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 <limits.h>
29 #include <zlib.h>
30 
31 #if defined(__FreeBSD__)
32 #include <unistd.h>
33 #endif
34 
35 #include "got_error.h"
36 #include "got_cancel.h"
37 #include "got_object.h"
38 #include "got_path.h"
39 #include "got_reference.h"
40 #include "got_repository_admin.h"
41 #include "got_opentemp.h"
42 
43 #include "got_lib_deltify.h"
44 #include "got_lib_delta.h"
45 #include "got_lib_object.h"
46 #include "got_lib_object_idset.h"
47 #include "got_lib_object_cache.h"
48 #include "got_lib_deflate.h"
49 #include "got_lib_pack.h"
50 #include "got_lib_privsep.h"
51 #include "got_lib_repository.h"
52 #include "got_lib_ratelimit.h"
53 
54 #ifndef MIN
55 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
56 #endif
57 
58 #ifndef MAX
59 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
60 #endif
61 
62 struct got_pack_meta {
63 	struct got_object_id id;
64 	char	*path;
65 	int	obj_type;
66 	off_t	size;
67 	time_t	mtime;
68 
69 	/* The best delta we picked */
70 	struct got_pack_meta *head;
71 	struct got_pack_meta *prev;
72 	unsigned char *delta_buf; /* if not encoded in delta cache file */
73 	off_t	delta_offset;	/* offset in delta cache file */
74 	off_t	delta_len;	/* encoded delta length */
75 	int	nchain;
76 
77 	int	have_reused_delta;
78 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
79 	struct got_object_id *base_obj_id;
80 
81 	/* Only used for delta window */
82 	struct got_delta_table *dtab;
83 
84 	/* Only used for writing offset deltas */
85 	off_t	off;
86 };
87 
88 struct got_pack_metavec {
89 	struct got_pack_meta **meta;
90 	int nmeta;
91 	int metasz;
92 };
93 
94 static const struct got_error *
95 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
96     const char *path, int obj_type, time_t mtime)
97 {
98 	const struct got_error *err = NULL;
99 	struct got_pack_meta *m;
100 
101 	*new = NULL;
102 
103 	m = calloc(1, sizeof(*m));
104 	if (m == NULL)
105 		return got_error_from_errno("calloc");
106 
107 	memcpy(&m->id, id, sizeof(m->id));
108 
109 	m->path = strdup(path);
110 	if (m->path == NULL) {
111 		err = got_error_from_errno("strdup");
112 		free(m);
113 		return err;
114 	}
115 
116 	m->obj_type = obj_type;
117 	m->mtime = mtime;
118 	*new = m;
119 	return NULL;
120 }
121 
122 static void
123 clear_meta(struct got_pack_meta *meta)
124 {
125 	if (meta == NULL)
126 		return;
127 	free(meta->path);
128 	meta->path = NULL;
129 	free(meta->delta_buf);
130 	meta->delta_buf = NULL;
131 	free(meta->base_obj_id);
132 	meta->base_obj_id = NULL;
133 }
134 
135 static void
136 free_nmeta(struct got_pack_meta **meta, int nmeta)
137 {
138 	int i;
139 
140 	for (i = 0; i < nmeta; i++)
141 		clear_meta(meta[i]);
142 	free(meta);
143 }
144 
145 static int
146 delta_order_cmp(const void *pa, const void *pb)
147 {
148 	struct got_pack_meta *a, *b;
149 	int cmp;
150 
151 	a = *(struct got_pack_meta **)pa;
152 	b = *(struct got_pack_meta **)pb;
153 
154 	if (a->obj_type != b->obj_type)
155 		return a->obj_type - b->obj_type;
156 	cmp = strcmp(a->path, b->path);
157 	if (cmp != 0)
158 		return cmp;
159 	if (a->mtime != b->mtime)
160 		return a->mtime - b->mtime;
161 	return got_object_id_cmp(&a->id, &b->id);
162 }
163 
164 static off_t
165 delta_size(struct got_delta_instruction *deltas, int ndeltas)
166 {
167 	int i;
168 	off_t size = 32;
169 	for (i = 0; i < ndeltas; i++) {
170 		if (deltas[i].copy)
171 			size += GOT_DELTA_SIZE_SHIFT;
172 		else
173 			size += deltas[i].len + 1;
174 	}
175 	return size;
176 }
177 
178 static const struct got_error *
179 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
180 {
181 	char *n;
182 
183 	if (*len + nseg >= *sz) {
184 		while (*len + nseg >= *sz)
185 			*sz += *sz / 2;
186 		n = realloc(*p, *sz);
187 		if (n == NULL)
188 			return got_error_from_errno("realloc");
189 		*p = n;
190 	}
191 	memcpy(*p + *len, seg, nseg);
192 	*len += nseg;
193 	return NULL;
194 }
195 
196 static const struct got_error *
197 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
198     struct got_delta_instruction *deltas, int ndeltas,
199     off_t delta_size, off_t base_size)
200 {
201 	const struct got_error *err;
202 	unsigned char buf[16], *bp;
203 	int i, j;
204 	size_t len = 0;
205 	off_t n;
206 	struct got_delta_instruction *d;
207 
208 	m->delta_buf = malloc(delta_size);
209 	if (m->delta_buf == NULL)
210 		return got_error_from_errno("calloc");
211 
212 	/* base object size */
213 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
214 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
215 	for (i = 1; n > 0; i++) {
216 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
217 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
218 		n >>= GOT_DELTA_SIZE_SHIFT;
219 	}
220 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
221 	if (err)
222 		return err;
223 
224 	/* target object size */
225 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
226 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
227 	for (i = 1; n > 0; i++) {
228 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
229 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
230 		n >>= GOT_DELTA_SIZE_SHIFT;
231 	}
232 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
233 	if (err)
234 		return err;
235 
236 	for (j = 0; j < ndeltas; j++) {
237 		d = &deltas[j];
238 		if (d->copy) {
239 			n = d->offset;
240 			bp = &buf[1];
241 			buf[0] = GOT_DELTA_BASE_COPY;
242 			for (i = 0; i < 4; i++) {
243 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
244 				buf[0] |= 1 << i;
245 				*bp++ = n & 0xff;
246 				n >>= 8;
247 				if (n == 0)
248 					break;
249 			}
250 
251 			n = d->len;
252 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
253 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
254 				for (i = 0; i < 3 && n > 0; i++) {
255 					buf[0] |= 1 << (i + 4);
256 					*bp++ = n & 0xff;
257 					n >>= 8;
258 				}
259 			}
260 			err = append(&m->delta_buf, &len, &delta_size,
261 			    buf, bp - buf);
262 			if (err)
263 				return err;
264 		} else if (o->f == NULL) {
265 			n = 0;
266 			while (n != d->len) {
267 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
268 				err = append(&m->delta_buf, &len, &delta_size,
269 				    buf, 1);
270 				if (err)
271 					return err;
272 				err = append(&m->delta_buf, &len, &delta_size,
273 				    o->data + o->hdrlen + d->offset + n,
274 				    buf[0]);
275 				if (err)
276 					return err;
277 				n += buf[0];
278 			}
279 		} else {
280 			char content[128];
281 			size_t r;
282 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
283 				return got_error_from_errno("fseeko");
284 			n = 0;
285 			while (n != d->len) {
286 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
287 				err = append(&m->delta_buf, &len, &delta_size,
288 				    buf, 1);
289 				if (err)
290 					return err;
291 				r = fread(content, 1, buf[0], o->f);
292 				if (r != buf[0])
293 					return got_ferror(o->f, GOT_ERR_IO);
294 				err = append(&m->delta_buf, &len, &delta_size,
295 				    content, buf[0]);
296 				if (err)
297 					return err;
298 				n += buf[0];
299 			}
300 		}
301 	}
302 
303 	m->delta_len = len;
304 	return NULL;
305 }
306 
307 static const struct got_error *
308 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
309     struct got_delta_instruction *deltas, int ndeltas,
310     off_t base_size, FILE *f)
311 {
312 	unsigned char buf[16], *bp;
313 	int i, j;
314 	off_t n;
315 	size_t w;
316 	struct got_delta_instruction *d;
317 
318 	/* base object size */
319 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
320 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
321 	for (i = 1; n > 0; i++) {
322 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
323 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
324 		n >>= GOT_DELTA_SIZE_SHIFT;
325 	}
326 	w = fwrite(buf, 1, i, f);
327 	if (w != i)
328 		return got_ferror(f, GOT_ERR_IO);
329 
330 	/* target object size */
331 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
332 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
333 	for (i = 1; n > 0; i++) {
334 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
335 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
336 		n >>= GOT_DELTA_SIZE_SHIFT;
337 	}
338 	w = fwrite(buf, 1, i, f);
339 	if (w != i)
340 		return got_ferror(f, GOT_ERR_IO);
341 
342 	for (j = 0; j < ndeltas; j++) {
343 		d = &deltas[j];
344 		if (d->copy) {
345 			n = d->offset;
346 			bp = &buf[1];
347 			buf[0] = GOT_DELTA_BASE_COPY;
348 			for (i = 0; i < 4; i++) {
349 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
350 				buf[0] |= 1 << i;
351 				*bp++ = n & 0xff;
352 				n >>= 8;
353 				if (n == 0)
354 					break;
355 			}
356 
357 			n = d->len;
358 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
359 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
360 				for (i = 0; i < 3 && n > 0; i++) {
361 					buf[0] |= 1 << (i + 4);
362 					*bp++ = n & 0xff;
363 					n >>= 8;
364 				}
365 			}
366 			w = fwrite(buf, 1, bp - buf, f);
367 			if (w != bp - buf)
368 				return got_ferror(f, GOT_ERR_IO);
369 		} else if (o->f == NULL) {
370 			n = 0;
371 			while (n != d->len) {
372 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
373 				w = fwrite(buf, 1, 1, f);
374 				if (w != 1)
375 					return got_ferror(f, GOT_ERR_IO);
376 				w = fwrite(o->data + o->hdrlen + d->offset + n,
377 				    1, buf[0], f);
378 				if (w != buf[0])
379 					return got_ferror(f, GOT_ERR_IO);
380 				n += buf[0];
381 			}
382 		} else {
383 			char content[128];
384 			size_t r;
385 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
386 				return got_error_from_errno("fseeko");
387 			n = 0;
388 			while (n != d->len) {
389 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
390 				w = fwrite(buf, 1, 1, f);
391 				if (w != 1)
392 					return got_ferror(f, GOT_ERR_IO);
393 				r = fread(content, 1, buf[0], o->f);
394 				if (r != buf[0])
395 					return got_ferror(o->f, GOT_ERR_IO);
396 				w = fwrite(content, 1, buf[0], f);
397 				if (w != buf[0])
398 					return got_ferror(f, GOT_ERR_IO);
399 				n += buf[0];
400 			}
401 		}
402 	}
403 
404 	m->delta_len = ftello(f) - m->delta_offset;
405 	return NULL;
406 }
407 
408 static const struct got_error *
409 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
410     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
411     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
412     int nobj_written)
413 {
414 	const struct got_error *err;
415 	int elapsed;
416 
417 	if (progress_cb == NULL)
418 		return NULL;
419 
420 	err = got_ratelimit_check(&elapsed, rl);
421 	if (err || !elapsed)
422 		return err;
423 
424 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
425 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
426 }
427 
428 static const struct got_error *
429 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
430 {
431 	if (v->nmeta == v->metasz){
432 		size_t newsize = 2 * v->metasz;
433 		struct got_pack_meta **new;
434 		new = reallocarray(v->meta, newsize, sizeof(*new));
435 		if (new == NULL)
436 			return got_error_from_errno("reallocarray");
437 		v->meta = new;
438 		v->metasz = newsize; 
439 	}
440 
441 	v->meta[v->nmeta++] = m;
442 	return NULL;
443 }
444 
445 static const struct got_error *
446 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
447     struct got_object_idset *idset, struct got_pack *pack,
448     struct got_packidx *packidx, int delta_cache_fd,
449     struct got_repository *repo)
450 {
451 	const struct got_error *err = NULL;
452 	struct got_pack_meta *base = NULL;
453 	struct got_object_id *base_obj_id = NULL;
454 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
455 	uint64_t base_size, result_size;
456 
457 	if (m->have_reused_delta)
458 		return NULL;
459 
460 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
461 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
462 	    packidx, idx, &m->id, repo);
463 	if (err)
464 		return err;
465 
466 	if (delta_offset + delta_len < delta_offset)
467 		return got_error(GOT_ERR_BAD_PACKFILE);
468 
469 	base = got_object_idset_get(idset, base_obj_id);
470 	if (base == NULL)
471 		goto done;
472 
473 	m->delta_len = delta_len;
474 	m->delta_offset = delta_cache_offset;
475 	m->prev = base;
476 	m->size = result_size;
477 	m->have_reused_delta = 1;
478 	m->reused_delta_offset = delta_offset;
479 	m->base_obj_id = base_obj_id;
480 	base_obj_id = NULL;
481 	err = add_meta(m, v);
482 done:
483 	free(base_obj_id);
484 	return err;
485 }
486 
487 static const struct got_error *
488 find_pack_for_reuse(struct got_packidx **best_packidx,
489     struct got_repository *repo)
490 {
491 	const struct got_error *err = NULL;
492 	struct got_pathlist_entry *pe;
493 	const char *best_packidx_path = NULL;
494 	int nobj_max = 0;
495 
496 	*best_packidx = NULL;
497 
498 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
499 		const char *path_packidx = pe->path;
500 		struct got_packidx *packidx;
501 		int nobj;
502 
503 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
504 		if (err)
505 			break;
506 
507 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
508 		if (nobj > nobj_max) {
509 			best_packidx_path = path_packidx;
510 			nobj_max = nobj;
511 		}
512 	}
513 
514 	if (best_packidx_path) {
515 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
516 		    repo);
517 	}
518 
519 	return err;
520 }
521 
522 struct search_deltas_arg {
523 	struct got_packidx *packidx;
524 	struct got_pack *pack;
525 	struct got_object_idset *idset;
526 	struct got_pack_metavec *v;
527 	int delta_cache_fd;
528 	struct got_repository *repo;
529 	got_pack_progress_cb progress_cb;
530 	void *progress_arg;
531 	struct got_ratelimit *rl;
532 	got_cancel_cb cancel_cb;
533 	void *cancel_arg;
534 	int ncolored;
535 	int nfound;
536 	int ntrees;
537 	int ncommits;
538 };
539 
540 static const struct got_error *
541 search_delta_for_object(struct got_object_id *id, void *data, void *arg)
542 {
543 	const struct got_error *err;
544 	struct got_pack_meta *m = data;
545 	struct search_deltas_arg *a = arg;
546 	int obj_idx;
547 	struct got_object *obj = NULL;
548 
549 	if (a->cancel_cb) {
550 		err = (*a->cancel_cb)(a->cancel_arg);
551 		if (err)
552 			return err;
553 	}
554 
555 	if (!got_repo_check_packidx_bloom_filter(a->repo,
556 	    a->packidx->path_packidx, id))
557 		return NULL;
558 
559 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
560 	if (obj_idx == -1)
561 		return NULL;
562 
563 	/* TODO:
564 	 * Opening and closing an object just to check its flags
565 	 * is a bit expensive. We could have an imsg which requests
566 	 * plain type/size information for an object without doing
567 	 * work such as traversing the object's entire delta chain
568 	 * to find the base object type, and other such info which
569 	 * we don't really need here.
570 	 */
571 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
572 	    a->packidx, obj_idx, a->repo);
573 	if (err)
574 		return err;
575 
576 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
577 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
578 		    a->delta_cache_fd, a->repo);
579 		if (err)
580 			goto done;
581 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
582 		    a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
583 		    got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
584 	}
585 done:
586 	got_object_close(obj);
587 	return err;
588 }
589 
590 static const struct got_error *
591 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
592     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
593     struct got_repository *repo,
594     got_pack_progress_cb progress_cb, void *progress_arg,
595     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
596 {
597 	const struct got_error *err = NULL;
598 	char *path_packfile = NULL;
599 	struct got_packidx *packidx;
600 	struct got_pack *pack;
601 	struct search_deltas_arg sda;
602 
603 	err = find_pack_for_reuse(&packidx, repo);
604 	if (err)
605 		return err;
606 
607 	if (packidx == NULL)
608 		return NULL;
609 
610 	err = got_packidx_get_packfile_path(&path_packfile,
611 	    packidx->path_packidx);
612 	if (err)
613 		return err;
614 
615 	pack = got_repo_get_cached_pack(repo, path_packfile);
616 	if (pack == NULL) {
617 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
618 		if (err)
619 			goto done;
620 	}
621 
622 	sda.packidx = packidx;
623 	sda.pack = pack;
624 	sda.idset = idset;
625 	sda.v = v;
626 	sda.delta_cache_fd = delta_cache_fd;
627 	sda.repo = repo;
628 	sda.progress_cb = progress_cb;
629 	sda.progress_arg = progress_arg;
630 	sda.rl = rl;
631 	sda.cancel_cb = cancel_cb;
632 	sda.cancel_arg = cancel_arg;
633 	sda.ncolored = ncolored;
634 	sda.nfound = nfound;
635 	sda.ntrees = ntrees;
636 	sda.ncommits = ncommits;
637 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
638 done:
639 	free(path_packfile);
640 	return err;
641 }
642 
643 static const struct got_error *
644 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
645     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
646     struct got_repository *repo,
647     got_pack_progress_cb progress_cb, void *progress_arg,
648     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
649 {
650 	const struct got_error *err = NULL;
651 	struct got_pack_meta *m = NULL, *base = NULL;
652 	struct got_raw_object *raw = NULL, *base_raw = NULL;
653 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
654 	int i, j, ndeltas, best_ndeltas;
655 	off_t size, best_size;
656 	const int max_base_candidates = 3;
657 	size_t delta_memsize = 0;
658 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
659 	int outfd = -1;
660 
661 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
662 	for (i = 0; i < nmeta; i++) {
663 		if (cancel_cb) {
664 			err = (*cancel_cb)(cancel_arg);
665 			if (err)
666 				break;
667 		}
668 		err = report_progress(progress_cb, progress_arg, rl,
669 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
670 		    nreused + i, 0);
671 		if (err)
672 			goto done;
673 		m = meta[i];
674 
675 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
676 		    m->obj_type == GOT_OBJ_TYPE_TAG)
677 			continue;
678 
679 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
680 		if (err)
681 			goto done;
682 		m->size = raw->size;
683 
684 		if (raw->f == NULL) {
685 			err = got_deltify_init_mem(&m->dtab, raw->data,
686 			    raw->hdrlen, raw->size + raw->hdrlen);
687 		} else {
688 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
689 			    raw->size + raw->hdrlen);
690 		}
691 		if (err)
692 			goto done;
693 
694 		if (i > max_base_candidates) {
695 			struct got_pack_meta *n = NULL;
696 			n = meta[i - (max_base_candidates + 1)];
697 			got_deltify_free(n->dtab);
698 			n->dtab = NULL;
699 		}
700 
701 		best_size = raw->size;
702 		best_ndeltas = 0;
703 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
704 			if (cancel_cb) {
705 				err = (*cancel_cb)(cancel_arg);
706 				if (err)
707 					goto done;
708 			}
709 			base = meta[j];
710 			/* long chains make unpacking slow, avoid such bases */
711 			if (base->nchain >= 128 ||
712 			    base->obj_type != m->obj_type)
713 				continue;
714 
715 			err = got_object_raw_open(&base_raw, &outfd, repo,
716 			    &base->id);
717 			if (err)
718 				goto done;
719 
720 			if (raw->f == NULL && base_raw->f == NULL) {
721 				err = got_deltify_mem_mem(&deltas, &ndeltas,
722 				    raw->data, raw->hdrlen,
723 				    raw->size + raw->hdrlen,
724 				    base->dtab, base_raw->data,
725 				    base_raw->hdrlen,
726 				    base_raw->size + base_raw->hdrlen);
727 			} else if (raw->f == NULL) {
728 				err = got_deltify_mem_file(&deltas, &ndeltas,
729 				    raw->data, raw->hdrlen,
730 				    raw->size + raw->hdrlen,
731 				    base->dtab, base_raw->f,
732 				    base_raw->hdrlen,
733 				    base_raw->size + base_raw->hdrlen);
734 			} else if (base_raw->f == NULL) {
735 				err = got_deltify_file_mem(&deltas, &ndeltas,
736 				    raw->f, raw->hdrlen,
737 				    raw->size + raw->hdrlen,
738 				    base->dtab, base_raw->data,
739 				    base_raw->hdrlen,
740 				    base_raw->size + base_raw->hdrlen);
741 			} else {
742 				err = got_deltify(&deltas, &ndeltas,
743 				    raw->f, raw->hdrlen,
744 				    raw->size + raw->hdrlen,
745 				    base->dtab, base_raw->f, base_raw->hdrlen,
746 				    base_raw->size + base_raw->hdrlen);
747 			}
748 			got_object_raw_close(base_raw);
749 			base_raw = NULL;
750 			if (err)
751 				goto done;
752 
753 			size = delta_size(deltas, ndeltas);
754 			if (size + 32 < best_size){
755 				/*
756 				 * if we already picked a best delta,
757 				 * replace it.
758 				 */
759 				best_size = size;
760 				free(best_deltas);
761 				best_deltas = deltas;
762 				best_ndeltas = ndeltas;
763 				deltas = NULL;
764 				m->nchain = base->nchain + 1;
765 				m->prev = base;
766 				m->head = base->head;
767 				if (m->head == NULL)
768 					m->head = base;
769 			} else {
770 				free(deltas);
771 				deltas = NULL;
772 				ndeltas = 0;
773 			}
774 		}
775 
776 		if (best_ndeltas > 0) {
777 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
778 			    delta_memsize + best_size <= max_delta_memsize) {
779 				delta_memsize += best_size;
780 				err = encode_delta_in_mem(m, raw, best_deltas,
781 				    best_ndeltas, best_size, m->prev->size);
782 			} else {
783 				m->delta_offset = ftello(delta_cache);
784 				/*
785 				 * TODO:
786 				 * Storing compressed delta data in the delta
787 				 * cache file would probably be more efficient
788 				 * than writing uncompressed delta data here
789 				 * and compressing it while writing the pack
790 				 * file. This would also allow for reusing
791 				 * deltas in their compressed form.
792 				 */
793 				err = encode_delta(m, raw, best_deltas,
794 				    best_ndeltas, m->prev->size, delta_cache);
795 			}
796 			free(best_deltas);
797 			best_deltas = NULL;
798 			best_ndeltas = 0;
799 			if (err)
800 				goto done;
801 		}
802 
803 		got_object_raw_close(raw);
804 		raw = NULL;
805 	}
806 done:
807 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
808 		got_deltify_free(meta[i]->dtab);
809 		meta[i]->dtab = NULL;
810 	}
811 	if (raw)
812 		got_object_raw_close(raw);
813 	if (base_raw)
814 		got_object_raw_close(base_raw);
815 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
816 		err = got_error_from_errno("close");
817 	free(deltas);
818 	free(best_deltas);
819 	return err;
820 }
821 
822 static const struct got_error *
823 search_packidx(int *found, struct got_object_id *id,
824     struct got_repository *repo)
825 {
826 	const struct got_error *err = NULL;
827 	struct got_packidx *packidx = NULL;
828 	int idx;
829 
830 	*found = 0;
831 
832 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
833 	if (err == NULL)
834 		*found = 1; /* object is already packed */
835 	else if (err->code == GOT_ERR_NO_OBJ)
836 		err = NULL;
837 	return err;
838 }
839 
840 static const int obj_types[] = {
841 	GOT_OBJ_TYPE_ANY,
842 	GOT_OBJ_TYPE_COMMIT,
843 	GOT_OBJ_TYPE_TREE,
844 	GOT_OBJ_TYPE_BLOB,
845 	GOT_OBJ_TYPE_TAG,
846 	GOT_OBJ_TYPE_OFFSET_DELTA,
847 	GOT_OBJ_TYPE_REF_DELTA
848 };
849 
850 static const struct got_error *
851 add_object(int want_meta, struct got_object_idset *idset,
852     struct got_object_id *id, const char *path, int obj_type,
853     time_t mtime, int loose_obj_only, struct got_repository *repo)
854 {
855 	const struct got_error *err;
856 	struct got_pack_meta *m = NULL;
857 
858 	if (loose_obj_only) {
859 		int is_packed;
860 		err = search_packidx(&is_packed, id, repo);
861 		if (err)
862 			return err;
863 		if (is_packed)
864 			return NULL;
865 	}
866 
867 	if (want_meta) {
868 		err = alloc_meta(&m, id, path, obj_type, mtime);
869 		if (err)
870 			return err;
871 	}
872 
873 	return got_object_idset_add(idset, id, m);
874 }
875 
876 static const struct got_error *
877 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
878     struct got_object_idset *idset, struct got_object_id *tree_id,
879     const char *dpath, time_t mtime, struct got_repository *repo,
880     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
881     got_pack_progress_cb progress_cb, void *progress_arg,
882     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
883 {
884 	const struct got_error *err;
885 	struct got_tree_object *tree;
886 	char *p = NULL;
887 	int i;
888 
889 	err = got_object_open_as_tree(&tree, repo, tree_id);
890 	if (err)
891 		return err;
892 
893 	(*ntrees)++;
894 	err = report_progress(progress_cb, progress_arg, rl,
895 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
896 	if (err)
897 		return err;
898 
899 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
900 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
901 		struct got_object_id *id = got_tree_entry_get_id(e);
902 		mode_t mode = got_tree_entry_get_mode(e);
903 
904 		if (cancel_cb) {
905 			err = (*cancel_cb)(cancel_arg);
906 			if (err)
907 				break;
908 		}
909 
910 		if (got_object_tree_entry_is_submodule(e) ||
911 		    got_object_idset_contains(idset, id))
912 			continue;
913 		
914 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
915 		    got_tree_entry_get_name(e)) == -1) {
916 			err = got_error_from_errno("asprintf");
917 			break;
918 		}
919 
920 		if (S_ISDIR(mode)) {
921 			struct got_object_qid *qid;
922 			err = got_object_qid_alloc(&qid, id);
923 			if (err)
924 				break;
925 			STAILQ_INSERT_TAIL(ids, qid, entry);
926 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
927 			err = add_object(want_meta, idset, id, p,
928 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo);
929 			if (err)
930 				break;
931 			(*nfound)++;
932 			err = report_progress(progress_cb, progress_arg, rl,
933 			    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
934 			if (err)
935 				break;
936 		}
937 		free(p);
938 		p = NULL;
939 	}
940 
941 	got_object_tree_close(tree);
942 	free(p);
943 	return err;
944 }
945 
946 static const struct got_error *
947 load_tree(int want_meta, struct got_object_idset *idset,
948     struct got_object_id *tree_id, const char *dpath, time_t mtime,
949     struct got_repository *repo, int loose_obj_only,
950     int *ncolored, int *nfound, int *ntrees,
951     got_pack_progress_cb progress_cb, void *progress_arg,
952     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
953 {
954 	const struct got_error *err = NULL;
955 	struct got_object_id_queue tree_ids;
956 	struct got_object_qid *qid;
957 
958 	if (got_object_idset_contains(idset, tree_id))
959 		return NULL;
960 
961 	err = got_object_qid_alloc(&qid, tree_id);
962 	if (err)
963 		return err;
964 
965 	STAILQ_INIT(&tree_ids);
966 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
967 
968 	while (!STAILQ_EMPTY(&tree_ids)) {
969 		if (cancel_cb) {
970 			err = (*cancel_cb)(cancel_arg);
971 			if (err)
972 				break;
973 		}
974 
975 		qid = STAILQ_FIRST(&tree_ids);
976 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
977 
978 		if (got_object_idset_contains(idset, qid->id)) {
979 			got_object_qid_free(qid);
980 			continue;
981 		}
982 
983 		err = add_object(want_meta, idset, qid->id, dpath,
984 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo);
985 		if (err) {
986 			got_object_qid_free(qid);
987 			break;
988 		}
989 
990 		(*nfound)++;
991 		err = report_progress(progress_cb, progress_arg, rl,
992 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
993 		if (err)
994 			break;
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)
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 	if (err)
1039 		goto done;
1040 
1041 	(*nfound)++;
1042 	err = report_progress(progress_cb, progress_arg, rl,
1043 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1044 	if (err)
1045 		goto done;
1046 
1047 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1048 	    "", got_object_commit_get_committer_time(commit),
1049 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1050 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1051 done:
1052 	got_object_commit_close(commit);
1053 	return err;
1054 }
1055 
1056 static const struct got_error *
1057 load_tag(int want_meta, struct got_object_idset *idset,
1058     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1059     int *ncolored, int *nfound, int *ntrees,
1060     got_pack_progress_cb progress_cb, void *progress_arg,
1061     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1062 {
1063 	const struct got_error *err;
1064 	struct got_tag_object *tag = NULL;
1065 
1066 	if (got_object_idset_contains(idset, id))
1067 		return NULL;
1068 
1069 	if (loose_obj_only) {
1070 		int is_packed;
1071 		err = search_packidx(&is_packed, id, repo);
1072 		if (err)
1073 			return err;
1074 		if (is_packed)
1075 			return NULL;
1076 	}
1077 
1078 	err = got_object_open_as_tag(&tag, repo, id);
1079 	if (err)
1080 		return err;
1081 
1082 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1083 	    got_object_tag_get_tagger_time(tag),
1084 	    loose_obj_only, repo);
1085 	if (err)
1086 		goto done;
1087 
1088 	(*nfound)++;
1089 	err = report_progress(progress_cb, progress_arg, rl,
1090 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1091 	if (err)
1092 		goto done;
1093 
1094 	switch (got_object_tag_get_object_type(tag)) {
1095 	case GOT_OBJ_TYPE_COMMIT:
1096 		err = load_commit(want_meta, idset,
1097 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1098 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1099 		    cancel_cb, cancel_arg);
1100 		break;
1101 	case GOT_OBJ_TYPE_TREE:
1102 		err = load_tree(want_meta, idset,
1103 		    got_object_tag_get_object_id(tag), "",
1104 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1105 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1106 		    cancel_cb, cancel_arg);
1107 		break;
1108 	default:
1109 		break;
1110 	}
1111 
1112 done:
1113 	got_object_tag_close(tag);
1114 	return err;
1115 }
1116 
1117 enum findtwixt_color {
1118 	COLOR_KEEP = 0,
1119 	COLOR_DROP,
1120 	COLOR_BLANK,
1121 };
1122 static const int findtwixt_colors[] = {
1123 	COLOR_KEEP,
1124 	COLOR_DROP,
1125 	COLOR_BLANK
1126 };
1127 
1128 static const struct got_error *
1129 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1130     int color, struct got_repository *repo)
1131 {
1132 	const struct got_error *err;
1133 	struct got_object_qid *qid;
1134 
1135 	err = got_object_qid_alloc(&qid, id);
1136 	if (err)
1137 		return err;
1138 
1139 	STAILQ_INSERT_TAIL(ids, qid, entry);
1140 	qid->data = (void *)&findtwixt_colors[color];
1141 	return NULL;
1142 }
1143 
1144 static const struct got_error *
1145 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1146     struct got_object_id *id, struct got_repository *repo,
1147     got_cancel_cb cancel_cb, void *cancel_arg)
1148 {
1149 	const struct got_error *err = NULL;
1150 	struct got_commit_object *commit;
1151 	const struct got_object_id_queue *parents;
1152 	struct got_object_id_queue ids;
1153 	struct got_object_qid *qid;
1154 
1155 	STAILQ_INIT(&ids);
1156 
1157 	err = got_object_qid_alloc(&qid, id);
1158 	if (err)
1159 		return err;
1160 	STAILQ_INSERT_HEAD(&ids, qid, entry);
1161 
1162 	while (!STAILQ_EMPTY(&ids)) {
1163 		if (cancel_cb) {
1164 			err = (*cancel_cb)(cancel_arg);
1165 			if (err)
1166 				break;
1167 		}
1168 
1169 		qid = STAILQ_FIRST(&ids);
1170 		STAILQ_REMOVE_HEAD(&ids, entry);
1171 
1172 		if (got_object_idset_contains(drop, qid->id)) {
1173 			got_object_qid_free(qid);
1174 			continue;
1175 		}
1176 
1177 		err = got_object_idset_add(drop, qid->id,
1178 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1179 		if (err) {
1180 			got_object_qid_free(qid);
1181 			break;
1182 		}
1183 
1184 		if (!got_object_idset_contains(keep, qid->id)) {
1185 			got_object_qid_free(qid);
1186 			continue;
1187 		}
1188 
1189 		err = got_object_open_as_commit(&commit, repo, qid->id);
1190 		got_object_qid_free(qid);
1191 		if (err)
1192 			break;
1193 
1194 		parents = got_object_commit_get_parent_ids(commit);
1195 		if (parents) {
1196 			err = got_object_id_queue_copy(parents, &ids);
1197 			if (err) {
1198 				got_object_commit_close(commit);
1199 				break;
1200 			}
1201 		}
1202 		got_object_commit_close(commit);
1203 	}
1204 
1205 	got_object_id_queue_free(&ids);
1206 	return err;
1207 }
1208 
1209 struct append_id_arg {
1210 	struct got_object_id **array;
1211 	int idx;
1212 };
1213 
1214 static const struct got_error *
1215 append_id(struct got_object_id *id, void *data, void *arg)
1216 {
1217 	struct append_id_arg *a = arg;
1218 
1219 	a->array[a->idx] = got_object_id_dup(id);
1220 	if (a->array[a->idx] == NULL)
1221 		return got_error_from_errno("got_object_id_dup");
1222 
1223 	a->idx++;
1224 	return NULL;
1225 }
1226 
1227 static const struct got_error *
1228 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1229     struct got_object_id **head, int nhead,
1230     struct got_object_id **tail, int ntail,
1231     struct got_repository *repo,
1232     got_pack_progress_cb progress_cb, void *progress_arg,
1233     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1234 {
1235 	const struct got_error *err = NULL;
1236 	struct got_object_id_queue ids;
1237 	struct got_object_idset *keep, *drop;
1238 	struct got_object_qid *qid;
1239 	int i, ncolor, nkeep, obj_type;
1240 
1241 	STAILQ_INIT(&ids);
1242 	*res = NULL;
1243 	*nres = 0;
1244 	*ncolored = 0;
1245 
1246 	keep = got_object_idset_alloc();
1247 	if (keep == NULL)
1248 		return got_error_from_errno("got_object_idset_alloc");
1249 
1250 	drop = got_object_idset_alloc();
1251 	if (drop == NULL) {
1252 		err = got_error_from_errno("got_object_idset_alloc");
1253 		goto done;
1254 	}
1255 
1256 	for (i = 0; i < nhead; i++) {
1257 		struct got_object_id *id = head[i];
1258 		if (id == NULL)
1259 			continue;
1260 		err = got_object_get_type(&obj_type, repo, id);
1261 		if (err)
1262 			return err;
1263 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1264 			continue;
1265 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1266 		if (err)
1267 			goto done;
1268 	}		
1269 	for (i = 0; i < ntail; i++) {
1270 		struct got_object_id *id = tail[i];
1271 		if (id == NULL)
1272 			continue;
1273 		err = got_object_get_type(&obj_type, repo, id);
1274 		if (err)
1275 			return err;
1276 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1277 			continue;
1278 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1279 		if (err)
1280 			goto done;
1281 	}
1282 
1283 	while (!STAILQ_EMPTY(&ids)) {
1284 		int qcolor;
1285 		qid = STAILQ_FIRST(&ids);
1286 		qcolor = *((int *)qid->data);
1287 
1288 		if (got_object_idset_contains(drop, qid->id))
1289 			ncolor = COLOR_DROP;
1290 		else if (got_object_idset_contains(keep, qid->id))
1291 			ncolor = COLOR_KEEP;
1292 		else
1293 			ncolor = COLOR_BLANK;
1294 
1295 		(*ncolored)++;
1296 		err = report_progress(progress_cb, progress_arg, rl,
1297 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1298 		if (err)
1299 			goto done;
1300 
1301 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1302 		    qcolor == COLOR_KEEP)) {
1303 			STAILQ_REMOVE_HEAD(&ids, entry);
1304 			got_object_qid_free(qid);
1305 			continue;
1306 		}
1307 
1308 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1309 			err = drop_commit(keep, drop, qid->id, repo,
1310 			    cancel_cb, cancel_arg);
1311 			if (err)
1312 				goto done;
1313 		} else if (ncolor == COLOR_BLANK) {
1314 			struct got_commit_object *commit;
1315 			struct got_object_id *id;
1316 			const struct got_object_id_queue *parents;
1317 			struct got_object_qid *pid;
1318 
1319 			id = got_object_id_dup(qid->id);
1320 			if (id == NULL) {
1321 				err = got_error_from_errno("got_object_id_dup");
1322 				goto done;
1323 			}
1324 			if (qcolor == COLOR_KEEP)
1325 				err = got_object_idset_add(keep, id,
1326 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1327 			else
1328 				err = got_object_idset_add(drop, id,
1329 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1330 			if (err) {
1331 				free(id);
1332 				goto done;
1333 			}
1334 
1335 			err = got_object_open_as_commit(&commit, repo, id);
1336 			if (err) {
1337 				free(id);
1338 				goto done;
1339 			}
1340 			parents = got_object_commit_get_parent_ids(commit);
1341 			if (parents) {
1342 				STAILQ_FOREACH(pid, parents, entry) {
1343 					err = queue_commit_id(&ids, pid->id,
1344 					    qcolor, repo);
1345 					if (err) {
1346 						free(id);
1347 						goto done;
1348 					}
1349 				}
1350 			}
1351 			got_object_commit_close(commit);
1352 			commit = NULL;
1353 		} else {
1354 			/* should not happen */
1355 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1356 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1357 			goto done;
1358 		}
1359 
1360 		STAILQ_REMOVE_HEAD(&ids, entry);
1361 		got_object_qid_free(qid);
1362 	}
1363 
1364 	nkeep = got_object_idset_num_elements(keep);
1365 	if (nkeep > 0) {
1366 		struct append_id_arg arg;
1367 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1368 		if (arg.array == NULL) {
1369 			err = got_error_from_errno("calloc");
1370 			goto done;
1371 		}
1372 		arg.idx = 0;
1373 		err = got_object_idset_for_each(keep, append_id, &arg);
1374 		if (err) {
1375 			free(arg.array);
1376 			goto done;
1377 		}
1378 		*res = arg.array;
1379 		*nres = nkeep;
1380 	}
1381 done:
1382 	got_object_idset_free(keep);
1383 	got_object_idset_free(drop);
1384 	got_object_id_queue_free(&ids);
1385 	return err;
1386 }
1387 
1388 static const struct got_error *
1389 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1390     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1391     struct got_object_id **ours, int nours, struct got_repository *repo,
1392     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1393     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1394 {
1395 	const struct got_error *err = NULL;
1396 	struct got_object_id **ids = NULL;
1397 	int i, nobj = 0, obj_type;
1398 
1399 	*ncolored = 0;
1400 	*nfound = 0;
1401 	*ntrees = 0;
1402 
1403 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1404 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1405 	if (err || nobj == 0)
1406 		goto done;
1407 
1408 	for (i = 0; i < ntheirs; i++) {
1409 		struct got_object_id *id = theirs[i];
1410 		if (id == NULL)
1411 			continue;
1412 		err = got_object_get_type(&obj_type, repo, id);
1413 		if (err)
1414 			return err;
1415 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1416 			continue;
1417 		err = load_commit(0, idset, id, repo, loose_obj_only,
1418 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1419 		    cancel_cb, cancel_arg);
1420 		if (err)
1421 			goto done;
1422 	}
1423 
1424 	for (i = 0; i < ntheirs; i++) {
1425 		struct got_object_id *id = theirs[i];
1426 		struct got_pack_meta *m;
1427 		if (id == NULL)
1428 			continue;
1429 		m = got_object_idset_get(idset, id);
1430 		if (m == NULL) {
1431 			err = got_object_get_type(&obj_type, repo, id);
1432 			if (err)
1433 				goto done;
1434 		} else
1435 			obj_type = m->obj_type;
1436 		if (obj_type != GOT_OBJ_TYPE_TAG)
1437 			continue;
1438 		err = load_tag(0, idset, id, repo, loose_obj_only,
1439 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1440 		    cancel_cb, cancel_arg);
1441 		if (err)
1442 			goto done;
1443 	}
1444 
1445 	for (i = 0; i < nobj; i++) {
1446 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1447 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1448 		    cancel_cb, cancel_arg);
1449 		if (err)
1450 			goto done;
1451 	}
1452 
1453 	for (i = 0; i < nours; i++) {
1454 		struct got_object_id *id = ours[i];
1455 		struct got_pack_meta *m;
1456 		if (id == NULL)
1457 			continue;
1458 		m = got_object_idset_get(idset, id);
1459 		if (m == NULL) {
1460 			err = got_object_get_type(&obj_type, repo, id);
1461 			if (err)
1462 				goto done;
1463 		} else
1464 			obj_type = m->obj_type;
1465 		if (obj_type != GOT_OBJ_TYPE_TAG)
1466 			continue;
1467 		err = load_tag(1, idset, id, repo, loose_obj_only,
1468 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1469 		    cancel_cb, cancel_arg);
1470 		if (err)
1471 			goto done;
1472 	}
1473 done:
1474 	for (i = 0; i < nobj; i++) {
1475 		free(ids[i]);
1476 	}
1477 	free(ids);
1478 	return err;
1479 }
1480 
1481 const struct got_error *
1482 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1483 {
1484 	size_t n;
1485 
1486 	SHA1Update(ctx, buf, len);
1487 	n = fwrite(buf, 1, len, f);
1488 	if (n != len)
1489 		return got_ferror(f, GOT_ERR_IO);
1490 	return NULL;
1491 }
1492 
1493 static void
1494 putbe32(char *b, uint32_t n)
1495 {
1496 	b[0] = n >> 24;
1497 	b[1] = n >> 16;
1498 	b[2] = n >> 8;
1499 	b[3] = n >> 0;
1500 }
1501 
1502 static int
1503 write_order_cmp(const void *pa, const void *pb)
1504 {
1505 	struct got_pack_meta *a, *b, *ahd, *bhd;
1506 
1507 	a = *(struct got_pack_meta **)pa;
1508 	b = *(struct got_pack_meta **)pb;
1509 	ahd = (a->head == NULL) ? a : a->head;
1510 	bhd = (b->head == NULL) ? b : b->head;
1511 	if (ahd->mtime != bhd->mtime)
1512 		return bhd->mtime - ahd->mtime;
1513 	if (ahd != bhd)
1514 		return (uintptr_t)bhd - (uintptr_t)ahd;
1515 	if (a->nchain != b->nchain)
1516 		return a->nchain - b->nchain;
1517 	return a->mtime - b->mtime;
1518 }
1519 
1520 static int
1521 reuse_write_order_cmp(const void *pa, const void *pb)
1522 {
1523 	struct got_pack_meta *a, *b;
1524 
1525 	a = *(struct got_pack_meta **)pa;
1526 	b = *(struct got_pack_meta **)pb;
1527 
1528 	if (a->reused_delta_offset < b->reused_delta_offset)
1529 		return -1;
1530 	if (a->reused_delta_offset > b->reused_delta_offset)
1531 		return 1;
1532 	return 0;
1533 }
1534 
1535 static const struct got_error *
1536 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1537 {
1538 	size_t i;
1539 
1540 	*hdrlen = 0;
1541 
1542 	hdr[0] = obj_type << 4;
1543 	hdr[0] |= len & 0xf;
1544 	len >>= 4;
1545 	for (i = 1; len != 0; i++){
1546 		if (i >= bufsize)
1547 			return got_error(GOT_ERR_NO_SPACE);
1548 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1549 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1550 		len >>= GOT_DELTA_SIZE_SHIFT;
1551 	}
1552 
1553 	*hdrlen = i;
1554 	return NULL;
1555 }
1556 
1557 static int
1558 packoff(char *hdr, off_t off)
1559 {
1560 	int i, j;
1561 	char rbuf[8];
1562 
1563 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1564 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1565 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1566 		    GOT_DELTA_SIZE_MORE;
1567 	}
1568 
1569 	j = 0;
1570 	while (i > 0)
1571 		hdr[j++] = rbuf[--i];
1572 	return j;
1573 }
1574 
1575 static const struct got_error *
1576 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1577     struct got_pack_meta *m)
1578 {
1579 	const struct got_error *err;
1580 	char buf[32];
1581 	int nh;
1582 
1583 	if (m->prev->off != 0) {
1584 		err = packhdr(&nh, buf, sizeof(buf),
1585 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1586 		if (err)
1587 			return err;
1588 		nh += packoff(buf + nh, m->off - m->prev->off);
1589 		err = hwrite(packfile, buf, nh, ctx);
1590 		if (err)
1591 			return err;
1592 		*packfile_size += nh;
1593 	} else {
1594 		err = packhdr(&nh, buf, sizeof(buf),
1595 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1596 		if (err)
1597 			return err;
1598 		err = hwrite(packfile, buf, nh, ctx);
1599 		if (err)
1600 			return err;
1601 		*packfile_size += nh;
1602 		err = hwrite(packfile, m->prev->id.sha1,
1603 		    sizeof(m->prev->id.sha1), ctx);
1604 		if (err)
1605 			return err;
1606 		*packfile_size += sizeof(m->prev->id.sha1);
1607 	}
1608 
1609 	return NULL;
1610 }
1611 
1612 static const struct got_error *
1613 write_packed_object(off_t *packfile_size, FILE *packfile,
1614     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1615     SHA1_CTX *ctx, struct got_repository *repo)
1616 {
1617 	const struct got_error *err = NULL;
1618 	struct got_deflate_checksum csum;
1619 	char buf[32];
1620 	int nh;
1621 	struct got_raw_object *raw = NULL;
1622 	off_t outlen;
1623 
1624 	csum.output_sha1 = ctx;
1625 	csum.output_crc = NULL;
1626 
1627 	m->off = ftello(packfile);
1628 	if (m->delta_len == 0) {
1629 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1630 		if (err)
1631 			goto done;
1632 		err = packhdr(&nh, buf, sizeof(buf),
1633 		    m->obj_type, raw->size);
1634 		if (err)
1635 			goto done;
1636 		err = hwrite(packfile, buf, nh, ctx);
1637 		if (err)
1638 			goto done;
1639 		*packfile_size += nh;
1640 		if (raw->f == NULL) {
1641 			err = got_deflate_to_file_mmap(&outlen,
1642 			    raw->data + raw->hdrlen, 0, raw->size,
1643 			    packfile, &csum);
1644 			if (err)
1645 				goto done;
1646 		} else {
1647 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1648 			    == -1) {
1649 				err = got_error_from_errno("fseeko");
1650 				goto done;
1651 			}
1652 			err = got_deflate_to_file(&outlen, raw->f,
1653 			    raw->size, packfile, &csum);
1654 			if (err)
1655 				goto done;
1656 		}
1657 		*packfile_size += outlen;
1658 		got_object_raw_close(raw);
1659 		raw = NULL;
1660 	} else if (m->delta_buf) {
1661 		err = deltahdr(packfile_size, ctx, packfile, m);
1662 		if (err)
1663 			goto done;
1664 		err = got_deflate_to_file_mmap(&outlen,
1665 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1666 		if (err)
1667 			goto done;
1668 		*packfile_size += outlen;
1669 		free(m->delta_buf);
1670 		m->delta_buf = NULL;
1671 	} else {
1672 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1673 		    == -1) {
1674 			err = got_error_from_errno("fseeko");
1675 			goto done;
1676 		}
1677 		err = deltahdr(packfile_size, ctx, packfile, m);
1678 		if (err)
1679 			goto done;
1680 		err = got_deflate_to_file(&outlen, delta_cache,
1681 		    m->delta_len, packfile, &csum);
1682 		if (err)
1683 			goto done;
1684 		*packfile_size += outlen;
1685 	}
1686 done:
1687 	if (raw)
1688 		got_object_raw_close(raw);
1689 	return err;
1690 }
1691 
1692 static const struct got_error *
1693 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1694     struct got_pack_meta **deltify, int ndeltify,
1695     struct got_pack_meta **reuse, int nreuse,
1696     int ncolored, int nfound, int ntrees, int nours,
1697     struct got_repository *repo,
1698     got_pack_progress_cb progress_cb, void *progress_arg,
1699     struct got_ratelimit *rl,
1700     got_cancel_cb cancel_cb, void *cancel_arg)
1701 {
1702 	const struct got_error *err = NULL;
1703 	int i;
1704 	SHA1_CTX ctx;
1705 	struct got_pack_meta *m;
1706 	char buf[32];
1707 	size_t n;
1708 	off_t packfile_size = 0;
1709 	int outfd = -1;
1710 
1711 	SHA1Init(&ctx);
1712 
1713 	err = hwrite(packfile, "PACK", 4, &ctx);
1714 	if (err)
1715 		return err;
1716 	putbe32(buf, GOT_PACKFILE_VERSION);
1717 	err = hwrite(packfile, buf, 4, &ctx);
1718 	if (err)
1719 		goto done;
1720 	putbe32(buf, ndeltify + nreuse);
1721 	err = hwrite(packfile, buf, 4, &ctx);
1722 	if (err)
1723 		goto done;
1724 
1725 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1726 	    write_order_cmp);
1727 	for (i = 0; i < ndeltify; i++) {
1728 		err = report_progress(progress_cb, progress_arg, rl,
1729 		    ncolored, nfound, ntrees, packfile_size, nours,
1730 		    ndeltify + nreuse, ndeltify + nreuse, i);
1731 		if (err)
1732 			goto done;
1733 		m = deltify[i];
1734 		err = write_packed_object(&packfile_size, packfile,
1735 		    delta_cache, m, &outfd, &ctx, repo);
1736 		if (err)
1737 			goto done;
1738 	}
1739 
1740 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1741 	    reuse_write_order_cmp);
1742 	for (i = 0; i < nreuse; i++) {
1743 		err = report_progress(progress_cb, progress_arg, rl,
1744 		    ncolored, nfound, ntrees, packfile_size, nours,
1745 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1746 		if (err)
1747 			goto done;
1748 		m = reuse[i];
1749 		err = write_packed_object(&packfile_size, packfile,
1750 		    delta_cache, m, &outfd, &ctx, repo);
1751 		if (err)
1752 			goto done;
1753 	}
1754 
1755 	SHA1Final(pack_sha1, &ctx);
1756 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1757 	if (n != SHA1_DIGEST_LENGTH)
1758 		err = got_ferror(packfile, GOT_ERR_IO);
1759 	packfile_size += SHA1_DIGEST_LENGTH;
1760 	packfile_size += sizeof(struct got_packfile_hdr);
1761 	if (progress_cb) {
1762 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1763 		    packfile_size, nours, ndeltify + nreuse,
1764 		    ndeltify + nreuse, ndeltify + nreuse);
1765 		if (err)
1766 			goto done;
1767 	}
1768 done:
1769 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1770 		err = got_error_from_errno("close");
1771 	return err;
1772 }
1773 
1774 static const struct got_error *
1775 remove_unused_object(struct got_object_idset_element *entry, void *arg)
1776 {
1777 	struct got_object_idset *idset = arg;
1778 
1779 	if (got_object_idset_get_element_data(entry) == NULL)
1780 		got_object_idset_remove_element(idset, entry);
1781 
1782 	return NULL;
1783 }
1784 
1785 static const struct got_error *
1786 remove_reused_object(struct got_object_idset_element *entry, void *arg)
1787 {
1788 	struct got_object_idset *idset = arg;
1789 	struct got_pack_meta *m;
1790 
1791 	m = got_object_idset_get_element_data(entry);
1792 	if (m->have_reused_delta)
1793 		got_object_idset_remove_element(idset, entry);
1794 
1795 	return NULL;
1796 }
1797 
1798 static const struct got_error *
1799 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1800 {
1801 	struct got_pack_meta *m = data;
1802 	struct got_pack_metavec *v = arg;
1803 
1804 	return add_meta(m, v);
1805 }
1806 
1807 const struct got_error *
1808 got_pack_create(uint8_t *packsha1, FILE *packfile,
1809     struct got_object_id **theirs, int ntheirs,
1810     struct got_object_id **ours, int nours,
1811     struct got_repository *repo, int loose_obj_only, int allow_empty,
1812     got_pack_progress_cb progress_cb, void *progress_arg,
1813     got_cancel_cb cancel_cb, void *cancel_arg)
1814 {
1815 	const struct got_error *err;
1816 	int delta_cache_fd = -1;
1817 	FILE *delta_cache = NULL;
1818 	struct got_object_idset *idset;
1819 	struct got_ratelimit rl;
1820 	struct got_pack_metavec deltify, reuse;
1821 	int ncolored = 0, nfound = 0, ntrees = 0;
1822 
1823 	memset(&deltify, 0, sizeof(deltify));
1824 	memset(&reuse, 0, sizeof(reuse));
1825 
1826 	got_ratelimit_init(&rl, 0, 500);
1827 
1828 	idset = got_object_idset_alloc();
1829 	if (idset == NULL)
1830 		return got_error_from_errno("got_object_idset_alloc");
1831 
1832 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1833 	    ntheirs, ours, nours, repo, loose_obj_only,
1834 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1835 	if (err)
1836 		return err;
1837 
1838 	err = got_object_idset_for_each_element(idset,
1839 	    remove_unused_object, idset);
1840 	if (err)
1841 		goto done;
1842 
1843 	if (progress_cb) {
1844 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1845 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1846 		if (err)
1847 			goto done;
1848 	}
1849 
1850 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1851 		err = got_error(GOT_ERR_CANNOT_PACK);
1852 		goto done;
1853 	}
1854 
1855 	delta_cache_fd = got_opentempfd();
1856 	if (delta_cache_fd == -1) {
1857 		err = got_error_from_errno("got_opentemp");
1858 		goto done;
1859 	}
1860 
1861 	reuse.metasz = 64;
1862 	reuse.meta = calloc(reuse.metasz,
1863 	    sizeof(struct got_pack_meta *));
1864 	if (reuse.meta == NULL) {
1865 		err = got_error_from_errno("calloc");
1866 		goto done;
1867 	}
1868 
1869 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1870 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1871 	    cancel_cb, cancel_arg);
1872 	if (err)
1873 		goto done;
1874 	if (reuse.nmeta > 0) {
1875 		err = got_object_idset_for_each_element(idset,
1876 		    remove_reused_object, idset);
1877 		if (err)
1878 			goto done;
1879 	}
1880 
1881 	delta_cache = fdopen(delta_cache_fd, "a+");
1882 	if (delta_cache == NULL) {
1883 		err = got_error_from_errno("fdopen");
1884 		goto done;
1885 	}
1886 	delta_cache_fd = -1;
1887 
1888 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1889 		err = got_error_from_errno("fseeko");
1890 		goto done;
1891 	}
1892 
1893 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1894 	    sizeof(struct got_pack_meta *));
1895 	if (deltify.meta == NULL) {
1896 		err = got_error_from_errno("calloc");
1897 		goto done;
1898 	}
1899 	deltify.metasz = got_object_idset_num_elements(idset);
1900 
1901 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1902 	if (err)
1903 		goto done;
1904 	if (deltify.nmeta > 0) {
1905 		err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1906 		    nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1907 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1908 		if (err)
1909 			goto done;
1910 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1911 			err = got_error_from_errno("fseeko");
1912 			goto done;
1913 		}
1914 	}
1915 
1916 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1917 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1918 	    nours, repo, progress_cb, progress_arg, &rl,
1919 	    cancel_cb, cancel_arg);
1920 	if (err)
1921 		goto done;
1922 done:
1923 	free_nmeta(deltify.meta, deltify.nmeta);
1924 	free_nmeta(reuse.meta, reuse.nmeta);
1925 	got_object_idset_free(idset);
1926 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1927 		err = got_error_from_errno("close");
1928 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1929 		err = got_error_from_errno("fclose");
1930 	return err;
1931 }