137 {
1138 	const struct got_error *err = NULL;
1139 	struct got_commit_object *commit;
1140 	const struct got_object_id_queue *parents;
1141 	struct got_object_id_queue ids;
1142 	struct got_object_qid *qid;
1143 
1144 	STAILQ_INIT(&ids);
1145 
1146 	err = got_object_qid_alloc(&qid, id);
1147 	if (err)
1148 		return err;
1149 	STAILQ_INSERT_HEAD(&ids, qid, entry);
1150 
1151 	while (!STAILQ_EMPTY(&ids)) {
1152 		if (cancel_cb) {
1153 			err = (*cancel_cb)(cancel_arg);
1154 			if (err)
1155 				break;
1156 		}
1157 
1158 		qid = STAILQ_FIRST(&ids);
1159 		STAILQ_REMOVE_HEAD(&ids, entry);
1160 
1161 		if (got_object_idset_contains(drop, qid->id)) {
1162 			got_object_qid_free(qid);
1163 			continue;
1164 		}
1165 
1166 		err = got_object_idset_add(drop, qid->id, NULL);
1167 		if (err) {
1168 			got_object_qid_free(qid);
1169 			break;
1170 		}
1171 
1172 		if (!got_object_idset_contains(keep, qid->id)) {
1173 			got_object_qid_free(qid);
1174 			continue;
1175 		}
1176 
1177 		err = got_object_open_as_commit(&commit, repo, qid->id);
1178 		got_object_qid_free(qid);
1179 		if (err)
1180 			break;
1181 
1182 		parents = got_object_commit_get_parent_ids(commit);
1183 		if (parents) {
1184 			err = got_object_id_queue_copy(parents, &ids);
1185 			if (err) {
1186 				got_object_commit_close(commit);
1187 				break;
1188 			}
1189 		}
1190 		got_object_commit_close(commit);
1191 	}
1192 
1193 	got_object_id_queue_free(&ids);
1194 	return err;
1195 }
1196 
1197 struct append_id_arg {
1198 	struct got_object_id **array;
1199 	int idx;
1200 };
1201 
1202 static const struct got_error *
1203 append_id(struct got_object_id *id, void *data, void *arg)
1204 {
1205 	struct append_id_arg *a = arg;
1206 
1207 	a->array[a->idx] = got_object_id_dup(id);
1208 	if (a->array[a->idx] == NULL)
1209 		return got_error_from_errno("got_object_id_dup");
1210 
1211 	a->idx++;
1212 	return NULL;
1213 }
1214 
1215 static const struct got_error *
1216 queue_commit_or_tag_id(struct got_object_id *id, int color,
1217     struct got_object_id_queue *ids, struct got_repository *repo)
1218 {
1219 	const struct got_error *err;
1220 	struct got_tag_object *tag = NULL;
1221 	int obj_type;
1222 
1223 	err = got_object_get_type(&obj_type, repo, id);
1224 	if (err)
1225 		return err;
1226 
1227 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1228 		err = got_object_open_as_tag(&tag, repo, id);
1229 		if (err)
1230 			return err;
1231 		obj_type = got_object_tag_get_object_type(tag);
1232 		id = got_object_tag_get_object_id(tag);
1233 	}
1234 
1235 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1236 		err = queue_commit_id(ids, id, color, repo);
1237 		if (err)
1238 			goto done;
1239 	}
1240 done:
1241 	if (tag)
1242 		got_object_tag_close(tag);
1243 	return err;
1244 }
1245 
1246 static const struct got_error *
1247 color_commits(int *ncolored, struct got_object_id_queue *ids,
1248     struct got_object_idset *keep, struct got_object_idset *drop,
1249     struct got_repository *repo,
1250     got_pack_progress_cb progress_cb, void *progress_arg,
1251     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1252 {
1253 	const struct got_error *err = NULL;
1254 	struct got_commit_object *commit = NULL;
1255 	struct got_object_qid *qid;
1256 
1257 	while (!STAILQ_EMPTY(ids)) {
1258 		int qcolor, ncolor;
1259 
1260 		if (cancel_cb) {
1261 			err = cancel_cb(cancel_arg);
1262 			if (err)
1263 				break;
1264 		}
1265 
1266 		qid = STAILQ_FIRST(ids);
1267 		qcolor = *((int *)qid->data);
1268 
1269 		if (got_object_idset_contains(drop, qid->id))
1270 			ncolor = COLOR_DROP;
1271 		else if (got_object_idset_contains(keep, qid->id))
1272 			ncolor = COLOR_KEEP;
1273 		else
1274 			ncolor = COLOR_BLANK;
1275 
1276 		(*ncolored)++;
1277 		err = report_progress(progress_cb, progress_arg, rl,
1278 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1279 		if (err)
1280 			break;
1281 
1282 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1283 		    qcolor == COLOR_KEEP)) {
1284 			STAILQ_REMOVE_HEAD(ids, entry);
1285 			got_object_qid_free(qid);
1286 			continue;
1287 		}
1288 
1289 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1290 			err = drop_commit(keep, drop, qid->id, repo,
1291 			    cancel_cb, cancel_arg);
1292 			if (err)
1293 				break;
1294 		} else if (ncolor == COLOR_BLANK) {
1295 			struct got_commit_object *commit;
1296 			const struct got_object_id_queue *parents;
1297 			struct got_object_qid *pid;
1298 
1299 			if (qcolor == COLOR_KEEP)
1300 				err = got_object_idset_add(keep, qid->id, NULL);
1301 			else
1302 				err = got_object_idset_add(drop, qid->id, NULL);
1303 			if (err)
1304 				break;
1305 
1306 			err = got_object_open_as_commit(&commit, repo, qid->id);
1307 			if (err)
1308 				break;
1309 
1310 			parents = got_object_commit_get_parent_ids(commit);
1311 			if (parents) {
1312 				STAILQ_FOREACH(pid, parents, entry) {
1313 					err = queue_commit_id(ids, pid->id,
1314 					    qcolor, repo);
1315 					if (err)
1316 						break;
1317 				}
1318 			}
1319 			got_object_commit_close(commit);
1320 			commit = NULL;
1321 		} else {
1322 			/* should not happen */
1323 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1324 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1325 			break;
1326 		}
1327 
1328 		STAILQ_REMOVE_HEAD(ids, entry);
1329 		got_object_qid_free(qid);
1330 	}
1331 
1332 	if (commit)
1333 		got_object_commit_close(commit);
1334 	return err;
1335 }
1336 
1337 static const struct got_error *
1338 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1339     struct got_object_id **head, int nhead,
1340     struct got_object_id **tail, int ntail,
1341     struct got_repository *repo,
1342     got_pack_progress_cb progress_cb, void *progress_arg,
1343     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1344 {
1345 	const struct got_error *err = NULL;
1346 	struct got_object_id_queue ids;
1347 	struct got_object_idset *keep, *drop;
1348 	int i, nkeep;
1349 
1350 	STAILQ_INIT(&ids);
1351 	*res = NULL;
1352 	*nres = 0;
1353 	*ncolored = 0;
1354 
1355 	keep = got_object_idset_alloc();
1356 	if (keep == NULL)
1357 		return got_error_from_errno("got_object_idset_alloc");
1358 
1359 	drop = got_object_idset_alloc();
1360 	if (drop == NULL) {
1361 		err = got_error_from_errno("got_object_idset_alloc");
1362 		goto done;
1363 	}
1364 
1365 	for (i = 0; i < nhead; i++) {
1366 		struct got_object_id *id = head[i];
1367 		if (id == NULL)
1368 			continue;
1369 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1370 		if (err)
1371 			goto done;
1372 	}		
1373 
1374 	for (i = 0; i < ntail; i++) {
1375 		struct got_object_id *id = tail[i];
1376 		if (id == NULL)
1377 			continue;
1378 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1379 		if (err)
1380 			goto done;
1381 	}
1382 
1383 	err = color_commits(ncolored, &ids, keep, drop, repo,
1384 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1385 	if (err)
1386 		goto done;
1387 
1388 	nkeep = got_object_idset_num_elements(keep);
1389 	if (nkeep > 0) {
1390 		struct append_id_arg arg;
1391 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1392 		if (arg.array == NULL) {
1393 			err = got_error_from_errno("calloc");
1394 			goto done;
1395 		}
1396 		arg.idx = 0;
1397 		err = got_object_idset_for_each(keep, append_id, &arg);
1398 		if (err) {
1399 			free(arg.array);
1400 			goto done;
1401 		}
1402 		*res = arg.array;
1403 		*nres = nkeep;
1404 	}
1405 done:
1406 	got_object_idset_free(keep);
1407 	got_object_idset_free(drop);
1408 	got_object_id_queue_free(&ids);
1409 	return err;
1410 }
1411 
1412 static const struct got_error *
1413 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1414     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1415     struct got_object_id **ours, int nours, struct got_repository *repo,
1416     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1417     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1418 {
1419 	const struct got_error *err = NULL;
1420 	struct got_object_id **ids = NULL;
1421 	int i, nobj = 0, obj_type;
1422 
1423 	*ncolored = 0;
1424 	*nfound = 0;
1425 	*ntrees = 0;
1426 
1427 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1428 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1429 	if (err || nobj == 0)
1430 		goto done;
1431 
1432 	for (i = 0; i < ntheirs; i++) {
1433 		struct got_object_id *id = theirs[i];
1434 		if (id == NULL)
1435 			continue;
1436 		err = got_object_get_type(&obj_type, repo, id);
1437 		if (err)
1438 			return err;
1439 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1440 			err = load_commit(0, idset, id, repo,
1441 			    loose_obj_only, ncolored, nfound, ntrees,
1442 			    progress_cb, progress_arg, rl,
1443 			    cancel_cb, cancel_arg);
1444 			if (err)
1445 				goto done;
1446 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1447 			err = load_tag(0, idset, id, repo,
1448 			    loose_obj_only, ncolored, nfound, ntrees,
1449 			    progress_cb, progress_arg, rl,
1450 			    cancel_cb, cancel_arg);
1451 			if (err)
1452 				goto done;
1453 		}
1454 	}
1455 
1456 	for (i = 0; i < nobj; i++) {
1457 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1458 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1459 		    cancel_cb, cancel_arg);
1460 		if (err)
1461 			goto done;
1462 	}
1463 
1464 	for (i = 0; i < nours; i++) {
1465 		struct got_object_id *id = ours[i];
1466 		struct got_pack_meta *m;
1467 		if (id == NULL)
1468 			continue;
1469 		m = got_object_idset_get(idset, id);
1470 		if (m == NULL) {
1471 			err = got_object_get_type(&obj_type, repo, id);
1472 			if (err)
1473 				goto done;
1474 		} else
1475 			obj_type = m->obj_type;
1476 		if (obj_type != GOT_OBJ_TYPE_TAG)
1477 			continue;
1478 		err = load_tag(1, idset, id, repo, loose_obj_only,
1479 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1480 		    cancel_cb, cancel_arg);
1481 		if (err)
1482 			goto done;
1483 	}
1484 done:
1485 	for (i = 0; i < nobj; i++) {
1486 		free(ids[i]);
1487 	}
1488 	free(ids);
1489 	return err;
1490 }
1491 
1492 const struct got_error *
1493 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1494 {
1495 	size_t n;
1496 
1497 	SHA1Update(ctx, buf, len);
1498 	n = fwrite(buf, 1, len, f);
1499 	if (n != len)
1500 		return got_ferror(f, GOT_ERR_IO);
1501 	return NULL;
1502 }
1503 
1504 static void
1505 putbe32(char *b, uint32_t n)
1506 {
1507 	b[0] = n >> 24;
1508 	b[1] = n >> 16;
1509 	b[2] = n >> 8;
1510 	b[3] = n >> 0;
1511 }
1512 
1513 static int
1514 write_order_cmp(const void *pa, const void *pb)
1515 {
1516 	struct got_pack_meta *a, *b, *ahd, *bhd;
1517 
1518 	a = *(struct got_pack_meta **)pa;
1519 	b = *(struct got_pack_meta **)pb;
1520 	ahd = (a->head == NULL) ? a : a->head;
1521 	bhd = (b->head == NULL) ? b : b->head;
1522 	if (ahd->mtime != bhd->mtime)
1523 		return bhd->mtime - ahd->mtime;
1524 	if (ahd != bhd)
1525 		return (uintptr_t)bhd - (uintptr_t)ahd;
1526 	if (a->nchain != b->nchain)
1527 		return a->nchain - b->nchain;
1528 	return a->mtime - b->mtime;
1529 }
1530 
1531 static int
1532 reuse_write_order_cmp(const void *pa, const void *pb)
1533 {
1534 	struct got_pack_meta *a, *b;
1535 
1536 	a = *(struct got_pack_meta **)pa;
1537 	b = *(struct got_pack_meta **)pb;
1538 
1539 	if (a->reused_delta_offset < b->reused_delta_offset)
1540 		return -1;
1541 	if (a->reused_delta_offset > b->reused_delta_offset)
1542 		return 1;
1543 	return 0;
1544 }
1545 
1546 static const struct got_error *
1547 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1548 {
1549 	size_t i;
1550 
1551 	*hdrlen = 0;
1552 
1553 	hdr[0] = obj_type << 4;
1554 	hdr[0] |= len & 0xf;
1555 	len >>= 4;
1556 	for (i = 1; len != 0; i++){
1557 		if (i >= bufsize)
1558 			return got_error(GOT_ERR_NO_SPACE);
1559 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1560 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1561 		len >>= GOT_DELTA_SIZE_SHIFT;
1562 	}
1563 
1564 	*hdrlen = i;
1565 	return NULL;
1566 }
1567 
1568 static int
1569 packoff(char *hdr, off_t off)
1570 {
1571 	int i, j;
1572 	char rbuf[8];
1573 
1574 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1575 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1576 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1577 		    GOT_DELTA_SIZE_MORE;
1578 	}
1579 
1580 	j = 0;
1581 	while (i > 0)
1582 		hdr[j++] = rbuf[--i];
1583 	return j;
1584 }
1585 
1586 static const struct got_error *
1587 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1588     struct got_pack_meta *m)
1589 {
1590 	const struct got_error *err;
1591 	char buf[32];
1592 	int nh;
1593 
1594 	if (m->prev->off != 0) {
1595 		err = packhdr(&nh, buf, sizeof(buf),
1596 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1597 		if (err)
1598 			return err;
1599 		nh += packoff(buf + nh, m->off - m->prev->off);
1600 		err = hwrite(packfile, buf, nh, ctx);
1601 		if (err)
1602 			return err;
1603 		*packfile_size += nh;
1604 	} else {
1605 		err = packhdr(&nh, buf, sizeof(buf),
1606 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1607 		if (err)
1608 			return err;
1609 		err = hwrite(packfile, buf, nh, ctx);
1610 		if (err)
1611 			return err;
1612 		*packfile_size += nh;
1613 		err = hwrite(packfile, m->prev->id.sha1,
1614 		    sizeof(m->prev->id.sha1), ctx);
1615 		if (err)
1616 			return err;
1617 		*packfile_size += sizeof(m->prev->id.sha1);
1618 	}
1619 
1620 	return NULL;
1621 }
1622 
1623 static const struct got_error *
1624 write_packed_object(off_t *packfile_size, FILE *packfile,
1625     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1626     SHA1_CTX *ctx, struct got_repository *repo)
1627 {
1628 	const struct got_error *err = NULL;
1629 	struct got_deflate_checksum csum;
1630 	char buf[32];
1631 	int nh;
1632 	struct got_raw_object *raw = NULL;
1633 	off_t outlen;
1634 
1635 	csum.output_sha1 = ctx;
1636 	csum.output_crc = NULL;
1637 
1638 	m->off = ftello(packfile);
1639 	if (m->delta_len == 0) {
1640 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1641 		if (err)
1642 			goto done;
1643 		err = packhdr(&nh, buf, sizeof(buf),
1644 		    m->obj_type, raw->size);
1645 		if (err)
1646 			goto done;
1647 		err = hwrite(packfile, buf, nh, ctx);
1648 		if (err)
1649 			goto done;
1650 		*packfile_size += nh;
1651 		if (raw->f == NULL) {
1652 			err = got_deflate_to_file_mmap(&outlen,
1653 			    raw->data + raw->hdrlen, 0, raw->size,
1654 			    packfile, &csum);
1655 			if (err)
1656 				goto done;
1657 		} else {
1658 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1659 			    == -1) {
1660 				err = got_error_from_errno("fseeko");
1661 				goto done;
1662 			}
1663 			err = got_deflate_to_file(&outlen, raw->f,
1664 			    raw->size, packfile, &csum);
1665 			if (err)
1666 				goto done;
1667 		}
1668 		*packfile_size += outlen;
1669 		got_object_raw_close(raw);
1670 		raw = NULL;
1671 	} else if (m->delta_buf) {
1672 		err = deltahdr(packfile_size, ctx, packfile, m);
1673 		if (err)
1674 			goto done;
1675 		err = got_deflate_to_file_mmap(&outlen,
1676 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1677 		if (err)
1678 			goto done;
1679 		*packfile_size += outlen;
1680 		free(m->delta_buf);
1681 		m->delta_buf = NULL;
1682 	} else {
1683 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1684 		    == -1) {
1685 			err = got_error_from_errno("fseeko");
1686 			goto done;
1687 		}
1688 		err = deltahdr(packfile_size, ctx, packfile, m);
1689 		if (err)
1690 			goto done;
1691 		err = got_deflate_to_file(&outlen, delta_cache,
1692 		    m->delta_len, packfile, &csum);
1693 		if (err)
1694 			goto done;
1695 		*packfile_size += outlen;
1696 	}
1697 done:
1698 	if (raw)
1699 		got_object_raw_close(raw);
1700 	return err;
1701 }
1702 
1703 static const struct got_error *
1704 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1705     struct got_pack_meta **deltify, int ndeltify,
1706     struct got_pack_meta **reuse, int nreuse,
1707     int ncolored, int nfound, int ntrees, int nours,
1708     struct got_repository *repo,
1709     got_pack_progress_cb progress_cb, void *progress_arg,
1710     struct got_ratelimit *rl,
1711     got_cancel_cb cancel_cb, void *cancel_arg)
1712 {
1713 	const struct got_error *err = NULL;
1714 	int i;
1715 	SHA1_CTX ctx;
1716 	struct got_pack_meta *m;
1717 	char buf[32];
1718 	size_t n;
1719 	off_t packfile_size = 0;
1720 	int outfd = -1;
1721 
1722 	SHA1Init(&ctx);
1723 
1724 	err = hwrite(packfile, "PACK", 4, &ctx);
1725 	if (err)
1726 		return err;
1727 	putbe32(buf, GOT_PACKFILE_VERSION);
1728 	err = hwrite(packfile, buf, 4, &ctx);
1729 	if (err)
1730 		goto done;
1731 	putbe32(buf, ndeltify + nreuse);
1732 	err = hwrite(packfile, buf, 4, &ctx);
1733 	if (err)
1734 		goto done;
1735 
1736 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1737 	    write_order_cmp);
1738 	for (i = 0; i < ndeltify; i++) {
1739 		err = report_progress(progress_cb, progress_arg, rl,
1740 		    ncolored, nfound, ntrees, packfile_size, nours,
1741 		    ndeltify + nreuse, ndeltify + nreuse, i);
1742 		if (err)
1743 			goto done;
1744 		m = deltify[i];
1745 		err = write_packed_object(&packfile_size, packfile,
1746 		    delta_cache, m, &outfd, &ctx, repo);
1747 		if (err)
1748 			goto done;
1749 	}
1750 
1751 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1752 	    reuse_write_order_cmp);
1753 	for (i = 0; i < nreuse; i++) {
1754 		err = report_progress(progress_cb, progress_arg, rl,
1755 		    ncolored, nfound, ntrees, packfile_size, nours,
1756 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1757 		if (err)
1758 			goto done;
1759 		m = reuse[i];
1760 		err = write_packed_object(&packfile_size, packfile,
1761 		    delta_cache, m, &outfd, &ctx, repo);
1762 		if (err)
1763 			goto done;
1764 	}
1765 
1766 	SHA1Final(pack_sha1, &ctx);
1767 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1768 	if (n != SHA1_DIGEST_LENGTH)
1769 		err = got_ferror(packfile, GOT_ERR_IO);
1770 	packfile_size += SHA1_DIGEST_LENGTH;
1771 	packfile_size += sizeof(struct got_packfile_hdr);
1772 	if (progress_cb) {
1773 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1774 		    packfile_size, nours, ndeltify + nreuse,
1775 		    ndeltify + nreuse, ndeltify + nreuse);
1776 		if (err)
1777 			goto done;
1778 	}
1779 done:
1780 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1781 		err = got_error_from_errno("close");
1782 	return err;
1783 }
1784 
1785 static const struct got_error *
1786 remove_unused_object(struct got_object_idset_element *entry, void *arg)
1787 {
1788 	struct got_object_idset *idset = arg;
1789 
1790 	if (got_object_idset_get_element_data(entry) == NULL)
1791 		got_object_idset_remove_element(idset, entry);
1792 
1793 	return NULL;
1794 }
1795 
1796 static const struct got_error *
1797 remove_reused_object(struct got_object_idset_element *entry, void *arg)
1798 {
1799 	struct got_object_idset *idset = arg;
1800 	struct got_pack_meta *m;
1801 
1802 	m = got_object_idset_get_element_data(entry);
1803 	if (m->have_reused_delta)
1804 		got_object_idset_remove_element(idset, entry);
1805 
1806 	return NULL;
1807 }
1808 
1809 static const struct got_error *
1810 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1811 {
1812 	struct got_pack_meta *m = data;
1813 	struct got_pack_metavec *v = arg;
1814 
1815 	return add_meta(m, v);
1816 }
1817 
1818 const struct got_error *
1819 got_pack_create(uint8_t *packsha1, FILE *packfile,
1820     struct got_object_id **theirs, int ntheirs,
1821     struct got_object_id **ours, int nours,
1822     struct got_repository *repo, int loose_obj_only, int allow_empty,
1823     got_pack_progress_cb progress_cb, void *progress_arg,
1824     got_cancel_cb cancel_cb, void *cancel_arg)
1825 {
1826 	const struct got_error *err;
1827 	int delta_cache_fd = -1;
1828 	FILE *delta_cache = NULL;
1829 	struct got_object_idset *idset;
1830 	struct got_ratelimit rl;
1831 	struct got_pack_metavec deltify, reuse;
1832 	int ncolored = 0, nfound = 0, ntrees = 0;
1833 
1834 	memset(&deltify, 0, sizeof(deltify));
1835 	memset(&reuse, 0, sizeof(reuse));
1836 
1837 	got_ratelimit_init(&rl, 0, 500);
1838 
1839 	idset = got_object_idset_alloc();
1840 	if (idset == NULL)
1841 		return got_error_from_errno("got_object_idset_alloc");
1842 
1843 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1844 	    ntheirs, ours, nours, repo, loose_obj_only,
1845 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1846 	if (err)
1847 		return err;
1848 
1849 	err = got_object_idset_for_each_element(idset,
1850 	    remove_unused_object, idset);
1851 	if (err)
1852 		goto done;
1853 
1854 	if (progress_cb) {
1855 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1856 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1857 		if (err)
1858 			goto done;
1859 	}
1860 
1861 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1862 		err = got_error(GOT_ERR_CANNOT_PACK);
1863 		goto done;
1864 	}
1865 
1866 	delta_cache_fd = got_opentempfd();
1867 	if (delta_cache_fd == -1) {
1868 		err = got_error_from_errno("got_opentemp");
1869 		goto done;
1870 	}
1871 
1872 	reuse.metasz = 64;
1873 	reuse.meta = calloc(reuse.metasz,
1874 	    sizeof(struct got_pack_meta *));
1875 	if (reuse.meta == NULL) {
1876 		err = got_error_from_errno("calloc");
1877 		goto done;
1878 	}
1879 
1880 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1881 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1882 	    cancel_cb, cancel_arg);
1883 	if (err)
1884 		goto done;
1885 	if (reuse.nmeta > 0) {
1886 		err = got_object_idset_for_each_element(idset,
1887 		    remove_reused_object, idset);
1888 		if (err)
1889 			goto done;
1890 	}
1891 
1892 	delta_cache = fdopen(delta_cache_fd, "a+");
1893 	if (delta_cache == NULL) {
1894 		err = got_error_from_errno("fdopen");
1895 		goto done;
1896 	}
1897 	delta_cache_fd = -1;
1898 
1899 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1900 		err = got_error_from_errno("fseeko");
1901 		goto done;
1902 	}
1903 
1904 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1905 	    sizeof(struct got_pack_meta *));
1906 	if (deltify.meta == NULL) {
1907 		err = got_error_from_errno("calloc");
1908 		goto done;
1909 	}
1910 	deltify.metasz = got_object_idset_num_elements(idset);
1911 
1912 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1913 	if (err)
1914 		goto done;
1915 	if (deltify.nmeta > 0) {
1916 		err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1917 		    nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1918 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1919 		if (err)
1920 			goto done;
1921 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1922 			err = got_error_from_errno("fseeko");
1923 			goto done;
1924 		}
1925 	}
1926 
1927 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1928 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1929 	    nours, repo, progress_cb, progress_arg, &rl,
1930 	    cancel_cb, cancel_arg);
1931 	if (err)
1932 		goto done;
1933 done:
1934 	free_nmeta(deltify.meta, deltify.nmeta);
1935 	free_nmeta(reuse.meta, reuse.nmeta);
1936 	got_object_idset_free(idset);
1937 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1938 		err = got_error_from_errno("close");
1939 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1940 		err = got_error_from_errno("fclose");
1941 	return err;
1942 }