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