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 #include <sys/mman.h>
25 
26 #include <endian.h>
27 #include <errno.h>
28 #include <stdint.h>
29 #include <imsg.h>
30 #include <inttypes.h>
31 #include <poll.h>
32 #include <stdio.h>
33 #include <stdlib.h>
34 #include <string.h>
35 #include <sha1.h>
36 #include <sha2.h>
37 #include <time.h>
38 #include <unistd.h>
39 #include <limits.h>
40 #include <zlib.h>
41 
42 #include "got_error.h"
43 #include "got_cancel.h"
44 #include "got_object.h"
45 #include "got_path.h"
46 #include "got_reference.h"
47 #include "got_repository_admin.h"
48 
49 #include "got_lib_deltify.h"
50 #include "got_lib_delta.h"
51 #include "got_lib_object.h"
52 #include "got_lib_object_idset.h"
53 #include "got_lib_object_cache.h"
54 #include "got_lib_deflate.h"
55 #include "got_lib_ratelimit.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_pack_create.h"
58 #include "got_lib_repository.h"
59 #include "got_lib_inflate.h"
60 #include "got_lib_poll.h"
61 
62 #include "murmurhash2.h"
63 
64 #ifndef MIN
65 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
66 #endif
67 
68 #ifndef MAX
69 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
70 #endif
71 
72 #ifndef nitems
73 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
74 #endif
75 
76 static const struct got_error *
77 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
78     const char *path, int obj_type, time_t mtime, uint32_t seed)
79 {
80 	struct got_pack_meta *m;
81 
82 	*new = NULL;
83 
84 	m = calloc(1, sizeof(*m));
85 	if (m == NULL)
86 		return got_error_from_errno("calloc");
87 
88 	memcpy(&m->id, id, sizeof(m->id));
89 
90 	m->path_hash = murmurhash2(path, strlen(path), seed);
91 	m->obj_type = obj_type;
92 	m->mtime = mtime;
93 	*new = m;
94 	return NULL;
95 }
96 
97 static void
98 clear_meta(struct got_pack_meta *meta)
99 {
100 	if (meta == NULL)
101 		return;
102 	meta->path_hash = 0;
103 	free(meta->delta_buf);
104 	meta->delta_buf = NULL;
105 	free(meta->base_obj_id);
106 	meta->base_obj_id = NULL;
107 	meta->reused_delta_offset = 0;
108 }
109 
110 static void
111 free_nmeta(struct got_pack_meta **meta, int nmeta)
112 {
113 	int i;
114 
115 	for (i = 0; i < nmeta; i++)
116 		clear_meta(meta[i]);
117 	free(meta);
118 }
119 
120 static int
121 delta_order_cmp(const void *pa, const void *pb)
122 {
123 	struct got_pack_meta *a, *b;
124 
125 	a = *(struct got_pack_meta **)pa;
126 	b = *(struct got_pack_meta **)pb;
127 
128 	if (a->obj_type != b->obj_type)
129 		return a->obj_type - b->obj_type;
130 	if (a->path_hash < b->path_hash)
131 		return -1;
132 	if (a->path_hash > b->path_hash)
133 		return 1;
134 	if (a->mtime < b->mtime)
135 		return -1;
136 	if (a->mtime > b->mtime)
137 		return 1;
138 	return got_object_id_cmp(&a->id, &b->id);
139 }
140 
141 static off_t
142 delta_size(struct got_delta_instruction *deltas, int ndeltas)
143 {
144 	int i;
145 	off_t size = 32;
146 	for (i = 0; i < ndeltas; i++) {
147 		if (deltas[i].copy)
148 			size += GOT_DELTA_SIZE_SHIFT;
149 		else
150 			size += deltas[i].len + 1;
151 	}
152 	return size;
153 }
154 
155 static const struct got_error *
156 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
157 {
158 	char *n;
159 
160 	if (*len + nseg >= *sz) {
161 		while (*len + nseg >= *sz)
162 			*sz += *sz / 2;
163 		n = realloc(*p, *sz);
164 		if (n == NULL)
165 			return got_error_from_errno("realloc");
166 		*p = n;
167 	}
168 	memcpy(*p + *len, seg, nseg);
169 	*len += nseg;
170 	return NULL;
171 }
172 
173 static const struct got_error *
174 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
175     struct got_delta_instruction *deltas, int ndeltas,
176     off_t delta_size, off_t base_size)
177 {
178 	const struct got_error *err;
179 	unsigned char buf[16], *bp;
180 	int i, j;
181 	size_t len = 0, compressed_len;
182 	off_t bufsize = delta_size;
183 	off_t n;
184 	struct got_delta_instruction *d;
185 	uint8_t *delta_buf;
186 
187 	delta_buf = malloc(bufsize);
188 	if (delta_buf == NULL)
189 		return got_error_from_errno("malloc");
190 
191 	/* base object size */
192 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
193 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
194 	for (i = 1; n > 0; i++) {
195 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
196 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
197 		n >>= GOT_DELTA_SIZE_SHIFT;
198 	}
199 	err = append(&delta_buf, &len, &bufsize, buf, i);
200 	if (err)
201 		goto done;
202 
203 	/* target object size */
204 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
205 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
206 	for (i = 1; n > 0; i++) {
207 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
208 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
209 		n >>= GOT_DELTA_SIZE_SHIFT;
210 	}
211 	err = append(&delta_buf, &len, &bufsize, buf, i);
212 	if (err)
213 		goto done;
214 
215 	for (j = 0; j < ndeltas; j++) {
216 		d = &deltas[j];
217 		if (d->copy) {
218 			n = d->offset;
219 			bp = &buf[1];
220 			buf[0] = GOT_DELTA_BASE_COPY;
221 			for (i = 0; i < 4; i++) {
222 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
223 				buf[0] |= 1 << i;
224 				*bp++ = n & 0xff;
225 				n >>= 8;
226 				if (n == 0)
227 					break;
228 			}
229 
230 			n = d->len;
231 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
232 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
233 				for (i = 0; i < 3 && n > 0; i++) {
234 					buf[0] |= 1 << (i + 4);
235 					*bp++ = n & 0xff;
236 					n >>= 8;
237 				}
238 			}
239 			err = append(&delta_buf, &len, &bufsize,
240 			    buf, bp - buf);
241 			if (err)
242 				goto done;
243 		} else if (o->f == NULL) {
244 			n = 0;
245 			while (n != d->len) {
246 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
247 				err = append(&delta_buf, &len, &bufsize,
248 				    buf, 1);
249 				if (err)
250 					goto done;
251 				err = append(&delta_buf, &len, &bufsize,
252 				    o->data + o->hdrlen + d->offset + n,
253 				    buf[0]);
254 				if (err)
255 					goto done;
256 				n += buf[0];
257 			}
258 		} else {
259 			char content[128];
260 			size_t r;
261 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
262 				err = got_error_from_errno("fseeko");
263 				goto done;
264 			}
265 			n = 0;
266 			while (n != d->len) {
267 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
268 				err = append(&delta_buf, &len, &bufsize,
269 				    buf, 1);
270 				if (err)
271 					goto done;
272 				r = fread(content, 1, buf[0], o->f);
273 				if (r != buf[0]) {
274 					err = got_ferror(o->f, GOT_ERR_IO);
275 					goto done;
276 				}
277 				err = append(&delta_buf, &len, &bufsize,
278 				    content, buf[0]);
279 				if (err)
280 					goto done;
281 				n += buf[0];
282 			}
283 		}
284 	}
285 
286 	err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
287 	    NULL, NULL, delta_buf, 0, len);
288 	if (err)
289 		goto done;
290 
291 	m->delta_len = len;
292 	m->delta_compressed_len = compressed_len;
293 done:
294 	free(delta_buf);
295 	return err;
296 }
297 
298 static const struct got_error *
299 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
300     struct got_delta_instruction *deltas, int ndeltas,
301     off_t base_size, FILE *f)
302 {
303 	const struct got_error *err;
304 	unsigned char buf[16], *bp;
305 	int i, j;
306 	off_t n;
307 	struct got_deflate_buf zb;
308 	struct got_delta_instruction *d;
309 	off_t delta_len = 0, compressed_len = 0;
310 
311 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
312 	if (err)
313 		return err;
314 
315 	/* base object size */
316 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
317 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
318 	for (i = 1; n > 0; i++) {
319 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
320 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
321 		n >>= GOT_DELTA_SIZE_SHIFT;
322 	}
323 
324 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
325 	    buf, 0, i, f, NULL);
326 	if (err)
327 		goto done;
328 	delta_len += i;
329 
330 	/* target object size */
331 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
332 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
333 	for (i = 1; n > 0; i++) {
334 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
335 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
336 		n >>= GOT_DELTA_SIZE_SHIFT;
337 	}
338 
339 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
340 	    buf, 0, i, f, NULL);
341 	if (err)
342 		goto done;
343 	delta_len += i;
344 
345 	for (j = 0; j < ndeltas; j++) {
346 		d = &deltas[j];
347 		if (d->copy) {
348 			n = d->offset;
349 			bp = &buf[1];
350 			buf[0] = GOT_DELTA_BASE_COPY;
351 			for (i = 0; i < 4; i++) {
352 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
353 				buf[0] |= 1 << i;
354 				*bp++ = n & 0xff;
355 				n >>= 8;
356 				if (n == 0)
357 					break;
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 			err = got_deflate_append_to_file_mmap(&zb,
369 			    &compressed_len, buf, 0, bp - buf, f, NULL);
370 			if (err)
371 				goto done;
372 			delta_len += (bp - buf);
373 		} else if (o->f == NULL) {
374 			n = 0;
375 			while (n != d->len) {
376 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
377 				err = got_deflate_append_to_file_mmap(&zb,
378 				    &compressed_len, buf, 0, 1, f, NULL);
379 				if (err)
380 					goto done;
381 				delta_len++;
382 				err = got_deflate_append_to_file_mmap(&zb,
383 				    &compressed_len,
384 				    o->data + o->hdrlen + d->offset + n, 0,
385 				    buf[0], f, NULL);
386 				if (err)
387 					goto done;
388 				delta_len += buf[0];
389 				n += buf[0];
390 			}
391 		} else {
392 			char content[128];
393 			size_t r;
394 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
395 				err = got_error_from_errno("fseeko");
396 				goto done;
397 			}
398 			n = 0;
399 			while (n != d->len) {
400 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
401 				err = got_deflate_append_to_file_mmap(&zb,
402 				    &compressed_len, buf, 0, 1, f, NULL);
403 				if (err)
404 					goto done;
405 				delta_len++;
406 				r = fread(content, 1, buf[0], o->f);
407 				if (r != buf[0]) {
408 					err = got_ferror(o->f, GOT_ERR_IO);
409 					goto done;
410 				}
411 				err = got_deflate_append_to_file_mmap(&zb,
412 				    &compressed_len, content, 0, buf[0], f,
413 				    NULL);
414 				if (err)
415 					goto done;
416 				delta_len += buf[0];
417 				n += buf[0];
418 			}
419 		}
420 	}
421 
422 	err = got_deflate_flush(&zb, f, NULL, &compressed_len);
423 	if (err)
424 		goto done;
425 
426 	/* sanity check */
427 	if (compressed_len != ftello(f) - m->delta_offset) {
428 		err = got_error(GOT_ERR_COMPRESSION);
429 		goto done;
430 	}
431 
432 	m->delta_len = delta_len;
433 	m->delta_compressed_len = compressed_len;
434 done:
435 	got_deflate_end(&zb);
436 	return err;
437 }
438 
439 const struct got_error *
440 got_pack_report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
441     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
442     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
443     int nobj_written)
444 {
445 	const struct got_error *err;
446 	int elapsed;
447 
448 	if (progress_cb == NULL)
449 		return NULL;
450 
451 	err = got_ratelimit_check(&elapsed, rl);
452 	if (err || !elapsed)
453 		return err;
454 
455 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
456 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
457 }
458 
459 const struct got_error *
460 got_pack_add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
461 {
462 	if (v->nmeta == v->metasz){
463 		size_t newsize = 2 * v->metasz;
464 		struct got_pack_meta **new;
465 		new = reallocarray(v->meta, newsize, sizeof(*new));
466 		if (new == NULL)
467 			return got_error_from_errno("reallocarray");
468 		v->meta = new;
469 		v->metasz = newsize;
470 	}
471 
472 	v->meta[v->nmeta++] = m;
473 	return NULL;
474 }
475 
476 const struct got_error *
477 got_pack_find_pack_for_reuse(struct got_packidx **best_packidx,
478     struct got_repository *repo)
479 {
480 	const struct got_error *err = NULL;
481 	struct got_pathlist_entry *pe;
482 	const char *best_packidx_path = NULL;
483 	int nobj_max = 0;
484 
485 	*best_packidx = NULL;
486 
487 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
488 		const char *path_packidx = pe->path;
489 		struct got_packidx *packidx;
490 		int nobj;
491 
492 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
493 		if (err)
494 			break;
495 
496 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
497 		if (nobj > nobj_max) {
498 			best_packidx_path = path_packidx;
499 			nobj_max = nobj;
500 		}
501 	}
502 
503 	if (best_packidx_path) {
504 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
505 		    repo);
506 	}
507 
508 	return err;
509 }
510 
511 const struct got_error *
512 got_pack_cache_pack_for_packidx(struct got_pack **pack,
513     struct got_packidx *packidx, struct got_repository *repo)
514 {
515 	const struct got_error *err;
516 	char *path_packfile = NULL;
517 
518 	err = got_packidx_get_packfile_path(&path_packfile,
519 	    packidx->path_packidx);
520 	if (err)
521 		return err;
522 
523 	*pack = got_repo_get_cached_pack(repo, path_packfile);
524 	if (*pack == NULL) {
525 		err = got_repo_cache_pack(pack, repo, path_packfile, packidx);
526 		if (err)
527 			goto done;
528 	}
529 done:
530 	free(path_packfile);
531 	return err;
532 }
533 
534 static const struct got_error *
535 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
536     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
537     struct got_repository *repo,
538     got_pack_progress_cb progress_cb, void *progress_arg,
539     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
540 {
541 	const struct got_error *err = NULL;
542 	struct got_pack_meta *m = NULL, *base = NULL;
543 	struct got_raw_object *raw = NULL, *base_raw = NULL;
544 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
545 	int i, j, ndeltas, best_ndeltas;
546 	off_t size, best_size;
547 	const int max_base_candidates = 3;
548 	size_t delta_memsize = 0;
549 	const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
550 	int outfd = -1;
551 	uint32_t delta_seed;
552 
553 	delta_seed = arc4random();
554 
555 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
556 	for (i = 0; i < nmeta; i++) {
557 		if (cancel_cb) {
558 			err = (*cancel_cb)(cancel_arg);
559 			if (err)
560 				break;
561 		}
562 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
563 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
564 		    nreused + i, 0);
565 		if (err)
566 			goto done;
567 		m = meta[i];
568 
569 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
570 		    m->obj_type == GOT_OBJ_TYPE_TAG)
571 			continue;
572 
573 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
574 		if (err)
575 			goto done;
576 		m->size = raw->size;
577 
578 		if (raw->f == NULL) {
579 			err = got_deltify_init_mem(&m->dtab, raw->data,
580 			    raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
581 		} else {
582 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
583 			    raw->size + raw->hdrlen, delta_seed);
584 		}
585 		if (err)
586 			goto done;
587 
588 		if (i > max_base_candidates) {
589 			struct got_pack_meta *n = NULL;
590 			n = meta[i - (max_base_candidates + 1)];
591 			got_deltify_free(n->dtab);
592 			n->dtab = NULL;
593 		}
594 
595 		best_size = raw->size;
596 		best_ndeltas = 0;
597 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
598 			if (cancel_cb) {
599 				err = (*cancel_cb)(cancel_arg);
600 				if (err)
601 					goto done;
602 			}
603 			base = meta[j];
604 			/* long chains make unpacking slow, avoid such bases */
605 			if (base->nchain >= 128 ||
606 			    base->obj_type != m->obj_type)
607 				continue;
608 
609 			err = got_object_raw_open(&base_raw, &outfd, repo,
610 			    &base->id);
611 			if (err)
612 				goto done;
613 
614 			if (raw->f == NULL && base_raw->f == NULL) {
615 				err = got_deltify_mem_mem(&deltas, &ndeltas,
616 				    raw->data, raw->hdrlen,
617 				    raw->size + raw->hdrlen, delta_seed,
618 				    base->dtab, base_raw->data,
619 				    base_raw->hdrlen,
620 				    base_raw->size + base_raw->hdrlen);
621 			} else if (raw->f == NULL) {
622 				err = got_deltify_mem_file(&deltas, &ndeltas,
623 				    raw->data, raw->hdrlen,
624 				    raw->size + raw->hdrlen, delta_seed,
625 				    base->dtab, base_raw->f,
626 				    base_raw->hdrlen,
627 				    base_raw->size + base_raw->hdrlen);
628 			} else if (base_raw->f == NULL) {
629 				err = got_deltify_file_mem(&deltas, &ndeltas,
630 				    raw->f, raw->hdrlen,
631 				    raw->size + raw->hdrlen, delta_seed,
632 				    base->dtab, base_raw->data,
633 				    base_raw->hdrlen,
634 				    base_raw->size + base_raw->hdrlen);
635 			} else {
636 				err = got_deltify(&deltas, &ndeltas,
637 				    raw->f, raw->hdrlen,
638 				    raw->size + raw->hdrlen, delta_seed,
639 				    base->dtab, base_raw->f, base_raw->hdrlen,
640 				    base_raw->size + base_raw->hdrlen);
641 			}
642 			got_object_raw_close(base_raw);
643 			base_raw = NULL;
644 			if (err)
645 				goto done;
646 
647 			size = delta_size(deltas, ndeltas);
648 			if (size + 32 < best_size){
649 				/*
650 				 * if we already picked a best delta,
651 				 * replace it.
652 				 */
653 				best_size = size;
654 				free(best_deltas);
655 				best_deltas = deltas;
656 				best_ndeltas = ndeltas;
657 				deltas = NULL;
658 				m->nchain = base->nchain + 1;
659 				m->prev = base;
660 				m->head = base->head;
661 				if (m->head == NULL)
662 					m->head = base;
663 			} else {
664 				free(deltas);
665 				deltas = NULL;
666 				ndeltas = 0;
667 			}
668 		}
669 
670 		if (best_ndeltas > 0) {
671 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
672 			    delta_memsize + best_size <= max_delta_memsize) {
673 				delta_memsize += best_size;
674 				err = encode_delta_in_mem(m, raw, best_deltas,
675 				    best_ndeltas, best_size, m->prev->size);
676 			} else {
677 				m->delta_offset = ftello(delta_cache);
678 				err = encode_delta(m, raw, best_deltas,
679 				    best_ndeltas, m->prev->size, delta_cache);
680 			}
681 			free(best_deltas);
682 			best_deltas = NULL;
683 			best_ndeltas = 0;
684 			if (err)
685 				goto done;
686 		}
687 
688 		got_object_raw_close(raw);
689 		raw = NULL;
690 	}
691 done:
692 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
693 		got_deltify_free(meta[i]->dtab);
694 		meta[i]->dtab = NULL;
695 	}
696 	if (raw)
697 		got_object_raw_close(raw);
698 	if (base_raw)
699 		got_object_raw_close(base_raw);
700 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
701 		err = got_error_from_errno("close");
702 	free(deltas);
703 	free(best_deltas);
704 	return err;
705 }
706 
707 static const struct got_error *
708 search_packidx(int *found, struct got_object_id *id,
709     struct got_repository *repo)
710 {
711 	const struct got_error *err = NULL;
712 	struct got_packidx *packidx = NULL;
713 	int idx;
714 
715 	*found = 0;
716 
717 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
718 	if (err == NULL)
719 		*found = 1; /* object is already packed */
720 	else if (err->code == GOT_ERR_NO_OBJ)
721 		err = NULL;
722 	return err;
723 }
724 
725 const struct got_error *
726 got_pack_add_object(int want_meta, struct got_object_idset *idset,
727     struct got_object_id *id, const char *path, int obj_type,
728     time_t mtime, uint32_t seed, int loose_obj_only,
729     struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
730     got_pack_progress_cb progress_cb, void *progress_arg,
731     struct got_ratelimit *rl)
732 {
733 	const struct got_error *err;
734 	struct got_pack_meta *m = NULL;
735 
736 	if (loose_obj_only) {
737 		int is_packed;
738 		err = search_packidx(&is_packed, id, repo);
739 		if (err)
740 			return err;
741 		if (is_packed && want_meta)
742 			return NULL;
743 	}
744 
745 	if (want_meta) {
746 		err = alloc_meta(&m, id, path, obj_type, mtime, seed);
747 		if (err)
748 			return err;
749 
750 		(*nfound)++;
751 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
752 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
753 		if (err) {
754 			clear_meta(m);
755 			free(m);
756 			return err;
757 		}
758 	}
759 
760 	err = got_object_idset_add(idset, id, m);
761 	if (err) {
762 		clear_meta(m);
763 		free(m);
764 	}
765 	return err;
766 }
767 
768 const struct got_error *
769 got_pack_load_tree_entries(struct got_object_id_queue *ids, int want_meta,
770     struct got_object_idset *idset, struct got_object_idset *idset_exclude,
771     struct got_tree_object *tree,
772     const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
773     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
774     got_pack_progress_cb progress_cb, void *progress_arg,
775     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
776 {
777 	const struct got_error *err;
778 	char *p = NULL;
779 	int i;
780 
781 	(*ntrees)++;
782 	err = got_pack_report_progress(progress_cb, progress_arg, rl,
783 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
784 	if (err)
785 		return err;
786 
787 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
788 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
789 		struct got_object_id *id = got_tree_entry_get_id(e);
790 		mode_t mode = got_tree_entry_get_mode(e);
791 
792 		if (cancel_cb) {
793 			err = (*cancel_cb)(cancel_arg);
794 			if (err)
795 				break;
796 		}
797 
798 		if (got_object_tree_entry_is_submodule(e) ||
799 		    got_object_idset_contains(idset, id) ||
800 		    got_object_idset_contains(idset_exclude, id))
801 			continue;
802 
803 		/*
804 		 * If got-read-pack is crawling trees for us then
805 		 * we are only here to collect blob IDs.
806 		 */
807 		if (ids == NULL && S_ISDIR(mode))
808 			continue;
809 
810 		if (asprintf(&p, "%s%s%s", dpath,
811 		    got_path_is_root_dir(dpath) ? "" : "/",
812 		    got_tree_entry_get_name(e)) == -1) {
813 			err = got_error_from_errno("asprintf");
814 			break;
815 		}
816 
817 		if (S_ISDIR(mode)) {
818 			struct got_object_qid *qid;
819 			err = got_object_qid_alloc(&qid, id);
820 			if (err)
821 				break;
822 			qid->data = p;
823 			p = NULL;
824 			STAILQ_INSERT_TAIL(ids, qid, entry);
825 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
826 			err = got_pack_add_object(want_meta,
827 			    want_meta ? idset : idset_exclude, id, p,
828 			    GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
829 			    repo, ncolored, nfound, ntrees,
830 			    progress_cb, progress_arg, rl);
831 			if (err)
832 				break;
833 			free(p);
834 			p = NULL;
835 		} else {
836 			free(p);
837 			p = NULL;
838 		}
839 	}
840 
841 	free(p);
842 	return err;
843 }
844 
845 const struct got_error *
846 got_pack_load_tree(int want_meta, struct got_object_idset *idset,
847     struct got_object_idset *idset_exclude,
848     struct got_object_id *tree_id, const char *dpath, time_t mtime,
849     uint32_t seed, struct got_repository *repo, int loose_obj_only,
850     int *ncolored, int *nfound, int *ntrees,
851     got_pack_progress_cb progress_cb, void *progress_arg,
852     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
853 {
854 	const struct got_error *err = NULL;
855 	struct got_object_id_queue tree_ids;
856 	struct got_object_qid *qid;
857 	struct got_tree_object *tree = NULL;
858 
859 	if (got_object_idset_contains(idset, tree_id) ||
860 	    got_object_idset_contains(idset_exclude, tree_id))
861 		return NULL;
862 
863 	err = got_object_qid_alloc(&qid, tree_id);
864 	if (err)
865 		return err;
866 	qid->data = strdup(dpath);
867 	if (qid->data == NULL) {
868 		err = got_error_from_errno("strdup");
869 		got_object_qid_free(qid);
870 		return err;
871 	}
872 
873 	STAILQ_INIT(&tree_ids);
874 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
875 
876 	while (!STAILQ_EMPTY(&tree_ids)) {
877 		const char *path;
878 		if (cancel_cb) {
879 			err = (*cancel_cb)(cancel_arg);
880 			if (err)
881 				break;
882 		}
883 
884 		qid = STAILQ_FIRST(&tree_ids);
885 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
886 		path = qid->data;
887 
888 		if (got_object_idset_contains(idset, &qid->id) ||
889 		    got_object_idset_contains(idset_exclude, &qid->id)) {
890 			free(qid->data);
891 			got_object_qid_free(qid);
892 			continue;
893 		}
894 
895 		err = got_pack_add_object(want_meta,
896 		    want_meta ? idset : idset_exclude,
897 		    &qid->id, path, GOT_OBJ_TYPE_TREE,
898 		    mtime, seed, loose_obj_only, repo,
899 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
900 		if (err) {
901 			free(qid->data);
902 			got_object_qid_free(qid);
903 			break;
904 		}
905 
906 		err = got_object_open_as_tree(&tree, repo, &qid->id);
907 		if (err) {
908 			free(qid->data);
909 			got_object_qid_free(qid);
910 			break;
911 		}
912 
913 		err = got_pack_load_tree_entries(&tree_ids, want_meta, idset,
914 		    idset_exclude, tree, path, mtime, seed, repo,
915 		    loose_obj_only, ncolored, nfound, ntrees,
916 		    progress_cb, progress_arg, rl,
917 		    cancel_cb, cancel_arg);
918 		free(qid->data);
919 		got_object_qid_free(qid);
920 		if (err)
921 			break;
922 
923 		got_object_tree_close(tree);
924 		tree = NULL;
925 	}
926 
927 	STAILQ_FOREACH(qid, &tree_ids, entry)
928 		free(qid->data);
929 	got_object_id_queue_free(&tree_ids);
930 	if (tree)
931 		got_object_tree_close(tree);
932 	return err;
933 }
934 
935 static const struct got_error *
936 load_commit(int want_meta, struct got_object_idset *idset,
937     struct got_object_idset *idset_exclude,
938     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
939     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
940     got_pack_progress_cb progress_cb, void *progress_arg,
941     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
942 {
943 	const struct got_error *err;
944 	struct got_commit_object *commit;
945 
946 	if (got_object_idset_contains(idset, id) ||
947 	    got_object_idset_contains(idset_exclude, id))
948 		return NULL;
949 
950 	if (loose_obj_only) {
951 		int is_packed;
952 		err = search_packidx(&is_packed, id, repo);
953 		if (err)
954 			return err;
955 		if (is_packed && want_meta)
956 			return NULL;
957 	}
958 
959 	err = got_object_open_as_commit(&commit, repo, id);
960 	if (err)
961 		return err;
962 
963 	err = got_pack_add_object(want_meta,
964 	    want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_COMMIT,
965 	    got_object_commit_get_committer_time(commit), seed,
966 	    loose_obj_only, repo,
967 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
968 	if (err)
969 		goto done;
970 
971 	err = got_pack_load_tree(want_meta, idset, idset_exclude,
972 	    got_object_commit_get_tree_id(commit),
973 	    "", got_object_commit_get_committer_time(commit), seed,
974 	    repo, loose_obj_only, ncolored, nfound, ntrees,
975 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
976 done:
977 	got_object_commit_close(commit);
978 	return err;
979 }
980 
981 static const struct got_error *
982 load_tag(int want_meta, struct got_object_idset *idset,
983     struct got_object_idset *idset_exclude,
984     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
985     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
986     got_pack_progress_cb progress_cb, void *progress_arg,
987     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
988 {
989 	const struct got_error *err;
990 	struct got_tag_object *tag = NULL;
991 
992 	if (got_object_idset_contains(idset, id) ||
993 	    got_object_idset_contains(idset_exclude, id))
994 		return NULL;
995 
996 	if (loose_obj_only) {
997 		int is_packed;
998 		err = search_packidx(&is_packed, id, repo);
999 		if (err)
1000 			return err;
1001 		if (is_packed && want_meta)
1002 			return NULL;
1003 	}
1004 
1005 	err = got_object_open_as_tag(&tag, repo, id);
1006 	if (err)
1007 		return err;
1008 
1009 	err = got_pack_add_object(want_meta,
1010 	    want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_TAG,
1011 	    got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1012 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1013 	if (err)
1014 		goto done;
1015 
1016 	switch (got_object_tag_get_object_type(tag)) {
1017 	case GOT_OBJ_TYPE_COMMIT:
1018 		err = load_commit(want_meta, idset, idset_exclude,
1019 		    got_object_tag_get_object_id(tag), repo, seed,
1020 		    loose_obj_only, ncolored, nfound, ntrees,
1021 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1022 		break;
1023 	case GOT_OBJ_TYPE_TREE:
1024 		err = got_pack_load_tree(want_meta, idset, idset_exclude,
1025 		    got_object_tag_get_object_id(tag), "",
1026 		    got_object_tag_get_tagger_time(tag), seed, repo,
1027 		    loose_obj_only, ncolored, nfound, ntrees,
1028 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1029 		break;
1030 	default:
1031 		break;
1032 	}
1033 
1034 done:
1035 	got_object_tag_close(tag);
1036 	return err;
1037 }
1038 
1039 const struct got_error *
1040 got_pack_paint_commit(struct got_object_qid *qid, intptr_t color)
1041 {
1042 	if (color < 0 || color >= COLOR_MAX)
1043 		return got_error(GOT_ERR_RANGE);
1044 
1045 	qid->data = (void *)color;
1046 	return NULL;
1047 }
1048 
1049 const struct got_error *
1050 got_pack_queue_commit_id(struct got_object_id_queue *ids,
1051     struct got_object_id *id, intptr_t color, struct got_repository *repo)
1052 {
1053 	const struct got_error *err;
1054 	struct got_object_qid *qid;
1055 
1056 	err = got_object_qid_alloc(&qid, id);
1057 	if (err)
1058 		return err;
1059 
1060 	STAILQ_INSERT_TAIL(ids, qid, entry);
1061 	return got_pack_paint_commit(qid, color);
1062 }
1063 
1064 struct append_id_arg {
1065 	struct got_object_id **array;
1066 	int idx;
1067 	struct got_object_idset *drop;
1068 	struct got_object_idset *skip;
1069 };
1070 
1071 static const struct got_error *
1072 append_id(struct got_object_id *id, void *data, void *arg)
1073 {
1074 	struct append_id_arg *a = arg;
1075 
1076 	if (got_object_idset_contains(a->skip, id) ||
1077 	    got_object_idset_contains(a->drop, id))
1078 		return NULL;
1079 
1080 	a->array[++a->idx] = got_object_id_dup(id);
1081 	if (a->array[a->idx] == NULL)
1082 		return got_error_from_errno("got_object_id_dup");
1083 
1084 	return NULL;
1085 }
1086 
1087 static const struct got_error *
1088 queue_commit_or_tag_id(struct got_object_id *id, intptr_t color,
1089     struct got_object_id_queue *ids, struct got_repository *repo)
1090 {
1091 	const struct got_error *err;
1092 	struct got_tag_object *tag = NULL;
1093 	int obj_type;
1094 
1095 	err = got_object_get_type(&obj_type, repo, id);
1096 	if (err)
1097 		return err;
1098 
1099 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1100 		err = got_object_open_as_tag(&tag, repo, id);
1101 		if (err)
1102 			return err;
1103 		obj_type = got_object_tag_get_object_type(tag);
1104 		id = got_object_tag_get_object_id(tag);
1105 	}
1106 
1107 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1108 		err = got_pack_queue_commit_id(ids, id, color, repo);
1109 		if (err)
1110 			goto done;
1111 	}
1112 done:
1113 	if (tag)
1114 		got_object_tag_close(tag);
1115 	return err;
1116 }
1117 
1118 const struct got_error *
1119 got_pack_find_pack_for_commit_painting(struct got_packidx **best_packidx,
1120     struct got_object_id_queue *ids, int nids, struct got_repository *repo)
1121 {
1122 	const struct got_error *err = NULL;
1123 	struct got_pathlist_entry *pe;
1124 	const char *best_packidx_path = NULL;
1125 	int nobj_max = 0;
1126 	int ncommits_max = 0;
1127 
1128 	*best_packidx = NULL;
1129 
1130 	/*
1131 	 * Find the largest pack which contains at least some of the
1132 	 * commits we are interested in.
1133 	 */
1134 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1135 		const char *path_packidx = pe->path;
1136 		struct got_packidx *packidx;
1137 		int nobj, idx, ncommits = 0;
1138 		struct got_object_qid *qid;
1139 
1140 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1141 		if (err)
1142 			break;
1143 
1144 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1145 		if (nobj <= nobj_max)
1146 			continue;
1147 
1148 		STAILQ_FOREACH(qid, ids, entry) {
1149 			idx = got_packidx_get_object_idx(packidx, &qid->id);
1150 			if (idx != -1)
1151 				ncommits++;
1152 		}
1153 		if (ncommits > ncommits_max) {
1154 			best_packidx_path = path_packidx;
1155 			nobj_max = nobj;
1156 			ncommits_max = ncommits;
1157 		}
1158 	}
1159 
1160 	if (best_packidx_path && err == NULL) {
1161 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1162 		    repo);
1163 	}
1164 
1165 	return err;
1166 }
1167 
1168 static const struct got_error *
1169 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1170     struct got_object_id **head, int nhead,
1171     struct got_object_id **tail, int ntail,
1172     struct got_repository *repo,
1173     got_pack_progress_cb progress_cb, void *progress_arg,
1174     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1175 {
1176 	const struct got_error *err = NULL;
1177 	struct got_object_id_queue ids;
1178 	struct got_object_idset *keep, *drop, *skip = NULL;
1179 	int i, nkeep;
1180 
1181 	STAILQ_INIT(&ids);
1182 	*res = NULL;
1183 	*nres = 0;
1184 	*ncolored = 0;
1185 
1186 	keep = got_object_idset_alloc();
1187 	if (keep == NULL)
1188 		return got_error_from_errno("got_object_idset_alloc");
1189 
1190 	drop = got_object_idset_alloc();
1191 	if (drop == NULL) {
1192 		err = got_error_from_errno("got_object_idset_alloc");
1193 		goto done;
1194 	}
1195 
1196 	skip = got_object_idset_alloc();
1197 	if (skip == NULL) {
1198 		err = got_error_from_errno("got_object_idset_alloc");
1199 		goto done;
1200 	}
1201 
1202 	for (i = 0; i < nhead; i++) {
1203 		struct got_object_id *id = head[i];
1204 		if (id == NULL)
1205 			continue;
1206 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1207 		if (err)
1208 			goto done;
1209 	}
1210 
1211 	for (i = 0; i < ntail; i++) {
1212 		struct got_object_id *id = tail[i];
1213 		if (id == NULL)
1214 			continue;
1215 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1216 		if (err)
1217 			goto done;
1218 	}
1219 
1220 	err = got_pack_paint_commits(ncolored, &ids, nhead + ntail,
1221 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1222 	    cancel_cb, cancel_arg);
1223 	if (err)
1224 		goto done;
1225 
1226 	nkeep = got_object_idset_num_elements(keep);
1227 	if (nkeep > 0) {
1228 		struct append_id_arg arg;
1229 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1230 		if (arg.array == NULL) {
1231 			err = got_error_from_errno("calloc");
1232 			goto done;
1233 		}
1234 		arg.idx = -1;
1235 		arg.skip = skip;
1236 		arg.drop = drop;
1237 		err = got_object_idset_for_each(keep, append_id, &arg);
1238 		if (err) {
1239 			free(arg.array);
1240 			goto done;
1241 		}
1242 		*res = arg.array;
1243 		*nres = arg.idx + 1;
1244 	}
1245 done:
1246 	got_object_idset_free(keep);
1247 	got_object_idset_free(drop);
1248 	if (skip)
1249 		got_object_idset_free(skip);
1250 	got_object_id_queue_free(&ids);
1251 	return err;
1252 }
1253 
1254 static const struct got_error *
1255 find_pack_for_enumeration(struct got_packidx **best_packidx,
1256     struct got_object_id **ids, int nids, struct got_repository *repo)
1257 {
1258 	const struct got_error *err = NULL;
1259 	struct got_pathlist_entry *pe;
1260 	const char *best_packidx_path = NULL;
1261 	int nobj_max = 0;
1262 	int ncommits_max = 0;
1263 
1264 	*best_packidx = NULL;
1265 
1266 	/*
1267 	 * Find the largest pack which contains at least some of the
1268 	 * commits and tags we are interested in.
1269 	 */
1270 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1271 		const char *path_packidx = pe->path;
1272 		struct got_packidx *packidx;
1273 		int nobj, i, idx, ncommits = 0;
1274 
1275 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1276 		if (err)
1277 			break;
1278 
1279 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1280 		if (nobj <= nobj_max)
1281 			continue;
1282 
1283 		for (i = 0; i < nids; i++) {
1284 			idx = got_packidx_get_object_idx(packidx, ids[i]);
1285 			if (idx != -1)
1286 				ncommits++;
1287 		}
1288 		if (ncommits > ncommits_max) {
1289 			best_packidx_path = path_packidx;
1290 			nobj_max = nobj;
1291 			ncommits_max = ncommits;
1292 		}
1293 	}
1294 
1295 	if (best_packidx_path && err == NULL) {
1296 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1297 		    repo);
1298 	}
1299 
1300 	return err;
1301 }
1302 
1303 static const struct got_error *
1304 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1305     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1306     struct got_object_id **ours, int nours, struct got_repository *repo,
1307     uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1308     void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1309     void *cancel_arg)
1310 {
1311 	const struct got_error *err = NULL;
1312 	struct got_object_id **ids = NULL;
1313 	struct got_packidx *packidx = NULL;
1314 	int i, nobj = 0, obj_type, found_all_objects = 0;
1315 	struct got_object_idset *idset_exclude;
1316 
1317 	idset_exclude = got_object_idset_alloc();
1318 	if (idset_exclude == NULL)
1319 		return got_error_from_errno("got_object_idset_alloc");
1320 
1321 	*ncolored = 0;
1322 	*nfound = 0;
1323 	*ntrees = 0;
1324 
1325 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1326 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1327 	if (err)
1328 		goto done;
1329 
1330 	err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
1331 	if (err)
1332 		goto done;
1333 	if (packidx) {
1334 		err = got_pack_load_packed_object_ids(&found_all_objects,
1335 		    theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
1336 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1337 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1338 		if (err)
1339 			goto done;
1340 	}
1341 
1342 	for (i = 0; i < ntheirs; i++) {
1343 		struct got_object_id *id = theirs[i];
1344 		if (id == NULL)
1345 			continue;
1346 		err = got_object_get_type(&obj_type, repo, id);
1347 		if (err)
1348 			return err;
1349 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1350 			if (!found_all_objects) {
1351 				err = load_commit(0, idset, idset_exclude,
1352 				    id, repo, seed, loose_obj_only,
1353 				    ncolored, nfound, ntrees,
1354 				    progress_cb, progress_arg, rl,
1355 				    cancel_cb, cancel_arg);
1356 				if (err)
1357 					goto done;
1358 			}
1359 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1360 			err = load_tag(0, idset, idset_exclude, id, repo,
1361 			    seed, loose_obj_only, ncolored, nfound, ntrees,
1362 			    progress_cb, progress_arg, rl,
1363 			    cancel_cb, cancel_arg);
1364 			if (err)
1365 				goto done;
1366 		}
1367 	}
1368 
1369 	found_all_objects = 0;
1370 	err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
1371 	if (err)
1372 		goto done;
1373 	if (packidx) {
1374 		err = got_pack_load_packed_object_ids(&found_all_objects, ids,
1375 		    nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
1376 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1377 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1378 		if (err)
1379 			goto done;
1380 	}
1381 
1382 	if (!found_all_objects) {
1383 		for (i = 0; i < nobj; i++) {
1384 			err = load_commit(1, idset, idset_exclude, ids[i],
1385 			    repo, seed, loose_obj_only, ncolored, nfound,
1386 			    ntrees, progress_cb, progress_arg, rl,
1387 			    cancel_cb, cancel_arg);
1388 			if (err)
1389 				goto done;
1390 		}
1391 	}
1392 
1393 	for (i = 0; i < nours; i++) {
1394 		struct got_object_id *id = ours[i];
1395 		struct got_pack_meta *m;
1396 		if (id == NULL)
1397 			continue;
1398 		m = got_object_idset_get(idset, id);
1399 		if (m == NULL) {
1400 			err = got_object_get_type(&obj_type, repo, id);
1401 			if (err)
1402 				goto done;
1403 		} else
1404 			obj_type = m->obj_type;
1405 		if (obj_type != GOT_OBJ_TYPE_TAG)
1406 			continue;
1407 		err = load_tag(1, idset, idset_exclude, id, repo,
1408 		    seed, loose_obj_only, ncolored, nfound, ntrees,
1409 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1410 		if (err)
1411 			goto done;
1412 	}
1413 done:
1414 	for (i = 0; i < nobj; i++) {
1415 		free(ids[i]);
1416 	}
1417 	free(ids);
1418 	got_object_idset_free(idset_exclude);
1419 	return err;
1420 }
1421 
1422 static const struct got_error *
1423 hwrite(int fd, const void *buf, off_t len, SHA1_CTX *ctx)
1424 {
1425 	SHA1Update(ctx, buf, len);
1426 	return got_poll_write_full(fd, buf, len);
1427 }
1428 
1429 static const struct got_error *
1430 hcopy(FILE *fsrc, int fd_dst, off_t len, SHA1_CTX *ctx)
1431 {
1432 	const struct got_error *err;
1433 	unsigned char buf[65536];
1434 	off_t remain = len;
1435 	size_t n;
1436 
1437 	while (remain > 0) {
1438 		size_t copylen = MIN(sizeof(buf), remain);
1439 		n = fread(buf, 1, copylen, fsrc);
1440 		if (n != copylen)
1441 			return got_ferror(fsrc, GOT_ERR_IO);
1442 		SHA1Update(ctx, buf, copylen);
1443 		err = got_poll_write_full(fd_dst, buf, copylen);
1444 		if (err)
1445 			return err;
1446 		remain -= copylen;
1447 	}
1448 
1449 	return NULL;
1450 }
1451 
1452 static const struct got_error *
1453 hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
1454     int fd, off_t len, SHA1_CTX *ctx)
1455 {
1456 	if (src_offset + len > src_size)
1457 		return got_error(GOT_ERR_RANGE);
1458 
1459 	SHA1Update(ctx, src + src_offset, len);
1460 	return got_poll_write_full(fd, src + src_offset, len);
1461 }
1462 
1463 static void
1464 putbe32(char *b, uint32_t n)
1465 {
1466 	b[0] = n >> 24;
1467 	b[1] = n >> 16;
1468 	b[2] = n >> 8;
1469 	b[3] = n >> 0;
1470 }
1471 
1472 static int
1473 write_order_cmp(const void *pa, const void *pb)
1474 {
1475 	struct got_pack_meta *a, *b, *ahd, *bhd;
1476 
1477 	a = *(struct got_pack_meta **)pa;
1478 	b = *(struct got_pack_meta **)pb;
1479 	ahd = (a->head == NULL) ? a : a->head;
1480 	bhd = (b->head == NULL) ? b : b->head;
1481 	if (bhd->mtime < ahd->mtime)
1482 		return -1;
1483 	if (bhd->mtime > ahd->mtime)
1484 		return 1;
1485 	if (bhd < ahd)
1486 		return -1;
1487 	if (bhd > ahd)
1488 		return 1;
1489 	if (a->nchain != b->nchain)
1490 		return a->nchain - b->nchain;
1491 	if (a->mtime < b->mtime)
1492 		return -1;
1493 	if (a->mtime > b->mtime)
1494 		return 1;
1495 	return got_object_id_cmp(&a->id, &b->id);
1496 }
1497 
1498 static int
1499 reuse_write_order_cmp(const void *pa, const void *pb)
1500 {
1501 	struct got_pack_meta *a, *b;
1502 
1503 	a = *(struct got_pack_meta **)pa;
1504 	b = *(struct got_pack_meta **)pb;
1505 
1506 	if (a->reused_delta_offset < b->reused_delta_offset)
1507 		return -1;
1508 	if (a->reused_delta_offset > b->reused_delta_offset)
1509 		return 1;
1510 	return 0;
1511 }
1512 
1513 static const struct got_error *
1514 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1515 {
1516 	size_t i;
1517 
1518 	*hdrlen = 0;
1519 
1520 	hdr[0] = obj_type << 4;
1521 	hdr[0] |= len & 0xf;
1522 	len >>= 4;
1523 	for (i = 1; len != 0; i++){
1524 		if (i >= bufsize)
1525 			return got_error(GOT_ERR_NO_SPACE);
1526 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1527 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1528 		len >>= GOT_DELTA_SIZE_SHIFT;
1529 	}
1530 
1531 	*hdrlen = i;
1532 	return NULL;
1533 }
1534 
1535 static int
1536 packoff(char *hdr, off_t off)
1537 {
1538 	int i, j;
1539 	char rbuf[8];
1540 
1541 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1542 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1543 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1544 		    GOT_DELTA_SIZE_MORE;
1545 	}
1546 
1547 	j = 0;
1548 	while (i > 0)
1549 		hdr[j++] = rbuf[--i];
1550 	return j;
1551 }
1552 
1553 static const struct got_error *
1554 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, int packfd,
1555     struct got_pack_meta *m)
1556 {
1557 	const struct got_error *err;
1558 	char buf[32];
1559 	int nh;
1560 
1561 	if (m->prev->off != 0) {
1562 		err = packhdr(&nh, buf, sizeof(buf),
1563 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1564 		if (err)
1565 			return err;
1566 		nh += packoff(buf + nh, m->off - m->prev->off);
1567 		err = hwrite(packfd, buf, nh, ctx);
1568 		if (err)
1569 			return err;
1570 		*packfile_size += nh;
1571 	} else {
1572 		err = packhdr(&nh, buf, sizeof(buf),
1573 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1574 		if (err)
1575 			return err;
1576 		err = hwrite(packfd, buf, nh, ctx);
1577 		if (err)
1578 			return err;
1579 		*packfile_size += nh;
1580 		err = hwrite(packfd, m->prev->id.hash,
1581 		    sizeof(m->prev->id.hash), ctx);
1582 		if (err)
1583 			return err;
1584 		*packfile_size += sizeof(m->prev->id.hash);
1585 	}
1586 
1587 	return NULL;
1588 }
1589 
1590 static const struct got_error *
1591 write_packed_object(off_t *packfile_size, int packfd,
1592     FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
1593     struct got_pack_meta *m, int *outfd, SHA1_CTX *ctx,
1594     struct got_repository *repo)
1595 {
1596 	const struct got_error *err = NULL;
1597 	struct got_deflate_checksum csum;
1598 	char buf[32];
1599 	int nh;
1600 	struct got_raw_object *raw = NULL;
1601 	off_t outlen, delta_offset;
1602 
1603 	csum.output_hash = ctx;
1604 	csum.output_crc = NULL;
1605 
1606 	if (m->reused_delta_offset)
1607 		delta_offset = m->reused_delta_offset;
1608 	else
1609 		delta_offset = m->delta_offset;
1610 
1611 	m->off = *packfile_size;
1612 	if (m->delta_len == 0) {
1613 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1614 		if (err)
1615 			goto done;
1616 		err = packhdr(&nh, buf, sizeof(buf),
1617 		    m->obj_type, raw->size);
1618 		if (err)
1619 			goto done;
1620 		err = hwrite(packfd, buf, nh, ctx);
1621 		if (err)
1622 			goto done;
1623 		*packfile_size += nh;
1624 		if (raw->f == NULL) {
1625 			err = got_deflate_to_fd_mmap(&outlen,
1626 			    raw->data + raw->hdrlen, 0, raw->size,
1627 			    packfd, &csum);
1628 			if (err)
1629 				goto done;
1630 		} else {
1631 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1632 			    == -1) {
1633 				err = got_error_from_errno("fseeko");
1634 				goto done;
1635 			}
1636 			err = got_deflate_to_fd(&outlen, raw->f,
1637 			    raw->size, packfd, &csum);
1638 			if (err)
1639 				goto done;
1640 		}
1641 		*packfile_size += outlen;
1642 		got_object_raw_close(raw);
1643 		raw = NULL;
1644 	} else if (m->delta_buf) {
1645 		err = deltahdr(packfile_size, ctx, packfd, m);
1646 		if (err)
1647 			goto done;
1648 		err = hwrite(packfd, m->delta_buf,
1649 		    m->delta_compressed_len, ctx);
1650 		if (err)
1651 			goto done;
1652 		*packfile_size += m->delta_compressed_len;
1653 		free(m->delta_buf);
1654 		m->delta_buf = NULL;
1655 	} else if (delta_cache_map) {
1656 		err = deltahdr(packfile_size, ctx, packfd, m);
1657 		if (err)
1658 			goto done;
1659 		err = hcopy_mmap(delta_cache_map, delta_offset,
1660 		    delta_cache_size, packfd, m->delta_compressed_len,
1661 		    ctx);
1662 		if (err)
1663 			goto done;
1664 		*packfile_size += m->delta_compressed_len;
1665 	} else {
1666 		if (fseeko(delta_cache, delta_offset, SEEK_SET) == -1) {
1667 			err = got_error_from_errno("fseeko");
1668 			goto done;
1669 		}
1670 		err = deltahdr(packfile_size, ctx, packfd, m);
1671 		if (err)
1672 			goto done;
1673 		err = hcopy(delta_cache, packfd,
1674 		    m->delta_compressed_len, ctx);
1675 		if (err)
1676 			goto done;
1677 		*packfile_size += m->delta_compressed_len;
1678 	}
1679 done:
1680 	if (raw)
1681 		got_object_raw_close(raw);
1682 	return err;
1683 }
1684 
1685 static const struct got_error *
1686 genpack(uint8_t *pack_hash, int packfd, struct got_pack *reuse_pack,
1687     FILE *delta_cache, struct got_pack_meta **deltify, int ndeltify,
1688     struct got_pack_meta **reuse, int nreuse,
1689     int ncolored, int nfound, int ntrees, int nours,
1690     struct got_repository *repo,
1691     got_pack_progress_cb progress_cb, void *progress_arg,
1692     struct got_ratelimit *rl,
1693     got_cancel_cb cancel_cb, void *cancel_arg)
1694 {
1695 	const struct got_error *err = NULL;
1696 	int i;
1697 	SHA1_CTX ctx;
1698 	struct got_pack_meta *m;
1699 	char buf[32];
1700 	off_t packfile_size = 0;
1701 	int outfd = -1;
1702 	int delta_cache_fd = -1;
1703 	uint8_t *delta_cache_map = NULL;
1704 	size_t delta_cache_size = 0;
1705 	FILE *packfile = NULL;
1706 
1707 	SHA1Init(&ctx);
1708 
1709 #ifndef GOT_PACK_NO_MMAP
1710 	delta_cache_fd = dup(fileno(delta_cache));
1711 	if (delta_cache_fd != -1) {
1712 		struct stat sb;
1713 		if (fstat(delta_cache_fd, &sb) == -1) {
1714 			err = got_error_from_errno("fstat");
1715 			goto done;
1716 		}
1717 		if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
1718 			delta_cache_map = mmap(NULL, sb.st_size,
1719 			    PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
1720 			if (delta_cache_map == MAP_FAILED) {
1721 				if (errno != ENOMEM) {
1722 					err = got_error_from_errno("mmap");
1723 					goto done;
1724 				}
1725 				delta_cache_map = NULL; /* fallback on stdio */
1726 			} else
1727 				delta_cache_size = (size_t)sb.st_size;
1728 		}
1729 	}
1730 #endif
1731 	err = hwrite(packfd, "PACK", 4, &ctx);
1732 	if (err)
1733 		goto done;
1734 	putbe32(buf, GOT_PACKFILE_VERSION);
1735 	err = hwrite(packfd, buf, 4, &ctx);
1736 	if (err)
1737 		goto done;
1738 	putbe32(buf, ndeltify + nreuse);
1739 	err = hwrite(packfd, buf, 4, &ctx);
1740 	if (err)
1741 		goto done;
1742 
1743 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1744 	    write_order_cmp);
1745 	for (i = 0; i < ndeltify; i++) {
1746 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
1747 		    ncolored, nfound, ntrees, packfile_size, nours,
1748 		    ndeltify + nreuse, ndeltify + nreuse, i);
1749 		if (err)
1750 			goto done;
1751 		m = deltify[i];
1752 		err = write_packed_object(&packfile_size, packfd,
1753 		    delta_cache, delta_cache_map, delta_cache_size,
1754 		    m, &outfd, &ctx, repo);
1755 		if (err)
1756 			goto done;
1757 	}
1758 
1759 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1760 	    reuse_write_order_cmp);
1761 	if (nreuse > 0 && reuse_pack->map == NULL) {
1762 		int fd = dup(reuse_pack->fd);
1763 		if (fd == -1) {
1764 			err = got_error_from_errno("dup");
1765 			goto done;
1766 		}
1767 		packfile = fdopen(fd, "r");
1768 		if (packfile == NULL) {
1769 			err = got_error_from_errno("fdopen");
1770 			close(fd);
1771 			goto done;
1772 		}
1773 	}
1774 	for (i = 0; i < nreuse; i++) {
1775 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
1776 		    ncolored, nfound, ntrees, packfile_size, nours,
1777 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1778 		if (err)
1779 			goto done;
1780 		m = reuse[i];
1781 		err = write_packed_object(&packfile_size, packfd,
1782 		    packfile, reuse_pack->map, reuse_pack->filesize,
1783 		    m, &outfd, &ctx, repo);
1784 		if (err)
1785 			goto done;
1786 	}
1787 
1788 	SHA1Final(pack_hash, &ctx);
1789 	err = got_poll_write_full(packfd, pack_hash, SHA1_DIGEST_LENGTH);
1790 	if (err)
1791 		goto done;
1792 	packfile_size += SHA1_DIGEST_LENGTH;
1793 	packfile_size += sizeof(struct got_packfile_hdr);
1794 	if (progress_cb) {
1795 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1796 		    packfile_size, nours, ndeltify + nreuse,
1797 		    ndeltify + nreuse, ndeltify + nreuse);
1798 		if (err)
1799 			goto done;
1800 	}
1801 done:
1802 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1803 		err = got_error_from_errno("close");
1804 	if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
1805 		err = got_error_from_errno("munmap");
1806 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1807 		err = got_error_from_errno("close");
1808 	if (packfile && fclose(packfile) == EOF && err == NULL)
1809 		err = got_error_from_errno("fclose");
1810 	return err;
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 	if (m->reused_delta_offset != 0)
1820 		return NULL;
1821 
1822 	return got_pack_add_meta(m, v);
1823 }
1824 
1825 const struct got_error *
1826 got_pack_create(uint8_t *packhash, int packfd, FILE *delta_cache,
1827     struct got_object_id **theirs, int ntheirs,
1828     struct got_object_id **ours, int nours,
1829     struct got_repository *repo, int loose_obj_only, int allow_empty,
1830     got_pack_progress_cb progress_cb, void *progress_arg,
1831     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1832 {
1833 	const struct got_error *err;
1834 	struct got_object_idset *idset;
1835 	struct got_packidx *reuse_packidx = NULL;
1836 	struct got_pack *reuse_pack = NULL;
1837 	struct got_pack_metavec deltify, reuse;
1838 	int ncolored = 0, nfound = 0, ntrees = 0;
1839 	size_t ndeltify;
1840 	uint32_t seed;
1841 
1842 	seed = arc4random();
1843 
1844 	memset(&deltify, 0, sizeof(deltify));
1845 	memset(&reuse, 0, sizeof(reuse));
1846 
1847 	idset = got_object_idset_alloc();
1848 	if (idset == NULL)
1849 		return got_error_from_errno("got_object_idset_alloc");
1850 
1851 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1852 	    ntheirs, ours, nours, repo, seed, loose_obj_only,
1853 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
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 	reuse.metasz = 64;
1870 	reuse.meta = calloc(reuse.metasz,
1871 	    sizeof(struct got_pack_meta *));
1872 	if (reuse.meta == NULL) {
1873 		err = got_error_from_errno("calloc");
1874 		goto done;
1875 	}
1876 
1877 	err = got_pack_search_deltas(&reuse_packidx, &reuse_pack,
1878 	    &reuse, idset, ncolored, nfound, ntrees, nours,
1879 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1880 	if (err)
1881 		goto done;
1882 
1883 	if (reuse_packidx && reuse_pack) {
1884 		err = got_repo_pin_pack(repo, reuse_packidx, reuse_pack);
1885 		if (err)
1886 			goto done;
1887 	}
1888 
1889 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1890 		err = got_error_from_errno("fseeko");
1891 		goto done;
1892 	}
1893 
1894 	ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
1895 	if (ndeltify > 0) {
1896 		deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
1897 		if (deltify.meta == NULL) {
1898 			err = got_error_from_errno("calloc");
1899 			goto done;
1900 		}
1901 		deltify.metasz = ndeltify;
1902 
1903 		err = got_object_idset_for_each(idset, add_meta_idset_cb,
1904 		    &deltify);
1905 		if (err)
1906 			goto done;
1907 		if (deltify.nmeta > 0) {
1908 			err = pick_deltas(deltify.meta, deltify.nmeta,
1909 			    ncolored, nfound, ntrees, nours, reuse.nmeta,
1910 			    delta_cache, repo, progress_cb, progress_arg, rl,
1911 			    cancel_cb, cancel_arg);
1912 			if (err)
1913 				goto done;
1914 		}
1915 	}
1916 
1917 	if (fflush(delta_cache) == EOF) {
1918 		err = got_error_from_errno("fflush");
1919 		goto done;
1920 	}
1921 
1922 	if (progress_cb) {
1923 		/*
1924 		 * Report a 1-byte packfile write to indicate we are about
1925 		 * to start sending packfile data. gotd(8) needs this.
1926 		 */
1927 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1928 		    1 /* packfile_size */, nours,
1929 		    got_object_idset_num_elements(idset),
1930 		    deltify.nmeta + reuse.nmeta, 0);
1931 		if (err)
1932 			goto done;
1933 	}
1934 
1935 	err = genpack(packhash, packfd, reuse_pack, delta_cache, deltify.meta,
1936 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1937 	    nours, repo, progress_cb, progress_arg, rl,
1938 	    cancel_cb, cancel_arg);
1939 	if (err)
1940 		goto done;
1941 done:
1942 	free_nmeta(deltify.meta, deltify.nmeta);
1943 	free_nmeta(reuse.meta, reuse.nmeta);
1944 	got_object_idset_free(idset);
1945 	got_repo_unpin_pack(repo);
1946 	return err;
1947 }