1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 #include <sys/queue.h>
20 #include <sys/tree.h>
21 #include <sys/uio.h>
22 #include <sys/stat.h>
23 #include <sys/time.h>
24 
25 #include <endian.h>
26 #include <stdint.h>
27 #include <imsg.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <sha1.h>
32 #include <time.h>
33 #include <unistd.h>
34 #include <limits.h>
35 #include <zlib.h>
36 
37 #include "got_error.h"
38 #include "got_cancel.h"
39 #include "got_object.h"
40 #include "got_path.h"
41 #include "got_reference.h"
42 #include "got_repository_admin.h"
43 #include "got_opentemp.h"
44 
45 #include "got_lib_deltify.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_idset.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_deflate.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_repository.h"
54 #include "got_lib_ratelimit.h"
55 
56 #ifndef MIN
57 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
58 #endif
59 
60 #ifndef MAX
61 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
62 #endif
63 
64 #ifndef nitems
65 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
66 #endif
67 
68 struct got_pack_meta {
69 	struct got_object_id id;
70 	char	*path;
71 	int	obj_type;
72 	off_t	size;
73 	time_t	mtime;
74 
75 	/* The best delta we picked */
76 	struct got_pack_meta *head;
77 	struct got_pack_meta *prev;
78 	unsigned char *delta_buf; /* if not encoded in delta cache file */
79 	off_t	delta_offset;	/* offset in delta cache file */
80 	off_t	delta_len;	/* encoded delta length */
81 	int	nchain;
82 
83 	int	have_reused_delta;
84 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
85 	struct got_object_id *base_obj_id;
86 
87 	/* Only used for delta window */
88 	struct got_delta_table *dtab;
89 
90 	/* Only used for writing offset deltas */
91 	off_t	off;
92 };
93 
94 struct got_pack_metavec {
95 	struct got_pack_meta **meta;
96 	int nmeta;
97 	int metasz;
98 };
99 
100 static const struct got_error *
101 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
102     const char *path, int obj_type, time_t mtime)
103 {
104 	const struct got_error *err = NULL;
105 	struct got_pack_meta *m;
106 
107 	*new = NULL;
108 
109 	m = calloc(1, sizeof(*m));
110 	if (m == NULL)
111 		return got_error_from_errno("calloc");
112 
113 	memcpy(&m->id, id, sizeof(m->id));
114 
115 	m->path = strdup(path);
116 	if (m->path == NULL) {
117 		err = got_error_from_errno("strdup");
118 		free(m);
119 		return err;
120 	}
121 
122 	m->obj_type = obj_type;
123 	m->mtime = mtime;
124 	*new = m;
125 	return NULL;
126 }
127 
128 static void
129 clear_meta(struct got_pack_meta *meta)
130 {
131 	if (meta == NULL)
132 		return;
133 	free(meta->path);
134 	meta->path = NULL;
135 	free(meta->delta_buf);
136 	meta->delta_buf = NULL;
137 	free(meta->base_obj_id);
138 	meta->base_obj_id = NULL;
139 }
140 
141 static void
142 free_nmeta(struct got_pack_meta **meta, int nmeta)
143 {
144 	int i;
145 
146 	for (i = 0; i < nmeta; i++)
147 		clear_meta(meta[i]);
148 	free(meta);
149 }
150 
151 static int
152 delta_order_cmp(const void *pa, const void *pb)
153 {
154 	struct got_pack_meta *a, *b;
155 	int cmp;
156 
157 	a = *(struct got_pack_meta **)pa;
158 	b = *(struct got_pack_meta **)pb;
159 
160 	if (a->obj_type != b->obj_type)
161 		return a->obj_type - b->obj_type;
162 	cmp = strcmp(a->path, b->path);
163 	if (cmp != 0)
164 		return cmp;
165 	if (a->mtime < b->mtime)
166 		return -1;
167 	if (a->mtime > b->mtime)
168 		return 1;
169 	return got_object_id_cmp(&a->id, &b->id);
170 }
171 
172 static off_t
173 delta_size(struct got_delta_instruction *deltas, int ndeltas)
174 {
175 	int i;
176 	off_t size = 32;
177 	for (i = 0; i < ndeltas; i++) {
178 		if (deltas[i].copy)
179 			size += GOT_DELTA_SIZE_SHIFT;
180 		else
181 			size += deltas[i].len + 1;
182 	}
183 	return size;
184 }
185 
186 static const struct got_error *
187 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
188 {
189 	char *n;
190 
191 	if (*len + nseg >= *sz) {
192 		while (*len + nseg >= *sz)
193 			*sz += *sz / 2;
194 		n = realloc(*p, *sz);
195 		if (n == NULL)
196 			return got_error_from_errno("realloc");
197 		*p = n;
198 	}
199 	memcpy(*p + *len, seg, nseg);
200 	*len += nseg;
201 	return NULL;
202 }
203 
204 static const struct got_error *
205 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
206     struct got_delta_instruction *deltas, int ndeltas,
207     off_t delta_size, off_t base_size)
208 {
209 	const struct got_error *err;
210 	unsigned char buf[16], *bp;
211 	int i, j;
212 	size_t len = 0;
213 	off_t n;
214 	struct got_delta_instruction *d;
215 
216 	m->delta_buf = malloc(delta_size);
217 	if (m->delta_buf == NULL)
218 		return got_error_from_errno("calloc");
219 
220 	/* base object size */
221 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
222 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
223 	for (i = 1; n > 0; i++) {
224 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
225 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
226 		n >>= GOT_DELTA_SIZE_SHIFT;
227 	}
228 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
229 	if (err)
230 		return err;
231 
232 	/* target object size */
233 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
234 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
235 	for (i = 1; n > 0; i++) {
236 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
237 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
238 		n >>= GOT_DELTA_SIZE_SHIFT;
239 	}
240 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
241 	if (err)
242 		return err;
243 
244 	for (j = 0; j < ndeltas; j++) {
245 		d = &deltas[j];
246 		if (d->copy) {
247 			n = d->offset;
248 			bp = &buf[1];
249 			buf[0] = GOT_DELTA_BASE_COPY;
250 			for (i = 0; i < 4; i++) {
251 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
252 				buf[0] |= 1 << i;
253 				*bp++ = n & 0xff;
254 				n >>= 8;
255 				if (n == 0)
256 					break;
257 			}
258 
259 			n = d->len;
260 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
261 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
262 				for (i = 0; i < 3 && n > 0; i++) {
263 					buf[0] |= 1 << (i + 4);
264 					*bp++ = n & 0xff;
265 					n >>= 8;
266 				}
267 			}
268 			err = append(&m->delta_buf, &len, &delta_size,
269 			    buf, bp - buf);
270 			if (err)
271 				return err;
272 		} else if (o->f == NULL) {
273 			n = 0;
274 			while (n != d->len) {
275 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
276 				err = append(&m->delta_buf, &len, &delta_size,
277 				    buf, 1);
278 				if (err)
279 					return err;
280 				err = append(&m->delta_buf, &len, &delta_size,
281 				    o->data + o->hdrlen + d->offset + n,
282 				    buf[0]);
283 				if (err)
284 					return err;
285 				n += buf[0];
286 			}
287 		} else {
288 			char content[128];
289 			size_t r;
290 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
291 				return got_error_from_errno("fseeko");
292 			n = 0;
293 			while (n != d->len) {
294 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
295 				err = append(&m->delta_buf, &len, &delta_size,
296 				    buf, 1);
297 				if (err)
298 					return err;
299 				r = fread(content, 1, buf[0], o->f);
300 				if (r != buf[0])
301 					return got_ferror(o->f, GOT_ERR_IO);
302 				err = append(&m->delta_buf, &len, &delta_size,
303 				    content, buf[0]);
304 				if (err)
305 					return err;
306 				n += buf[0];
307 			}
308 		}
309 	}
310 
311 	m->delta_len = len;
312 	return NULL;
313 }
314 
315 static const struct got_error *
316 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
317     struct got_delta_instruction *deltas, int ndeltas,
318     off_t base_size, FILE *f)
319 {
320 	unsigned char buf[16], *bp;
321 	int i, j;
322 	off_t n;
323 	size_t w;
324 	struct got_delta_instruction *d;
325 
326 	/* base object size */
327 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
328 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
329 	for (i = 1; n > 0; i++) {
330 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
331 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
332 		n >>= GOT_DELTA_SIZE_SHIFT;
333 	}
334 	w = fwrite(buf, 1, i, f);
335 	if (w != i)
336 		return got_ferror(f, GOT_ERR_IO);
337 
338 	/* target object size */
339 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
340 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
341 	for (i = 1; n > 0; i++) {
342 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
343 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
344 		n >>= GOT_DELTA_SIZE_SHIFT;
345 	}
346 	w = fwrite(buf, 1, i, f);
347 	if (w != i)
348 		return got_ferror(f, GOT_ERR_IO);
349 
350 	for (j = 0; j < ndeltas; j++) {
351 		d = &deltas[j];
352 		if (d->copy) {
353 			n = d->offset;
354 			bp = &buf[1];
355 			buf[0] = GOT_DELTA_BASE_COPY;
356 			for (i = 0; i < 4; i++) {
357 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
358 				buf[0] |= 1 << i;
359 				*bp++ = n & 0xff;
360 				n >>= 8;
361 				if (n == 0)
362 					break;
363 			}
364 
365 			n = d->len;
366 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
367 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
368 				for (i = 0; i < 3 && n > 0; i++) {
369 					buf[0] |= 1 << (i + 4);
370 					*bp++ = n & 0xff;
371 					n >>= 8;
372 				}
373 			}
374 			w = fwrite(buf, 1, bp - buf, f);
375 			if (w != bp - buf)
376 				return got_ferror(f, GOT_ERR_IO);
377 		} else if (o->f == NULL) {
378 			n = 0;
379 			while (n != d->len) {
380 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
381 				w = fwrite(buf, 1, 1, f);
382 				if (w != 1)
383 					return got_ferror(f, GOT_ERR_IO);
384 				w = fwrite(o->data + o->hdrlen + d->offset + n,
385 				    1, buf[0], f);
386 				if (w != buf[0])
387 					return got_ferror(f, GOT_ERR_IO);
388 				n += buf[0];
389 			}
390 		} else {
391 			char content[128];
392 			size_t r;
393 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
394 				return got_error_from_errno("fseeko");
395 			n = 0;
396 			while (n != d->len) {
397 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
398 				w = fwrite(buf, 1, 1, f);
399 				if (w != 1)
400 					return got_ferror(f, GOT_ERR_IO);
401 				r = fread(content, 1, buf[0], o->f);
402 				if (r != buf[0])
403 					return got_ferror(o->f, GOT_ERR_IO);
404 				w = fwrite(content, 1, buf[0], f);
405 				if (w != buf[0])
406 					return got_ferror(f, GOT_ERR_IO);
407 				n += buf[0];
408 			}
409 		}
410 	}
411 
412 	m->delta_len = ftello(f) - m->delta_offset;
413 	return NULL;
414 }
415 
416 static const struct got_error *
417 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
418     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
419     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
420     int nobj_written)
421 {
422 	const struct got_error *err;
423 	int elapsed;
424 
425 	if (progress_cb == NULL)
426 		return NULL;
427 
428 	err = got_ratelimit_check(&elapsed, rl);
429 	if (err || !elapsed)
430 		return err;
431 
432 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
433 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
434 }
435 
436 static const struct got_error *
437 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
438 {
439 	if (v->nmeta == v->metasz){
440 		size_t newsize = 2 * v->metasz;
441 		struct got_pack_meta **new;
442 		new = reallocarray(v->meta, newsize, sizeof(*new));
443 		if (new == NULL)
444 			return got_error_from_errno("reallocarray");
445 		v->meta = new;
446 		v->metasz = newsize; 
447 	}
448 
449 	v->meta[v->nmeta++] = m;
450 	return NULL;
451 }
452 
453 static const struct got_error *
454 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
455     struct got_object_idset *idset, struct got_pack *pack,
456     struct got_packidx *packidx, int delta_cache_fd,
457     struct got_repository *repo)
458 {
459 	const struct got_error *err = NULL;
460 	struct got_pack_meta *base = NULL;
461 	struct got_object_id *base_obj_id = NULL;
462 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
463 	uint64_t base_size, result_size;
464 
465 	if (m->have_reused_delta)
466 		return NULL;
467 
468 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
469 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
470 	    packidx, idx, &m->id, repo);
471 	if (err)
472 		return err;
473 
474 	if (delta_offset + delta_len < delta_offset)
475 		return got_error(GOT_ERR_BAD_PACKFILE);
476 
477 	base = got_object_idset_get(idset, base_obj_id);
478 	if (base == NULL)
479 		goto done;
480 
481 	m->delta_len = delta_len;
482 	m->delta_offset = delta_cache_offset;
483 	m->prev = base;
484 	m->size = result_size;
485 	m->have_reused_delta = 1;
486 	m->reused_delta_offset = delta_offset;
487 	m->base_obj_id = base_obj_id;
488 	base_obj_id = NULL;
489 	err = add_meta(m, v);
490 done:
491 	free(base_obj_id);
492 	return err;
493 }
494 
495 static const struct got_error *
496 find_pack_for_reuse(struct got_packidx **best_packidx,
497     struct got_repository *repo)
498 {
499 	const struct got_error *err = NULL;
500 	struct got_pathlist_entry *pe;
501 	const char *best_packidx_path = NULL;
502 	int nobj_max = 0;
503 
504 	*best_packidx = NULL;
505 
506 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
507 		const char *path_packidx = pe->path;
508 		struct got_packidx *packidx;
509 		int nobj;
510 
511 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
512 		if (err)
513 			break;
514 
515 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
516 		if (nobj > nobj_max) {
517 			best_packidx_path = path_packidx;
518 			nobj_max = nobj;
519 		}
520 	}
521 
522 	if (best_packidx_path) {
523 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
524 		    repo);
525 	}
526 
527 	return err;
528 }
529 
530 struct search_deltas_arg {
531 	struct got_packidx *packidx;
532 	struct got_pack *pack;
533 	struct got_object_idset *idset;
534 	struct got_pack_metavec *v;
535 	int delta_cache_fd;
536 	struct got_repository *repo;
537 	got_pack_progress_cb progress_cb;
538 	void *progress_arg;
539 	struct got_ratelimit *rl;
540 	got_cancel_cb cancel_cb;
541 	void *cancel_arg;
542 	int ncolored;
543 	int nfound;
544 	int ntrees;
545 	int ncommits;
546 };
547 
548 static const struct got_error *
549 search_delta_for_object(struct got_object_id *id, void *data, void *arg)
550 {
551 	const struct got_error *err;
552 	struct got_pack_meta *m = data;
553 	struct search_deltas_arg *a = arg;
554 	int obj_idx;
555 	struct got_object *obj = NULL;
556 
557 	if (a->cancel_cb) {
558 		err = (*a->cancel_cb)(a->cancel_arg);
559 		if (err)
560 			return err;
561 	}
562 
563 	if (!got_repo_check_packidx_bloom_filter(a->repo,
564 	    a->packidx->path_packidx, id))
565 		return NULL;
566 
567 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
568 	if (obj_idx == -1)
569 		return NULL;
570 
571 	/* TODO:
572 	 * Opening and closing an object just to check its flags
573 	 * is a bit expensive. We could have an imsg which requests
574 	 * plain type/size information for an object without doing
575 	 * work such as traversing the object's entire delta chain
576 	 * to find the base object type, and other such info which
577 	 * we don't really need here.
578 	 */
579 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
580 	    a->packidx, obj_idx, a->repo);
581 	if (err)
582 		return err;
583 
584 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
585 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
586 		    a->delta_cache_fd, a->repo);
587 		if (err)
588 			goto done;
589 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
590 		    a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
591 		    got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
592 	}
593 done:
594 	got_object_close(obj);
595 	return err;
596 }
597 
598 static const struct got_error *
599 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
600     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
601     struct got_repository *repo,
602     got_pack_progress_cb progress_cb, void *progress_arg,
603     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
604 {
605 	const struct got_error *err = NULL;
606 	char *path_packfile = NULL;
607 	struct got_packidx *packidx;
608 	struct got_pack *pack;
609 	struct search_deltas_arg sda;
610 
611 	err = find_pack_for_reuse(&packidx, repo);
612 	if (err)
613 		return err;
614 
615 	if (packidx == NULL)
616 		return NULL;
617 
618 	err = got_packidx_get_packfile_path(&path_packfile,
619 	    packidx->path_packidx);
620 	if (err)
621 		return err;
622 
623 	pack = got_repo_get_cached_pack(repo, path_packfile);
624 	if (pack == NULL) {
625 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
626 		if (err)
627 			goto done;
628 	}
629 
630 	sda.packidx = packidx;
631 	sda.pack = pack;
632 	sda.idset = idset;
633 	sda.v = v;
634 	sda.delta_cache_fd = delta_cache_fd;
635 	sda.repo = repo;
636 	sda.progress_cb = progress_cb;
637 	sda.progress_arg = progress_arg;
638 	sda.rl = rl;
639 	sda.cancel_cb = cancel_cb;
640 	sda.cancel_arg = cancel_arg;
641 	sda.ncolored = ncolored;
642 	sda.nfound = nfound;
643 	sda.ntrees = ntrees;
644 	sda.ncommits = ncommits;
645 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
646 done:
647 	free(path_packfile);
648 	return err;
649 }
650 
651 static const struct got_error *
652 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
653     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
654     struct got_repository *repo,
655     got_pack_progress_cb progress_cb, void *progress_arg,
656     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
657 {
658 	const struct got_error *err = NULL;
659 	struct got_pack_meta *m = NULL, *base = NULL;
660 	struct got_raw_object *raw = NULL, *base_raw = NULL;
661 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
662 	int i, j, ndeltas, best_ndeltas;
663 	off_t size, best_size;
664 	const int max_base_candidates = 3;
665 	size_t delta_memsize = 0;
666 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
667 	int outfd = -1;
668 
669 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
670 	for (i = 0; i < nmeta; i++) {
671 		if (cancel_cb) {
672 			err = (*cancel_cb)(cancel_arg);
673 			if (err)
674 				break;
675 		}
676 		err = report_progress(progress_cb, progress_arg, rl,
677 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
678 		    nreused + i, 0);
679 		if (err)
680 			goto done;
681 		m = meta[i];
682 
683 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
684 		    m->obj_type == GOT_OBJ_TYPE_TAG)
685 			continue;
686 
687 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
688 		if (err)
689 			goto done;
690 		m->size = raw->size;
691 
692 		if (raw->f == NULL) {
693 			err = got_deltify_init_mem(&m->dtab, raw->data,
694 			    raw->hdrlen, raw->size + raw->hdrlen);
695 		} else {
696 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
697 			    raw->size + raw->hdrlen);
698 		}
699 		if (err)
700 			goto done;
701 
702 		if (i > max_base_candidates) {
703 			struct got_pack_meta *n = NULL;
704 			n = meta[i - (max_base_candidates + 1)];
705 			got_deltify_free(n->dtab);
706 			n->dtab = NULL;
707 		}
708 
709 		best_size = raw->size;
710 		best_ndeltas = 0;
711 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
712 			if (cancel_cb) {
713 				err = (*cancel_cb)(cancel_arg);
714 				if (err)
715 					goto done;
716 			}
717 			base = meta[j];
718 			/* long chains make unpacking slow, avoid such bases */
719 			if (base->nchain >= 128 ||
720 			    base->obj_type != m->obj_type)
721 				continue;
722 
723 			err = got_object_raw_open(&base_raw, &outfd, repo,
724 			    &base->id);
725 			if (err)
726 				goto done;
727 
728 			if (raw->f == NULL && base_raw->f == NULL) {
729 				err = got_deltify_mem_mem(&deltas, &ndeltas,
730 				    raw->data, raw->hdrlen,
731 				    raw->size + raw->hdrlen,
732 				    base->dtab, base_raw->data,
733 				    base_raw->hdrlen,
734 				    base_raw->size + base_raw->hdrlen);
735 			} else if (raw->f == NULL) {
736 				err = got_deltify_mem_file(&deltas, &ndeltas,
737 				    raw->data, raw->hdrlen,
738 				    raw->size + raw->hdrlen,
739 				    base->dtab, base_raw->f,
740 				    base_raw->hdrlen,
741 				    base_raw->size + base_raw->hdrlen);
742 			} else if (base_raw->f == NULL) {
743 				err = got_deltify_file_mem(&deltas, &ndeltas,
744 				    raw->f, raw->hdrlen,
745 				    raw->size + raw->hdrlen,
746 				    base->dtab, base_raw->data,
747 				    base_raw->hdrlen,
748 				    base_raw->size + base_raw->hdrlen);
749 			} else {
750 				err = got_deltify(&deltas, &ndeltas,
751 				    raw->f, raw->hdrlen,
752 				    raw->size + raw->hdrlen,
753 				    base->dtab, base_raw->f, base_raw->hdrlen,
754 				    base_raw->size + base_raw->hdrlen);
755 			}
756 			got_object_raw_close(base_raw);
757 			base_raw = NULL;
758 			if (err)
759 				goto done;
760 
761 			size = delta_size(deltas, ndeltas);
762 			if (size + 32 < best_size){
763 				/*
764 				 * if we already picked a best delta,
765 				 * replace it.
766 				 */
767 				best_size = size;
768 				free(best_deltas);
769 				best_deltas = deltas;
770 				best_ndeltas = ndeltas;
771 				deltas = NULL;
772 				m->nchain = base->nchain + 1;
773 				m->prev = base;
774 				m->head = base->head;
775 				if (m->head == NULL)
776 					m->head = base;
777 			} else {
778 				free(deltas);
779 				deltas = NULL;
780 				ndeltas = 0;
781 			}
782 		}
783 
784 		if (best_ndeltas > 0) {
785 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
786 			    delta_memsize + best_size <= max_delta_memsize) {
787 				delta_memsize += best_size;
788 				err = encode_delta_in_mem(m, raw, best_deltas,
789 				    best_ndeltas, best_size, m->prev->size);
790 			} else {
791 				m->delta_offset = ftello(delta_cache);
792 				/*
793 				 * TODO:
794 				 * Storing compressed delta data in the delta
795 				 * cache file would probably be more efficient
796 				 * than writing uncompressed delta data here
797 				 * and compressing it while writing the pack
798 				 * file. This would also allow for reusing
799 				 * deltas in their compressed form.
800 				 */
801 				err = encode_delta(m, raw, best_deltas,
802 				    best_ndeltas, m->prev->size, delta_cache);
803 			}
804 			free(best_deltas);
805 			best_deltas = NULL;
806 			best_ndeltas = 0;
807 			if (err)
808 				goto done;
809 		}
810 
811 		got_object_raw_close(raw);
812 		raw = NULL;
813 	}
814 done:
815 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
816 		got_deltify_free(meta[i]->dtab);
817 		meta[i]->dtab = NULL;
818 	}
819 	if (raw)
820 		got_object_raw_close(raw);
821 	if (base_raw)
822 		got_object_raw_close(base_raw);
823 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
824 		err = got_error_from_errno("close");
825 	free(deltas);
826 	free(best_deltas);
827 	return err;
828 }
829 
830 static const struct got_error *
831 search_packidx(int *found, struct got_object_id *id,
832     struct got_repository *repo)
833 {
834 	const struct got_error *err = NULL;
835 	struct got_packidx *packidx = NULL;
836 	int idx;
837 
838 	*found = 0;
839 
840 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
841 	if (err == NULL)
842 		*found = 1; /* object is already packed */
843 	else if (err->code == GOT_ERR_NO_OBJ)
844 		err = NULL;
845 	return err;
846 }
847 
848 static const struct got_error *
849 add_object(int want_meta, struct got_object_idset *idset,
850     struct got_object_id *id, const char *path, int obj_type,
851     time_t mtime, int loose_obj_only, struct got_repository *repo,
852     int *ncolored, int *nfound, int *ntrees,
853     got_pack_progress_cb progress_cb, void *progress_arg,
854     struct got_ratelimit *rl)
855 {
856 	const struct got_error *err;
857 	struct got_pack_meta *m = NULL;
858 
859 	if (loose_obj_only) {
860 		int is_packed;
861 		err = search_packidx(&is_packed, id, repo);
862 		if (err)
863 			return err;
864 		if (is_packed && want_meta)
865 			return NULL;
866 	}
867 
868 	if (want_meta) {
869 		err = alloc_meta(&m, id, path, obj_type, mtime);
870 		if (err)
871 			return err;
872 
873 		(*nfound)++;
874 		err = report_progress(progress_cb, progress_arg, rl,
875 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
876 		if (err) {
877 			clear_meta(m);
878 			free(m);
879 			return err;
880 		}
881 	}
882 
883 	err = got_object_idset_add(idset, id, m);
884 	if (err) {
885 		clear_meta(m);
886 		free(m);
887 	}
888 	return err;
889 }
890 
891 static const struct got_error *
892 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
893     struct got_object_idset *idset, struct got_object_id *tree_id,
894     const char *dpath, time_t mtime, struct got_repository *repo,
895     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
896     got_pack_progress_cb progress_cb, void *progress_arg,
897     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
898 {
899 	const struct got_error *err;
900 	struct got_tree_object *tree;
901 	char *p = NULL;
902 	int i;
903 
904 	err = got_object_open_as_tree(&tree, repo, tree_id);
905 	if (err)
906 		return err;
907 
908 	(*ntrees)++;
909 	err = report_progress(progress_cb, progress_arg, rl,
910 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
911 	if (err)
912 		return err;
913 
914 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
915 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
916 		struct got_object_id *id = got_tree_entry_get_id(e);
917 		mode_t mode = got_tree_entry_get_mode(e);
918 
919 		if (cancel_cb) {
920 			err = (*cancel_cb)(cancel_arg);
921 			if (err)
922 				break;
923 		}
924 
925 		if (got_object_tree_entry_is_submodule(e) ||
926 		    got_object_idset_contains(idset, id))
927 			continue;
928 		
929 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
930 		    got_tree_entry_get_name(e)) == -1) {
931 			err = got_error_from_errno("asprintf");
932 			break;
933 		}
934 
935 		if (S_ISDIR(mode)) {
936 			struct got_object_qid *qid;
937 			err = got_object_qid_alloc(&qid, id);
938 			if (err)
939 				break;
940 			STAILQ_INSERT_TAIL(ids, qid, entry);
941 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
942 			err = add_object(want_meta, idset, id, p,
943 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo,
944 			    ncolored, nfound, ntrees,
945 			    progress_cb, progress_arg, rl);
946 			if (err)
947 				break;
948 		}
949 		free(p);
950 		p = NULL;
951 	}
952 
953 	got_object_tree_close(tree);
954 	free(p);
955 	return err;
956 }
957 
958 static const struct got_error *
959 load_tree(int want_meta, struct got_object_idset *idset,
960     struct got_object_id *tree_id, const char *dpath, time_t mtime,
961     struct got_repository *repo, int loose_obj_only,
962     int *ncolored, int *nfound, int *ntrees,
963     got_pack_progress_cb progress_cb, void *progress_arg,
964     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
965 {
966 	const struct got_error *err = NULL;
967 	struct got_object_id_queue tree_ids;
968 	struct got_object_qid *qid;
969 
970 	if (got_object_idset_contains(idset, tree_id))
971 		return NULL;
972 
973 	err = got_object_qid_alloc(&qid, tree_id);
974 	if (err)
975 		return err;
976 
977 	STAILQ_INIT(&tree_ids);
978 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
979 
980 	while (!STAILQ_EMPTY(&tree_ids)) {
981 		if (cancel_cb) {
982 			err = (*cancel_cb)(cancel_arg);
983 			if (err)
984 				break;
985 		}
986 
987 		qid = STAILQ_FIRST(&tree_ids);
988 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
989 
990 		if (got_object_idset_contains(idset, &qid->id)) {
991 			got_object_qid_free(qid);
992 			continue;
993 		}
994 
995 		err = add_object(want_meta, idset, &qid->id, dpath,
996 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo,
997 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
998 		if (err) {
999 			got_object_qid_free(qid);
1000 			break;
1001 		}
1002 
1003 		err = load_tree_entries(&tree_ids, want_meta, idset, &qid->id,
1004 		    dpath, mtime, repo, loose_obj_only, ncolored, nfound,
1005 		    ntrees, progress_cb, progress_arg, rl,
1006 		    cancel_cb, cancel_arg);
1007 		got_object_qid_free(qid);
1008 		if (err)
1009 			break;
1010 	}
1011 
1012 	got_object_id_queue_free(&tree_ids);
1013 	return err;
1014 }
1015 
1016 static const struct got_error *
1017 load_commit(int want_meta, struct got_object_idset *idset,
1018     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1019     int *ncolored, int *nfound, int *ntrees,
1020     got_pack_progress_cb progress_cb, void *progress_arg,
1021     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1022 {
1023 	const struct got_error *err;
1024 	struct got_commit_object *commit;
1025 
1026 	if (got_object_idset_contains(idset, id))
1027 		return NULL;
1028 
1029 	if (loose_obj_only) {
1030 		int is_packed;
1031 		err = search_packidx(&is_packed, id, repo);
1032 		if (err)
1033 			return err;
1034 		if (is_packed && want_meta)
1035 			return NULL;
1036 	}
1037 
1038 	err = got_object_open_as_commit(&commit, repo, id);
1039 	if (err)
1040 		return err;
1041 
1042 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1043 	    got_object_commit_get_committer_time(commit),
1044 	    loose_obj_only, repo,
1045 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1046 	if (err)
1047 		goto done;
1048 
1049 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1050 	    "", got_object_commit_get_committer_time(commit),
1051 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1052 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1053 done:
1054 	got_object_commit_close(commit);
1055 	return err;
1056 }
1057 
1058 static const struct got_error *
1059 load_tag(int want_meta, struct got_object_idset *idset,
1060     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1061     int *ncolored, int *nfound, int *ntrees,
1062     got_pack_progress_cb progress_cb, void *progress_arg,
1063     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1064 {
1065 	const struct got_error *err;
1066 	struct got_tag_object *tag = NULL;
1067 
1068 	if (got_object_idset_contains(idset, id))
1069 		return NULL;
1070 
1071 	if (loose_obj_only) {
1072 		int is_packed;
1073 		err = search_packidx(&is_packed, id, repo);
1074 		if (err)
1075 			return err;
1076 		if (is_packed && want_meta)
1077 			return NULL;
1078 	}
1079 
1080 	err = got_object_open_as_tag(&tag, repo, id);
1081 	if (err)
1082 		return err;
1083 
1084 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1085 	    got_object_tag_get_tagger_time(tag), loose_obj_only, repo,
1086 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1087 	if (err)
1088 		goto done;
1089 
1090 	switch (got_object_tag_get_object_type(tag)) {
1091 	case GOT_OBJ_TYPE_COMMIT:
1092 		err = load_commit(want_meta, idset,
1093 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1094 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1095 		    cancel_cb, cancel_arg);
1096 		break;
1097 	case GOT_OBJ_TYPE_TREE:
1098 		err = load_tree(want_meta, idset,
1099 		    got_object_tag_get_object_id(tag), "",
1100 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1101 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1102 		    cancel_cb, cancel_arg);
1103 		break;
1104 	default:
1105 		break;
1106 	}
1107 
1108 done:
1109 	got_object_tag_close(tag);
1110 	return err;
1111 }
1112 
1113 enum findtwixt_color {
1114 	COLOR_KEEP = 0,
1115 	COLOR_DROP,
1116 	COLOR_BLANK,
1117 	COLOR_SKIP,
1118 };
1119 
1120 static const int findtwixt_colors[] = {
1121 	COLOR_KEEP,
1122 	COLOR_DROP,
1123 	COLOR_BLANK,
1124 	COLOR_SKIP,
1125 };
1126 
1127 static const struct got_error *
1128 paint_commit(struct got_object_qid *qid, int color)
1129 {
1130 	if (color < 0 || color >= nitems(findtwixt_colors))
1131 		return got_error(GOT_ERR_RANGE);
1132 
1133 	qid->data = (void *)&findtwixt_colors[color];
1134 	return NULL;
1135 }
1136 
1137 static const struct got_error *
1138 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1139     int color, struct got_repository *repo)
1140 {
1141 	const struct got_error *err;
1142 	struct got_object_qid *qid;
1143 
1144 	err = got_object_qid_alloc(&qid, id);
1145 	if (err)
1146 		return err;
1147 
1148 	STAILQ_INSERT_TAIL(ids, qid, entry);
1149 	return paint_commit(qid, color);
1150 }
1151 
1152 struct append_id_arg {
1153 	struct got_object_id **array;
1154 	int idx;
1155 	struct got_object_idset *drop;
1156 	struct got_object_idset *skip;
1157 };
1158 
1159 static const struct got_error *
1160 append_id(struct got_object_id *id, void *data, void *arg)
1161 {
1162 	struct append_id_arg *a = arg;
1163 
1164 	if (got_object_idset_contains(a->skip, id) ||
1165 	    got_object_idset_contains(a->drop, id))
1166 		return NULL;
1167 
1168 	a->array[++a->idx] = got_object_id_dup(id);
1169 	if (a->array[a->idx] == NULL)
1170 		return got_error_from_errno("got_object_id_dup");
1171 
1172 	return NULL;
1173 }
1174 
1175 static const struct got_error *
1176 queue_commit_or_tag_id(struct got_object_id *id, int color,
1177     struct got_object_id_queue *ids, struct got_repository *repo)
1178 {
1179 	const struct got_error *err;
1180 	struct got_tag_object *tag = NULL;
1181 	int obj_type;
1182 
1183 	err = got_object_get_type(&obj_type, repo, id);
1184 	if (err)
1185 		return err;
1186 
1187 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1188 		err = got_object_open_as_tag(&tag, repo, id);
1189 		if (err)
1190 			return err;
1191 		obj_type = got_object_tag_get_object_type(tag);
1192 		id = got_object_tag_get_object_id(tag);
1193 	}
1194 
1195 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1196 		err = queue_commit_id(ids, id, color, repo);
1197 		if (err)
1198 			goto done;
1199 	}
1200 done:
1201 	if (tag)
1202 		got_object_tag_close(tag);
1203 	return err;
1204 }
1205 
1206 static const struct got_error *
1207 paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1208     struct got_object_idset *keep, struct got_object_idset *drop,
1209     struct got_object_idset *skip, struct got_repository *repo,
1210     got_pack_progress_cb progress_cb, void *progress_arg,
1211     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1212 {
1213 	const struct got_error *err = NULL;
1214 	struct got_commit_object *commit = NULL;
1215 	const struct got_object_id_queue *parents;
1216 	struct got_object_qid *qid;
1217 	int nqueued = nids, nskip = 0;
1218 
1219 	while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1220 		int color;
1221 
1222 		if (cancel_cb) {
1223 			err = cancel_cb(cancel_arg);
1224 			if (err)
1225 				break;
1226 		}
1227 
1228 		qid = STAILQ_FIRST(ids);
1229 		STAILQ_REMOVE_HEAD(ids, entry);
1230 		nqueued--;
1231 		color = *((int *)qid->data);
1232 		if (color == COLOR_SKIP)
1233 			nskip--;
1234 
1235 		if (got_object_idset_contains(skip, &qid->id)) {
1236 			got_object_qid_free(qid);
1237 			continue;
1238 		}
1239 
1240 		switch (color) {
1241 		case COLOR_KEEP:
1242 			if (got_object_idset_contains(keep, &qid->id)) {
1243 				got_object_qid_free(qid);
1244 				continue;
1245 			}
1246 			if (got_object_idset_contains(drop, &qid->id)) {
1247 				err = paint_commit(qid, COLOR_SKIP);
1248 				if (err)
1249 					goto done;
1250 				nskip++;
1251 			} else
1252 				(*ncolored)++;
1253 			err = got_object_idset_add(keep, &qid->id, NULL);
1254 			if (err)
1255 				goto done;
1256 			break;
1257 		case COLOR_DROP:
1258 			if (got_object_idset_contains(drop, &qid->id)) {
1259 				got_object_qid_free(qid);
1260 				continue;
1261 			}
1262 			if (got_object_idset_contains(keep, &qid->id)) {
1263 				err = paint_commit(qid, COLOR_SKIP);
1264 				if (err)
1265 					goto done;
1266 				nskip++;
1267 			} else
1268 				(*ncolored)++;
1269 			err = got_object_idset_add(drop, &qid->id, NULL);
1270 			if (err)
1271 				goto done;
1272 			break;
1273 		case COLOR_SKIP:
1274 			if (!got_object_idset_contains(skip, &qid->id)) {
1275 				err = got_object_idset_add(skip, &qid->id,
1276 				    NULL);
1277 				if (err)
1278 					goto done;
1279 			}
1280 			break;
1281 		default:
1282 			/* should not happen */
1283 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1284 			    "%s invalid commit color %d", __func__, color);
1285 			goto done;
1286 		}
1287 
1288 		err = report_progress(progress_cb, progress_arg, rl,
1289 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1290 		if (err)
1291 			break;
1292 
1293 
1294 		err = got_object_open_as_commit(&commit, repo, &qid->id);
1295 		if (err)
1296 			break;
1297 
1298 		parents = got_object_commit_get_parent_ids(commit);
1299 		if (parents) {
1300 			struct got_object_qid *pid;
1301 			color = *((int *)qid->data);
1302 			STAILQ_FOREACH(pid, parents, entry) {
1303 				err = queue_commit_id(ids, &pid->id, color,
1304 				    repo);
1305 				if (err)
1306 					break;
1307 				nqueued++;
1308 				if (color == COLOR_SKIP)
1309 					nskip++;
1310 			}
1311 		}
1312 
1313 		got_object_commit_close(commit);
1314 		commit = NULL;
1315 		got_object_qid_free(qid);
1316 	}
1317 done:
1318 	if (commit)
1319 		got_object_commit_close(commit);
1320 	return err;
1321 }
1322 
1323 static const struct got_error *
1324 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1325     struct got_object_id **head, int nhead,
1326     struct got_object_id **tail, int ntail,
1327     struct got_repository *repo,
1328     got_pack_progress_cb progress_cb, void *progress_arg,
1329     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1330 {
1331 	const struct got_error *err = NULL;
1332 	struct got_object_id_queue ids;
1333 	struct got_object_idset *keep, *drop, *skip = NULL;
1334 	int i, nkeep;
1335 
1336 	STAILQ_INIT(&ids);
1337 	*res = NULL;
1338 	*nres = 0;
1339 	*ncolored = 0;
1340 
1341 	keep = got_object_idset_alloc();
1342 	if (keep == NULL)
1343 		return got_error_from_errno("got_object_idset_alloc");
1344 
1345 	drop = got_object_idset_alloc();
1346 	if (drop == NULL) {
1347 		err = got_error_from_errno("got_object_idset_alloc");
1348 		goto done;
1349 	}
1350 
1351 	skip = got_object_idset_alloc();
1352 	if (skip == NULL) {
1353 		err = got_error_from_errno("got_object_idset_alloc");
1354 		goto done;
1355 	}
1356 
1357 	for (i = 0; i < nhead; i++) {
1358 		struct got_object_id *id = head[i];
1359 		if (id == NULL)
1360 			continue;
1361 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1362 		if (err)
1363 			goto done;
1364 	}		
1365 
1366 	for (i = 0; i < ntail; i++) {
1367 		struct got_object_id *id = tail[i];
1368 		if (id == NULL)
1369 			continue;
1370 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1371 		if (err)
1372 			goto done;
1373 	}
1374 
1375 	err = paint_commits(ncolored, &ids, nhead + ntail,
1376 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1377 	    cancel_cb, cancel_arg);
1378 	if (err)
1379 		goto done;
1380 
1381 	nkeep = got_object_idset_num_elements(keep);
1382 	if (nkeep > 0) {
1383 		struct append_id_arg arg;
1384 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1385 		if (arg.array == NULL) {
1386 			err = got_error_from_errno("calloc");
1387 			goto done;
1388 		}
1389 		arg.idx = -1;
1390 		arg.skip = skip;
1391 		arg.drop = drop;
1392 		err = got_object_idset_for_each(keep, append_id, &arg);
1393 		if (err) {
1394 			free(arg.array);
1395 			goto done;
1396 		}
1397 		*res = arg.array;
1398 		*nres = arg.idx + 1;
1399 	}
1400 done:
1401 	got_object_idset_free(keep);
1402 	got_object_idset_free(drop);
1403 	if (skip)
1404 		got_object_idset_free(skip);
1405 	got_object_id_queue_free(&ids);
1406 	return err;
1407 }
1408 
1409 static const struct got_error *
1410 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1411     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1412     struct got_object_id **ours, int nours, struct got_repository *repo,
1413     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1414     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1415 {
1416 	const struct got_error *err = NULL;
1417 	struct got_object_id **ids = NULL;
1418 	int i, nobj = 0, obj_type;
1419 
1420 	*ncolored = 0;
1421 	*nfound = 0;
1422 	*ntrees = 0;
1423 
1424 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1425 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1426 	if (err || nobj == 0)
1427 		goto done;
1428 
1429 	for (i = 0; i < ntheirs; i++) {
1430 		struct got_object_id *id = theirs[i];
1431 		if (id == NULL)
1432 			continue;
1433 		err = got_object_get_type(&obj_type, repo, id);
1434 		if (err)
1435 			return err;
1436 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1437 			err = load_commit(0, idset, id, repo,
1438 			    loose_obj_only, ncolored, nfound, ntrees,
1439 			    progress_cb, progress_arg, rl,
1440 			    cancel_cb, cancel_arg);
1441 			if (err)
1442 				goto done;
1443 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1444 			err = load_tag(0, idset, id, repo,
1445 			    loose_obj_only, ncolored, nfound, ntrees,
1446 			    progress_cb, progress_arg, rl,
1447 			    cancel_cb, cancel_arg);
1448 			if (err)
1449 				goto done;
1450 		}
1451 	}
1452 
1453 	for (i = 0; i < nobj; i++) {
1454 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1455 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1456 		    cancel_cb, cancel_arg);
1457 		if (err)
1458 			goto done;
1459 	}
1460 
1461 	for (i = 0; i < nours; i++) {
1462 		struct got_object_id *id = ours[i];
1463 		struct got_pack_meta *m;
1464 		if (id == NULL)
1465 			continue;
1466 		m = got_object_idset_get(idset, id);
1467 		if (m == NULL) {
1468 			err = got_object_get_type(&obj_type, repo, id);
1469 			if (err)
1470 				goto done;
1471 		} else
1472 			obj_type = m->obj_type;
1473 		if (obj_type != GOT_OBJ_TYPE_TAG)
1474 			continue;
1475 		err = load_tag(1, idset, id, repo, loose_obj_only,
1476 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1477 		    cancel_cb, cancel_arg);
1478 		if (err)
1479 			goto done;
1480 	}
1481 done:
1482 	for (i = 0; i < nobj; i++) {
1483 		free(ids[i]);
1484 	}
1485 	free(ids);
1486 	return err;
1487 }
1488 
1489 const struct got_error *
1490 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1491 {
1492 	size_t n;
1493 
1494 	SHA1Update(ctx, buf, len);
1495 	n = fwrite(buf, 1, len, f);
1496 	if (n != len)
1497 		return got_ferror(f, GOT_ERR_IO);
1498 	return NULL;
1499 }
1500 
1501 static void
1502 putbe32(char *b, uint32_t n)
1503 {
1504 	b[0] = n >> 24;
1505 	b[1] = n >> 16;
1506 	b[2] = n >> 8;
1507 	b[3] = n >> 0;
1508 }
1509 
1510 static int
1511 write_order_cmp(const void *pa, const void *pb)
1512 {
1513 	struct got_pack_meta *a, *b, *ahd, *bhd;
1514 
1515 	a = *(struct got_pack_meta **)pa;
1516 	b = *(struct got_pack_meta **)pb;
1517 	ahd = (a->head == NULL) ? a : a->head;
1518 	bhd = (b->head == NULL) ? b : b->head;
1519 	if (bhd->mtime < ahd->mtime)
1520 		return -1;
1521 	if (bhd->mtime > ahd->mtime)
1522 		return 1;
1523 	if (bhd < ahd)
1524 		return -1;
1525 	if (bhd > ahd)
1526 		return 1;
1527 	if (a->nchain != b->nchain)
1528 		return a->nchain - b->nchain;
1529 	if (a->mtime < b->mtime)
1530 		return -1;
1531 	if (a->mtime > b->mtime)
1532 		return 1;
1533 	return got_object_id_cmp(&a->id, &b->id);
1534 }
1535 
1536 static int
1537 reuse_write_order_cmp(const void *pa, const void *pb)
1538 {
1539 	struct got_pack_meta *a, *b;
1540 
1541 	a = *(struct got_pack_meta **)pa;
1542 	b = *(struct got_pack_meta **)pb;
1543 
1544 	if (a->reused_delta_offset < b->reused_delta_offset)
1545 		return -1;
1546 	if (a->reused_delta_offset > b->reused_delta_offset)
1547 		return 1;
1548 	return 0;
1549 }
1550 
1551 static const struct got_error *
1552 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1553 {
1554 	size_t i;
1555 
1556 	*hdrlen = 0;
1557 
1558 	hdr[0] = obj_type << 4;
1559 	hdr[0] |= len & 0xf;
1560 	len >>= 4;
1561 	for (i = 1; len != 0; i++){
1562 		if (i >= bufsize)
1563 			return got_error(GOT_ERR_NO_SPACE);
1564 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1565 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1566 		len >>= GOT_DELTA_SIZE_SHIFT;
1567 	}
1568 
1569 	*hdrlen = i;
1570 	return NULL;
1571 }
1572 
1573 static int
1574 packoff(char *hdr, off_t off)
1575 {
1576 	int i, j;
1577 	char rbuf[8];
1578 
1579 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1580 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1581 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1582 		    GOT_DELTA_SIZE_MORE;
1583 	}
1584 
1585 	j = 0;
1586 	while (i > 0)
1587 		hdr[j++] = rbuf[--i];
1588 	return j;
1589 }
1590 
1591 static const struct got_error *
1592 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1593     struct got_pack_meta *m)
1594 {
1595 	const struct got_error *err;
1596 	char buf[32];
1597 	int nh;
1598 
1599 	if (m->prev->off != 0) {
1600 		err = packhdr(&nh, buf, sizeof(buf),
1601 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1602 		if (err)
1603 			return err;
1604 		nh += packoff(buf + nh, m->off - m->prev->off);
1605 		err = hwrite(packfile, buf, nh, ctx);
1606 		if (err)
1607 			return err;
1608 		*packfile_size += nh;
1609 	} else {
1610 		err = packhdr(&nh, buf, sizeof(buf),
1611 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1612 		if (err)
1613 			return err;
1614 		err = hwrite(packfile, buf, nh, ctx);
1615 		if (err)
1616 			return err;
1617 		*packfile_size += nh;
1618 		err = hwrite(packfile, m->prev->id.sha1,
1619 		    sizeof(m->prev->id.sha1), ctx);
1620 		if (err)
1621 			return err;
1622 		*packfile_size += sizeof(m->prev->id.sha1);
1623 	}
1624 
1625 	return NULL;
1626 }
1627 
1628 static const struct got_error *
1629 write_packed_object(off_t *packfile_size, FILE *packfile,
1630     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1631     SHA1_CTX *ctx, struct got_repository *repo)
1632 {
1633 	const struct got_error *err = NULL;
1634 	struct got_deflate_checksum csum;
1635 	char buf[32];
1636 	int nh;
1637 	struct got_raw_object *raw = NULL;
1638 	off_t outlen;
1639 
1640 	csum.output_sha1 = ctx;
1641 	csum.output_crc = NULL;
1642 
1643 	m->off = ftello(packfile);
1644 	if (m->delta_len == 0) {
1645 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1646 		if (err)
1647 			goto done;
1648 		err = packhdr(&nh, buf, sizeof(buf),
1649 		    m->obj_type, raw->size);
1650 		if (err)
1651 			goto done;
1652 		err = hwrite(packfile, buf, nh, ctx);
1653 		if (err)
1654 			goto done;
1655 		*packfile_size += nh;
1656 		if (raw->f == NULL) {
1657 			err = got_deflate_to_file_mmap(&outlen,
1658 			    raw->data + raw->hdrlen, 0, raw->size,
1659 			    packfile, &csum);
1660 			if (err)
1661 				goto done;
1662 		} else {
1663 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1664 			    == -1) {
1665 				err = got_error_from_errno("fseeko");
1666 				goto done;
1667 			}
1668 			err = got_deflate_to_file(&outlen, raw->f,
1669 			    raw->size, packfile, &csum);
1670 			if (err)
1671 				goto done;
1672 		}
1673 		*packfile_size += outlen;
1674 		got_object_raw_close(raw);
1675 		raw = NULL;
1676 	} else if (m->delta_buf) {
1677 		err = deltahdr(packfile_size, ctx, packfile, m);
1678 		if (err)
1679 			goto done;
1680 		err = got_deflate_to_file_mmap(&outlen,
1681 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1682 		if (err)
1683 			goto done;
1684 		*packfile_size += outlen;
1685 		free(m->delta_buf);
1686 		m->delta_buf = NULL;
1687 	} else {
1688 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1689 		    == -1) {
1690 			err = got_error_from_errno("fseeko");
1691 			goto done;
1692 		}
1693 		err = deltahdr(packfile_size, ctx, packfile, m);
1694 		if (err)
1695 			goto done;
1696 		err = got_deflate_to_file(&outlen, delta_cache,
1697 		    m->delta_len, packfile, &csum);
1698 		if (err)
1699 			goto done;
1700 		*packfile_size += outlen;
1701 	}
1702 done:
1703 	if (raw)
1704 		got_object_raw_close(raw);
1705 	return err;
1706 }
1707 
1708 static const struct got_error *
1709 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1710     struct got_pack_meta **deltify, int ndeltify,
1711     struct got_pack_meta **reuse, int nreuse,
1712     int ncolored, int nfound, int ntrees, int nours,
1713     struct got_repository *repo,
1714     got_pack_progress_cb progress_cb, void *progress_arg,
1715     struct got_ratelimit *rl,
1716     got_cancel_cb cancel_cb, void *cancel_arg)
1717 {
1718 	const struct got_error *err = NULL;
1719 	int i;
1720 	SHA1_CTX ctx;
1721 	struct got_pack_meta *m;
1722 	char buf[32];
1723 	size_t n;
1724 	off_t packfile_size = 0;
1725 	int outfd = -1;
1726 
1727 	SHA1Init(&ctx);
1728 
1729 	err = hwrite(packfile, "PACK", 4, &ctx);
1730 	if (err)
1731 		return err;
1732 	putbe32(buf, GOT_PACKFILE_VERSION);
1733 	err = hwrite(packfile, buf, 4, &ctx);
1734 	if (err)
1735 		goto done;
1736 	putbe32(buf, ndeltify + nreuse);
1737 	err = hwrite(packfile, buf, 4, &ctx);
1738 	if (err)
1739 		goto done;
1740 
1741 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1742 	    write_order_cmp);
1743 	for (i = 0; i < ndeltify; i++) {
1744 		err = report_progress(progress_cb, progress_arg, rl,
1745 		    ncolored, nfound, ntrees, packfile_size, nours,
1746 		    ndeltify + nreuse, ndeltify + nreuse, i);
1747 		if (err)
1748 			goto done;
1749 		m = deltify[i];
1750 		err = write_packed_object(&packfile_size, packfile,
1751 		    delta_cache, m, &outfd, &ctx, repo);
1752 		if (err)
1753 			goto done;
1754 	}
1755 
1756 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1757 	    reuse_write_order_cmp);
1758 	for (i = 0; i < nreuse; i++) {
1759 		err = report_progress(progress_cb, progress_arg, rl,
1760 		    ncolored, nfound, ntrees, packfile_size, nours,
1761 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1762 		if (err)
1763 			goto done;
1764 		m = reuse[i];
1765 		err = write_packed_object(&packfile_size, packfile,
1766 		    delta_cache, m, &outfd, &ctx, repo);
1767 		if (err)
1768 			goto done;
1769 	}
1770 
1771 	SHA1Final(pack_sha1, &ctx);
1772 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1773 	if (n != SHA1_DIGEST_LENGTH)
1774 		err = got_ferror(packfile, GOT_ERR_IO);
1775 	packfile_size += SHA1_DIGEST_LENGTH;
1776 	packfile_size += sizeof(struct got_packfile_hdr);
1777 	if (progress_cb) {
1778 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1779 		    packfile_size, nours, ndeltify + nreuse,
1780 		    ndeltify + nreuse, ndeltify + nreuse);
1781 		if (err)
1782 			goto done;
1783 	}
1784 done:
1785 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1786 		err = got_error_from_errno("close");
1787 	return err;
1788 }
1789 
1790 static const struct got_error *
1791 remove_unused_object(struct got_object_id *id, void *data, void *arg)
1792 {
1793 	struct got_object_idset *idset = arg;
1794 
1795 	if (data == NULL)
1796 		got_object_idset_remove(NULL, idset, id);
1797 
1798 	return NULL;
1799 }
1800 
1801 static const struct got_error *
1802 remove_reused_object(struct got_object_id *id, void *data, void *arg)
1803 {
1804 	struct got_object_idset *idset = arg;
1805 	struct got_pack_meta *m = data;
1806 
1807 	if (m->have_reused_delta)
1808 		got_object_idset_remove(NULL, idset, id);
1809 
1810 	return NULL;
1811 }
1812 
1813 static const struct got_error *
1814 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1815 {
1816 	struct got_pack_meta *m = data;
1817 	struct got_pack_metavec *v = arg;
1818 
1819 	return add_meta(m, v);
1820 }
1821 
1822 const struct got_error *
1823 got_pack_create(uint8_t *packsha1, FILE *packfile,
1824     struct got_object_id **theirs, int ntheirs,
1825     struct got_object_id **ours, int nours,
1826     struct got_repository *repo, int loose_obj_only, int allow_empty,
1827     got_pack_progress_cb progress_cb, void *progress_arg,
1828     got_cancel_cb cancel_cb, void *cancel_arg)
1829 {
1830 	const struct got_error *err;
1831 	int delta_cache_fd = -1;
1832 	FILE *delta_cache = NULL;
1833 	struct got_object_idset *idset;
1834 	struct got_ratelimit rl;
1835 	struct got_pack_metavec deltify, reuse;
1836 	int ncolored = 0, nfound = 0, ntrees = 0;
1837 
1838 	memset(&deltify, 0, sizeof(deltify));
1839 	memset(&reuse, 0, sizeof(reuse));
1840 
1841 	got_ratelimit_init(&rl, 0, 500);
1842 
1843 	idset = got_object_idset_alloc();
1844 	if (idset == NULL)
1845 		return got_error_from_errno("got_object_idset_alloc");
1846 
1847 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1848 	    ntheirs, ours, nours, repo, loose_obj_only,
1849 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1850 	if (err)
1851 		return err;
1852 
1853 	err = got_object_idset_for_each(idset, remove_unused_object, idset);
1854 	if (err)
1855 		goto done;
1856 
1857 	if (progress_cb) {
1858 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1859 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1860 		if (err)
1861 			goto done;
1862 	}
1863 
1864 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1865 		err = got_error(GOT_ERR_CANNOT_PACK);
1866 		goto done;
1867 	}
1868 
1869 	delta_cache_fd = got_opentempfd();
1870 	if (delta_cache_fd == -1) {
1871 		err = got_error_from_errno("got_opentemp");
1872 		goto done;
1873 	}
1874 
1875 	reuse.metasz = 64;
1876 	reuse.meta = calloc(reuse.metasz,
1877 	    sizeof(struct got_pack_meta *));
1878 	if (reuse.meta == NULL) {
1879 		err = got_error_from_errno("calloc");
1880 		goto done;
1881 	}
1882 
1883 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1884 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1885 	    cancel_cb, cancel_arg);
1886 	if (err)
1887 		goto done;
1888 	if (reuse.nmeta > 0) {
1889 		err = got_object_idset_for_each(idset,
1890 		    remove_reused_object, idset);
1891 		if (err)
1892 			goto done;
1893 	}
1894 
1895 	delta_cache = fdopen(delta_cache_fd, "a+");
1896 	if (delta_cache == NULL) {
1897 		err = got_error_from_errno("fdopen");
1898 		goto done;
1899 	}
1900 	delta_cache_fd = -1;
1901 
1902 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1903 		err = got_error_from_errno("fseeko");
1904 		goto done;
1905 	}
1906 
1907 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1908 	    sizeof(struct got_pack_meta *));
1909 	if (deltify.meta == NULL) {
1910 		err = got_error_from_errno("calloc");
1911 		goto done;
1912 	}
1913 	deltify.metasz = got_object_idset_num_elements(idset);
1914 
1915 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1916 	if (err)
1917 		goto done;
1918 	if (deltify.nmeta > 0) {
1919 		err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1920 		    nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1921 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1922 		if (err)
1923 			goto done;
1924 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1925 			err = got_error_from_errno("fseeko");
1926 			goto done;
1927 		}
1928 	}
1929 
1930 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1931 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1932 	    nours, repo, progress_cb, progress_arg, &rl,
1933 	    cancel_cb, cancel_arg);
1934 	if (err)
1935 		goto done;
1936 done:
1937 	free_nmeta(deltify.meta, deltify.nmeta);
1938 	free_nmeta(reuse.meta, reuse.nmeta);
1939 	got_object_idset_free(idset);
1940 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1941 		err = got_error_from_errno("close");
1942 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1943 		err = got_error_from_errno("fclose");
1944 	return err;
1945 }