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_hash.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_idset.h"
54 #include "got_lib_object_cache.h"
55 #include "got_lib_deflate.h"
56 #include "got_lib_ratelimit.h"
57 #include "got_lib_pack.h"
58 #include "got_lib_pack_create.h"
59 #include "got_lib_repository.h"
60 #include "got_lib_inflate.h"
61 #include "got_lib_poll.h"
62 
63 #include "murmurhash2.h"
64 
65 #ifndef MIN
66 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
67 #endif
68 
69 #ifndef MAX
70 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
71 #endif
72 
73 #ifndef nitems
74 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
75 #endif
76 
77 static const struct got_error *
78 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
79     const char *path, int obj_type, time_t mtime, uint32_t seed)
80 {
81 	struct got_pack_meta *m;
82 
83 	*new = NULL;
84 
85 	m = calloc(1, sizeof(*m));
86 	if (m == NULL)
87 		return got_error_from_errno("calloc");
88 
89 	memcpy(&m->id, id, sizeof(m->id));
90 
91 	m->path_hash = murmurhash2(path, strlen(path), seed);
92 	m->obj_type = obj_type;
93 	m->mtime = mtime;
94 	*new = m;
95 	return NULL;
96 }
97 
98 static void
99 clear_meta(struct got_pack_meta *meta)
100 {
101 	if (meta == NULL)
102 		return;
103 	meta->path_hash = 0;
104 	free(meta->delta_buf);
105 	meta->delta_buf = NULL;
106 	free(meta->base_obj_id);
107 	meta->base_obj_id = NULL;
108 	meta->reused_delta_offset = 0;
109 }
110 
111 static void
112 free_nmeta(struct got_pack_meta **meta, int nmeta)
113 {
114 	int i;
115 
116 	for (i = 0; i < nmeta; i++)
117 		clear_meta(meta[i]);
118 	free(meta);
119 }
120 
121 static int
122 delta_order_cmp(const void *pa, const void *pb)
123 {
124 	struct got_pack_meta *a, *b;
125 
126 	a = *(struct got_pack_meta **)pa;
127 	b = *(struct got_pack_meta **)pb;
128 
129 	if (a->obj_type != b->obj_type)
130 		return a->obj_type - b->obj_type;
131 	if (a->path_hash < b->path_hash)
132 		return -1;
133 	if (a->path_hash > b->path_hash)
134 		return 1;
135 	if (a->mtime < b->mtime)
136 		return -1;
137 	if (a->mtime > b->mtime)
138 		return 1;
139 	return got_object_id_cmp(&a->id, &b->id);
140 }
141 
142 static off_t
143 delta_size(struct got_delta_instruction *deltas, int ndeltas)
144 {
145 	int i;
146 	off_t size = 32;
147 	for (i = 0; i < ndeltas; i++) {
148 		if (deltas[i].copy)
149 			size += GOT_DELTA_SIZE_SHIFT;
150 		else
151 			size += deltas[i].len + 1;
152 	}
153 	return size;
154 }
155 
156 static const struct got_error *
157 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
158 {
159 	char *n;
160 
161 	if (*len + nseg >= *sz) {
162 		while (*len + nseg >= *sz)
163 			*sz += *sz / 2;
164 		n = realloc(*p, *sz);
165 		if (n == NULL)
166 			return got_error_from_errno("realloc");
167 		*p = n;
168 	}
169 	memcpy(*p + *len, seg, nseg);
170 	*len += nseg;
171 	return NULL;
172 }
173 
174 static const struct got_error *
175 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
176     struct got_delta_instruction *deltas, int ndeltas,
177     off_t delta_size, off_t base_size)
178 {
179 	const struct got_error *err;
180 	unsigned char buf[16], *bp;
181 	int i, j;
182 	size_t len = 0, compressed_len;
183 	off_t bufsize = delta_size;
184 	off_t n;
185 	struct got_delta_instruction *d;
186 	uint8_t *delta_buf;
187 
188 	delta_buf = malloc(bufsize);
189 	if (delta_buf == NULL)
190 		return got_error_from_errno("malloc");
191 
192 	/* base object size */
193 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
194 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
195 	for (i = 1; n > 0; i++) {
196 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
197 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
198 		n >>= GOT_DELTA_SIZE_SHIFT;
199 	}
200 	err = append(&delta_buf, &len, &bufsize, buf, i);
201 	if (err)
202 		goto done;
203 
204 	/* target object size */
205 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
206 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
207 	for (i = 1; n > 0; i++) {
208 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
209 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
210 		n >>= GOT_DELTA_SIZE_SHIFT;
211 	}
212 	err = append(&delta_buf, &len, &bufsize, buf, i);
213 	if (err)
214 		goto done;
215 
216 	for (j = 0; j < ndeltas; j++) {
217 		d = &deltas[j];
218 		if (d->copy) {
219 			n = d->offset;
220 			bp = &buf[1];
221 			buf[0] = GOT_DELTA_BASE_COPY;
222 			for (i = 0; i < 4; i++) {
223 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
224 				buf[0] |= 1 << i;
225 				*bp++ = n & 0xff;
226 				n >>= 8;
227 				if (n == 0)
228 					break;
229 			}
230 
231 			n = d->len;
232 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
233 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
234 				for (i = 0; i < 3 && n > 0; i++) {
235 					buf[0] |= 1 << (i + 4);
236 					*bp++ = n & 0xff;
237 					n >>= 8;
238 				}
239 			}
240 			err = append(&delta_buf, &len, &bufsize,
241 			    buf, bp - buf);
242 			if (err)
243 				goto done;
244 		} else if (o->f == NULL) {
245 			n = 0;
246 			while (n != d->len) {
247 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
248 				err = append(&delta_buf, &len, &bufsize,
249 				    buf, 1);
250 				if (err)
251 					goto done;
252 				err = append(&delta_buf, &len, &bufsize,
253 				    o->data + o->hdrlen + d->offset + n,
254 				    buf[0]);
255 				if (err)
256 					goto done;
257 				n += buf[0];
258 			}
259 		} else {
260 			char content[128];
261 			size_t r;
262 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
263 				err = got_error_from_errno("fseeko");
264 				goto done;
265 			}
266 			n = 0;
267 			while (n != d->len) {
268 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
269 				err = append(&delta_buf, &len, &bufsize,
270 				    buf, 1);
271 				if (err)
272 					goto done;
273 				r = fread(content, 1, buf[0], o->f);
274 				if (r != buf[0]) {
275 					err = got_ferror(o->f, GOT_ERR_IO);
276 					goto done;
277 				}
278 				err = append(&delta_buf, &len, &bufsize,
279 				    content, buf[0]);
280 				if (err)
281 					goto done;
282 				n += buf[0];
283 			}
284 		}
285 	}
286 
287 	err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
288 	    NULL, NULL, delta_buf, 0, len);
289 	if (err)
290 		goto done;
291 
292 	m->delta_len = len;
293 	m->delta_compressed_len = compressed_len;
294 done:
295 	free(delta_buf);
296 	return err;
297 }
298 
299 static const struct got_error *
300 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
301     struct got_delta_instruction *deltas, int ndeltas,
302     off_t base_size, FILE *f)
303 {
304 	const struct got_error *err;
305 	unsigned char buf[16], *bp;
306 	int i, j;
307 	off_t n;
308 	struct got_deflate_buf zb;
309 	struct got_delta_instruction *d;
310 	off_t delta_len = 0, compressed_len = 0;
311 
312 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
313 	if (err)
314 		return err;
315 
316 	/* base object size */
317 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
318 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
319 	for (i = 1; n > 0; i++) {
320 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
321 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
322 		n >>= GOT_DELTA_SIZE_SHIFT;
323 	}
324 
325 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
326 	    buf, 0, i, f, NULL);
327 	if (err)
328 		goto done;
329 	delta_len += i;
330 
331 	/* target object size */
332 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
333 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
334 	for (i = 1; n > 0; i++) {
335 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
336 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
337 		n >>= GOT_DELTA_SIZE_SHIFT;
338 	}
339 
340 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
341 	    buf, 0, i, f, NULL);
342 	if (err)
343 		goto done;
344 	delta_len += i;
345 
346 	for (j = 0; j < ndeltas; j++) {
347 		d = &deltas[j];
348 		if (d->copy) {
349 			n = d->offset;
350 			bp = &buf[1];
351 			buf[0] = GOT_DELTA_BASE_COPY;
352 			for (i = 0; i < 4; i++) {
353 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
354 				buf[0] |= 1 << i;
355 				*bp++ = n & 0xff;
356 				n >>= 8;
357 				if (n == 0)
358 					break;
359 			}
360 			n = d->len;
361 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
362 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
363 				for (i = 0; i < 3 && n > 0; i++) {
364 					buf[0] |= 1 << (i + 4);
365 					*bp++ = n & 0xff;
366 					n >>= 8;
367 				}
368 			}
369 			err = got_deflate_append_to_file_mmap(&zb,
370 			    &compressed_len, buf, 0, bp - buf, f, NULL);
371 			if (err)
372 				goto done;
373 			delta_len += (bp - buf);
374 		} else if (o->f == NULL) {
375 			n = 0;
376 			while (n != d->len) {
377 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
378 				err = got_deflate_append_to_file_mmap(&zb,
379 				    &compressed_len, buf, 0, 1, f, NULL);
380 				if (err)
381 					goto done;
382 				delta_len++;
383 				err = got_deflate_append_to_file_mmap(&zb,
384 				    &compressed_len,
385 				    o->data + o->hdrlen + d->offset + n, 0,
386 				    buf[0], f, NULL);
387 				if (err)
388 					goto done;
389 				delta_len += buf[0];
390 				n += buf[0];
391 			}
392 		} else {
393 			char content[128];
394 			size_t r;
395 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
396 				err = got_error_from_errno("fseeko");
397 				goto done;
398 			}
399 			n = 0;
400 			while (n != d->len) {
401 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
402 				err = got_deflate_append_to_file_mmap(&zb,
403 				    &compressed_len, buf, 0, 1, f, NULL);
404 				if (err)
405 					goto done;
406 				delta_len++;
407 				r = fread(content, 1, buf[0], o->f);
408 				if (r != buf[0]) {
409 					err = got_ferror(o->f, GOT_ERR_IO);
410 					goto done;
411 				}
412 				err = got_deflate_append_to_file_mmap(&zb,
413 				    &compressed_len, content, 0, buf[0], f,
414 				    NULL);
415 				if (err)
416 					goto done;
417 				delta_len += buf[0];
418 				n += buf[0];
419 			}
420 		}
421 	}
422 
423 	err = got_deflate_flush(&zb, f, NULL, &compressed_len);
424 	if (err)
425 		goto done;
426 
427 	/* sanity check */
428 	if (compressed_len != ftello(f) - m->delta_offset) {
429 		err = got_error(GOT_ERR_COMPRESSION);
430 		goto done;
431 	}
432 
433 	m->delta_len = delta_len;
434 	m->delta_compressed_len = compressed_len;
435 done:
436 	got_deflate_end(&zb);
437 	return err;
438 }
439 
440 const struct got_error *
441 got_pack_report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
442     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
443     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
444     int nobj_written)
445 {
446 	const struct got_error *err;
447 	int elapsed;
448 
449 	if (progress_cb == NULL)
450 		return NULL;
451 
452 	err = got_ratelimit_check(&elapsed, rl);
453 	if (err || !elapsed)
454 		return err;
455 
456 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
457 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
458 }
459 
460 const struct got_error *
461 got_pack_add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
462 {
463 	if (v->nmeta == v->metasz){
464 		size_t newsize = 2 * v->metasz;
465 		struct got_pack_meta **new;
466 		new = reallocarray(v->meta, newsize, sizeof(*new));
467 		if (new == NULL)
468 			return got_error_from_errno("reallocarray");
469 		v->meta = new;
470 		v->metasz = newsize;
471 	}
472 
473 	v->meta[v->nmeta++] = m;
474 	return NULL;
475 }
476 
477 const struct got_error *
478 got_pack_find_pack_for_reuse(struct got_packidx **best_packidx,
479     struct got_repository *repo)
480 {
481 	const struct got_error *err = NULL;
482 	struct got_pathlist_entry *pe;
483 	const char *best_packidx_path = NULL;
484 	int nobj_max = 0;
485 
486 	*best_packidx = NULL;
487 
488 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
489 		const char *path_packidx = pe->path;
490 		struct got_packidx *packidx;
491 		int nobj;
492 
493 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
494 		if (err)
495 			break;
496 
497 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
498 		if (nobj > nobj_max) {
499 			best_packidx_path = path_packidx;
500 			nobj_max = nobj;
501 		}
502 	}
503 
504 	if (best_packidx_path) {
505 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
506 		    repo);
507 	}
508 
509 	return err;
510 }
511 
512 const struct got_error *
513 got_pack_cache_pack_for_packidx(struct got_pack **pack,
514     struct got_packidx *packidx, struct got_repository *repo)
515 {
516 	const struct got_error *err;
517 	char *path_packfile = NULL;
518 
519 	err = got_packidx_get_packfile_path(&path_packfile,
520 	    packidx->path_packidx);
521 	if (err)
522 		return err;
523 
524 	*pack = got_repo_get_cached_pack(repo, path_packfile);
525 	if (*pack == NULL) {
526 		err = got_repo_cache_pack(pack, repo, path_packfile, packidx);
527 		if (err)
528 			goto done;
529 	}
530 done:
531 	free(path_packfile);
532 	return err;
533 }
534 
535 static const struct got_error *
536 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
537     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
538     struct got_repository *repo,
539     got_pack_progress_cb progress_cb, void *progress_arg,
540     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
541 {
542 	const struct got_error *err = NULL;
543 	struct got_pack_meta *m = NULL, *base = NULL;
544 	struct got_raw_object *raw = NULL, *base_raw = NULL;
545 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
546 	int i, j, ndeltas, best_ndeltas;
547 	off_t size, best_size;
548 	const int max_base_candidates = 3;
549 	size_t delta_memsize = 0;
550 	const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
551 	int outfd = -1;
552 	uint32_t delta_seed;
553 
554 	delta_seed = arc4random();
555 
556 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
557 	for (i = 0; i < nmeta; i++) {
558 		if (cancel_cb) {
559 			err = (*cancel_cb)(cancel_arg);
560 			if (err)
561 				break;
562 		}
563 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
564 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
565 		    nreused + i, 0);
566 		if (err)
567 			goto done;
568 		m = meta[i];
569 
570 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
571 		    m->obj_type == GOT_OBJ_TYPE_TAG)
572 			continue;
573 
574 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
575 		if (err)
576 			goto done;
577 		m->size = raw->size;
578 
579 		if (raw->f == NULL) {
580 			err = got_deltify_init_mem(&m->dtab, raw->data,
581 			    raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
582 		} else {
583 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
584 			    raw->size + raw->hdrlen, delta_seed);
585 		}
586 		if (err)
587 			goto done;
588 
589 		if (i > max_base_candidates) {
590 			struct got_pack_meta *n = NULL;
591 			n = meta[i - (max_base_candidates + 1)];
592 			got_deltify_free(n->dtab);
593 			n->dtab = NULL;
594 		}
595 
596 		best_size = raw->size;
597 		best_ndeltas = 0;
598 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
599 			if (cancel_cb) {
600 				err = (*cancel_cb)(cancel_arg);
601 				if (err)
602 					goto done;
603 			}
604 			base = meta[j];
605 			/* long chains make unpacking slow, avoid such bases */
606 			if (base->nchain >= 128 ||
607 			    base->obj_type != m->obj_type)
608 				continue;
609 
610 			err = got_object_raw_open(&base_raw, &outfd, repo,
611 			    &base->id);
612 			if (err)
613 				goto done;
614 
615 			if (raw->f == NULL && base_raw->f == NULL) {
616 				err = got_deltify_mem_mem(&deltas, &ndeltas,
617 				    raw->data, raw->hdrlen,
618 				    raw->size + raw->hdrlen, delta_seed,
619 				    base->dtab, base_raw->data,
620 				    base_raw->hdrlen,
621 				    base_raw->size + base_raw->hdrlen);
622 			} else if (raw->f == NULL) {
623 				err = got_deltify_mem_file(&deltas, &ndeltas,
624 				    raw->data, raw->hdrlen,
625 				    raw->size + raw->hdrlen, delta_seed,
626 				    base->dtab, base_raw->f,
627 				    base_raw->hdrlen,
628 				    base_raw->size + base_raw->hdrlen);
629 			} else if (base_raw->f == NULL) {
630 				err = got_deltify_file_mem(&deltas, &ndeltas,
631 				    raw->f, raw->hdrlen,
632 				    raw->size + raw->hdrlen, delta_seed,
633 				    base->dtab, base_raw->data,
634 				    base_raw->hdrlen,
635 				    base_raw->size + base_raw->hdrlen);
636 			} else {
637 				err = got_deltify(&deltas, &ndeltas,
638 				    raw->f, raw->hdrlen,
639 				    raw->size + raw->hdrlen, delta_seed,
640 				    base->dtab, base_raw->f, base_raw->hdrlen,
641 				    base_raw->size + base_raw->hdrlen);
642 			}
643 			got_object_raw_close(base_raw);
644 			base_raw = NULL;
645 			if (err)
646 				goto done;
647 
648 			size = delta_size(deltas, ndeltas);
649 			if (size + 32 < best_size){
650 				/*
651 				 * if we already picked a best delta,
652 				 * replace it.
653 				 */
654 				best_size = size;
655 				free(best_deltas);
656 				best_deltas = deltas;
657 				best_ndeltas = ndeltas;
658 				deltas = NULL;
659 				m->nchain = base->nchain + 1;
660 				m->prev = base;
661 				m->head = base->head;
662 				if (m->head == NULL)
663 					m->head = base;
664 			} else {
665 				free(deltas);
666 				deltas = NULL;
667 				ndeltas = 0;
668 			}
669 		}
670 
671 		if (best_ndeltas > 0) {
672 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
673 			    delta_memsize + best_size <= max_delta_memsize) {
674 				delta_memsize += best_size;
675 				err = encode_delta_in_mem(m, raw, best_deltas,
676 				    best_ndeltas, best_size, m->prev->size);
677 			} else {
678 				m->delta_offset = ftello(delta_cache);
679 				err = encode_delta(m, raw, best_deltas,
680 				    best_ndeltas, m->prev->size, delta_cache);
681 			}
682 			free(best_deltas);
683 			best_deltas = NULL;
684 			best_ndeltas = 0;
685 			if (err)
686 				goto done;
687 		}
688 
689 		got_object_raw_close(raw);
690 		raw = NULL;
691 	}
692 done:
693 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
694 		got_deltify_free(meta[i]->dtab);
695 		meta[i]->dtab = NULL;
696 	}
697 	if (raw)
698 		got_object_raw_close(raw);
699 	if (base_raw)
700 		got_object_raw_close(base_raw);
701 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
702 		err = got_error_from_errno("close");
703 	free(deltas);
704 	free(best_deltas);
705 	return err;
706 }
707 
708 static const struct got_error *
709 search_packidx(int *found, struct got_object_id *id,
710     struct got_repository *repo)
711 {
712 	const struct got_error *err = NULL;
713 	struct got_packidx *packidx = NULL;
714 	int idx;
715 
716 	*found = 0;
717 
718 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
719 	if (err == NULL)
720 		*found = 1; /* object is already packed */
721 	else if (err->code == GOT_ERR_NO_OBJ)
722 		err = NULL;
723 	return err;
724 }
725 
726 const struct got_error *
727 got_pack_add_object(int want_meta, struct got_object_idset *idset,
728     struct got_object_id *id, const char *path, int obj_type,
729     time_t mtime, uint32_t seed, int loose_obj_only,
730     struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
731     got_pack_progress_cb progress_cb, void *progress_arg,
732     struct got_ratelimit *rl)
733 {
734 	const struct got_error *err;
735 	struct got_pack_meta *m = NULL;
736 
737 	if (loose_obj_only) {
738 		int is_packed;
739 		err = search_packidx(&is_packed, id, repo);
740 		if (err)
741 			return err;
742 		if (is_packed && want_meta)
743 			return NULL;
744 	}
745 
746 	if (want_meta) {
747 		err = alloc_meta(&m, id, path, obj_type, mtime, seed);
748 		if (err)
749 			return err;
750 
751 		(*nfound)++;
752 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
753 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
754 		if (err) {
755 			clear_meta(m);
756 			free(m);
757 			return err;
758 		}
759 	}
760 
761 	err = got_object_idset_add(idset, id, m);
762 	if (err) {
763 		clear_meta(m);
764 		free(m);
765 	}
766 	return err;
767 }
768 
769 const struct got_error *
770 got_pack_load_tree_entries(struct got_object_id_queue *ids, int want_meta,
771     struct got_object_idset *idset, struct got_object_idset *idset_exclude,
772     struct got_tree_object *tree,
773     const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
774     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
775     got_pack_progress_cb progress_cb, void *progress_arg,
776     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
777 {
778 	const struct got_error *err;
779 	char *p = NULL;
780 	int i;
781 
782 	(*ntrees)++;
783 	err = got_pack_report_progress(progress_cb, progress_arg, rl,
784 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
785 	if (err)
786 		return err;
787 
788 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
789 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
790 		struct got_object_id *id = got_tree_entry_get_id(e);
791 		mode_t mode = got_tree_entry_get_mode(e);
792 
793 		if (cancel_cb) {
794 			err = (*cancel_cb)(cancel_arg);
795 			if (err)
796 				break;
797 		}
798 
799 		if (got_object_tree_entry_is_submodule(e) ||
800 		    got_object_idset_contains(idset, id) ||
801 		    got_object_idset_contains(idset_exclude, id))
802 			continue;
803 
804 		/*
805 		 * If got-read-pack is crawling trees for us then
806 		 * we are only here to collect blob IDs.
807 		 */
808 		if (ids == NULL && S_ISDIR(mode))
809 			continue;
810 
811 		if (asprintf(&p, "%s%s%s", dpath,
812 		    got_path_is_root_dir(dpath) ? "" : "/",
813 		    got_tree_entry_get_name(e)) == -1) {
814 			err = got_error_from_errno("asprintf");
815 			break;
816 		}
817 
818 		if (S_ISDIR(mode)) {
819 			struct got_object_qid *qid;
820 			err = got_object_qid_alloc(&qid, id);
821 			if (err)
822 				break;
823 			qid->data = p;
824 			p = NULL;
825 			STAILQ_INSERT_TAIL(ids, qid, entry);
826 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
827 			err = got_pack_add_object(want_meta,
828 			    want_meta ? idset : idset_exclude, id, p,
829 			    GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
830 			    repo, ncolored, nfound, ntrees,
831 			    progress_cb, progress_arg, rl);
832 			if (err)
833 				break;
834 			free(p);
835 			p = NULL;
836 		} else {
837 			free(p);
838 			p = NULL;
839 		}
840 	}
841 
842 	free(p);
843 	return err;
844 }
845 
846 const struct got_error *
847 got_pack_load_tree(int want_meta, struct got_object_idset *idset,
848     struct got_object_idset *idset_exclude,
849     struct got_object_id *tree_id, const char *dpath, time_t mtime,
850     uint32_t seed, struct got_repository *repo, int loose_obj_only,
851     int *ncolored, int *nfound, int *ntrees,
852     got_pack_progress_cb progress_cb, void *progress_arg,
853     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
854 {
855 	const struct got_error *err = NULL;
856 	struct got_object_id_queue tree_ids;
857 	struct got_object_qid *qid;
858 	struct got_tree_object *tree = NULL;
859 
860 	if (got_object_idset_contains(idset, tree_id) ||
861 	    got_object_idset_contains(idset_exclude, tree_id))
862 		return NULL;
863 
864 	err = got_object_qid_alloc(&qid, tree_id);
865 	if (err)
866 		return err;
867 	qid->data = strdup(dpath);
868 	if (qid->data == NULL) {
869 		err = got_error_from_errno("strdup");
870 		got_object_qid_free(qid);
871 		return err;
872 	}
873 
874 	STAILQ_INIT(&tree_ids);
875 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
876 
877 	while (!STAILQ_EMPTY(&tree_ids)) {
878 		const char *path;
879 		if (cancel_cb) {
880 			err = (*cancel_cb)(cancel_arg);
881 			if (err)
882 				break;
883 		}
884 
885 		qid = STAILQ_FIRST(&tree_ids);
886 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
887 		path = qid->data;
888 
889 		if (got_object_idset_contains(idset, &qid->id) ||
890 		    got_object_idset_contains(idset_exclude, &qid->id)) {
891 			free(qid->data);
892 			got_object_qid_free(qid);
893 			continue;
894 		}
895 
896 		err = got_pack_add_object(want_meta,
897 		    want_meta ? idset : idset_exclude,
898 		    &qid->id, path, GOT_OBJ_TYPE_TREE,
899 		    mtime, seed, loose_obj_only, repo,
900 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
901 		if (err) {
902 			free(qid->data);
903 			got_object_qid_free(qid);
904 			break;
905 		}
906 
907 		err = got_object_open_as_tree(&tree, repo, &qid->id);
908 		if (err) {
909 			free(qid->data);
910 			got_object_qid_free(qid);
911 			break;
912 		}
913 
914 		err = got_pack_load_tree_entries(&tree_ids, want_meta, idset,
915 		    idset_exclude, tree, path, mtime, seed, repo,
916 		    loose_obj_only, ncolored, nfound, ntrees,
917 		    progress_cb, progress_arg, rl,
918 		    cancel_cb, cancel_arg);
919 		free(qid->data);
920 		got_object_qid_free(qid);
921 		if (err)
922 			break;
923 
924 		got_object_tree_close(tree);
925 		tree = NULL;
926 	}
927 
928 	STAILQ_FOREACH(qid, &tree_ids, entry)
929 		free(qid->data);
930 	got_object_id_queue_free(&tree_ids);
931 	if (tree)
932 		got_object_tree_close(tree);
933 	return err;
934 }
935 
936 static const struct got_error *
937 load_commit(int want_meta, struct got_object_idset *idset,
938     struct got_object_idset *idset_exclude,
939     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
940     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
941     got_pack_progress_cb progress_cb, void *progress_arg,
942     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
943 {
944 	const struct got_error *err;
945 	struct got_commit_object *commit;
946 
947 	if (got_object_idset_contains(idset, id) ||
948 	    got_object_idset_contains(idset_exclude, id))
949 		return NULL;
950 
951 	if (loose_obj_only) {
952 		int is_packed;
953 		err = search_packidx(&is_packed, id, repo);
954 		if (err)
955 			return err;
956 		if (is_packed && want_meta)
957 			return NULL;
958 	}
959 
960 	err = got_object_open_as_commit(&commit, repo, id);
961 	if (err)
962 		return err;
963 
964 	err = got_pack_add_object(want_meta,
965 	    want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_COMMIT,
966 	    got_object_commit_get_committer_time(commit), seed,
967 	    loose_obj_only, repo,
968 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
969 	if (err)
970 		goto done;
971 
972 	err = got_pack_load_tree(want_meta, idset, idset_exclude,
973 	    got_object_commit_get_tree_id(commit),
974 	    "", got_object_commit_get_committer_time(commit), seed,
975 	    repo, loose_obj_only, ncolored, nfound, ntrees,
976 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
977 done:
978 	got_object_commit_close(commit);
979 	return err;
980 }
981 
982 static const struct got_error *
983 load_tag(int want_meta, struct got_object_idset *idset,
984     struct got_object_idset *idset_exclude,
985     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
986     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
987     got_pack_progress_cb progress_cb, void *progress_arg,
988     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
989 {
990 	const struct got_error *err;
991 	struct got_tag_object *tag = NULL;
992 
993 	if (got_object_idset_contains(idset, id) ||
994 	    got_object_idset_contains(idset_exclude, id))
995 		return NULL;
996 
997 	if (loose_obj_only) {
998 		int is_packed;
999 		err = search_packidx(&is_packed, id, repo);
1000 		if (err)
1001 			return err;
1002 		if (is_packed && want_meta)
1003 			return NULL;
1004 	}
1005 
1006 	err = got_object_open_as_tag(&tag, repo, id);
1007 	if (err)
1008 		return err;
1009 
1010 	err = got_pack_add_object(want_meta,
1011 	    want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_TAG,
1012 	    got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1013 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1014 	if (err)
1015 		goto done;
1016 
1017 	switch (got_object_tag_get_object_type(tag)) {
1018 	case GOT_OBJ_TYPE_COMMIT:
1019 		err = load_commit(want_meta, idset, idset_exclude,
1020 		    got_object_tag_get_object_id(tag), repo, seed,
1021 		    loose_obj_only, ncolored, nfound, ntrees,
1022 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1023 		break;
1024 	case GOT_OBJ_TYPE_TREE:
1025 		err = got_pack_load_tree(want_meta, idset, idset_exclude,
1026 		    got_object_tag_get_object_id(tag), "",
1027 		    got_object_tag_get_tagger_time(tag), seed, repo,
1028 		    loose_obj_only, ncolored, nfound, ntrees,
1029 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1030 		break;
1031 	default:
1032 		break;
1033 	}
1034 
1035 done:
1036 	got_object_tag_close(tag);
1037 	return err;
1038 }
1039 
1040 const struct got_error *
1041 got_pack_paint_commit(struct got_object_qid *qid, intptr_t color)
1042 {
1043 	if (color < 0 || color >= COLOR_MAX)
1044 		return got_error(GOT_ERR_RANGE);
1045 
1046 	qid->data = (void *)color;
1047 	return NULL;
1048 }
1049 
1050 const struct got_error *
1051 got_pack_queue_commit_id(struct got_object_id_queue *ids,
1052     struct got_object_id *id, intptr_t color, struct got_repository *repo)
1053 {
1054 	const struct got_error *err;
1055 	struct got_object_qid *qid;
1056 
1057 	err = got_object_qid_alloc(&qid, id);
1058 	if (err)
1059 		return err;
1060 
1061 	STAILQ_INSERT_TAIL(ids, qid, entry);
1062 	return got_pack_paint_commit(qid, color);
1063 }
1064 
1065 struct append_id_arg {
1066 	struct got_object_id **array;
1067 	int idx;
1068 	struct got_object_idset *drop;
1069 	struct got_object_idset *skip;
1070 };
1071 
1072 static const struct got_error *
1073 append_id(struct got_object_id *id, void *data, void *arg)
1074 {
1075 	struct append_id_arg *a = arg;
1076 
1077 	if (got_object_idset_contains(a->skip, id) ||
1078 	    got_object_idset_contains(a->drop, id))
1079 		return NULL;
1080 
1081 	a->array[++a->idx] = got_object_id_dup(id);
1082 	if (a->array[a->idx] == NULL)
1083 		return got_error_from_errno("got_object_id_dup");
1084 
1085 	return NULL;
1086 }
1087 
1088 static const struct got_error *
1089 queue_commit_or_tag_id(struct got_object_id *id, intptr_t color,
1090     struct got_object_id_queue *ids, struct got_repository *repo)
1091 {
1092 	const struct got_error *err;
1093 	struct got_tag_object *tag = NULL;
1094 	int obj_type;
1095 
1096 	err = got_object_get_type(&obj_type, repo, id);
1097 	if (err)
1098 		return err;
1099 
1100 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1101 		err = got_object_open_as_tag(&tag, repo, id);
1102 		if (err)
1103 			return err;
1104 		obj_type = got_object_tag_get_object_type(tag);
1105 		id = got_object_tag_get_object_id(tag);
1106 	}
1107 
1108 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1109 		err = got_pack_queue_commit_id(ids, id, color, repo);
1110 		if (err)
1111 			goto done;
1112 	}
1113 done:
1114 	if (tag)
1115 		got_object_tag_close(tag);
1116 	return err;
1117 }
1118 
1119 const struct got_error *
1120 got_pack_find_pack_for_commit_painting(struct got_packidx **best_packidx,
1121     struct got_object_id_queue *ids, int nids, struct got_repository *repo)
1122 {
1123 	const struct got_error *err = NULL;
1124 	struct got_pathlist_entry *pe;
1125 	const char *best_packidx_path = NULL;
1126 	int nobj_max = 0;
1127 	int ncommits_max = 0;
1128 
1129 	*best_packidx = NULL;
1130 
1131 	/*
1132 	 * Find the largest pack which contains at least some of the
1133 	 * commits we are interested in.
1134 	 */
1135 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1136 		const char *path_packidx = pe->path;
1137 		struct got_packidx *packidx;
1138 		int nobj, idx, ncommits = 0;
1139 		struct got_object_qid *qid;
1140 
1141 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1142 		if (err)
1143 			break;
1144 
1145 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1146 		if (nobj <= nobj_max)
1147 			continue;
1148 
1149 		STAILQ_FOREACH(qid, ids, entry) {
1150 			idx = got_packidx_get_object_idx(packidx, &qid->id);
1151 			if (idx != -1)
1152 				ncommits++;
1153 		}
1154 		if (ncommits > ncommits_max) {
1155 			best_packidx_path = path_packidx;
1156 			nobj_max = nobj;
1157 			ncommits_max = ncommits;
1158 		}
1159 	}
1160 
1161 	if (best_packidx_path && err == NULL) {
1162 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1163 		    repo);
1164 	}
1165 
1166 	return err;
1167 }
1168 
1169 static const struct got_error *
1170 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1171     struct got_object_id **head, int nhead,
1172     struct got_object_id **tail, int ntail,
1173     struct got_repository *repo,
1174     got_pack_progress_cb progress_cb, void *progress_arg,
1175     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1176 {
1177 	const struct got_error *err = NULL;
1178 	struct got_object_id_queue ids;
1179 	struct got_object_idset *keep, *drop, *skip = NULL;
1180 	int i, nkeep;
1181 
1182 	STAILQ_INIT(&ids);
1183 	*res = NULL;
1184 	*nres = 0;
1185 	*ncolored = 0;
1186 
1187 	keep = got_object_idset_alloc();
1188 	if (keep == NULL)
1189 		return got_error_from_errno("got_object_idset_alloc");
1190 
1191 	drop = got_object_idset_alloc();
1192 	if (drop == NULL) {
1193 		err = got_error_from_errno("got_object_idset_alloc");
1194 		goto done;
1195 	}
1196 
1197 	skip = got_object_idset_alloc();
1198 	if (skip == NULL) {
1199 		err = got_error_from_errno("got_object_idset_alloc");
1200 		goto done;
1201 	}
1202 
1203 	for (i = 0; i < nhead; i++) {
1204 		struct got_object_id *id = head[i];
1205 		if (id == NULL)
1206 			continue;
1207 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1208 		if (err)
1209 			goto done;
1210 	}
1211 
1212 	for (i = 0; i < ntail; i++) {
1213 		struct got_object_id *id = tail[i];
1214 		if (id == NULL)
1215 			continue;
1216 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1217 		if (err)
1218 			goto done;
1219 	}
1220 
1221 	err = got_pack_paint_commits(ncolored, &ids, nhead + ntail,
1222 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1223 	    cancel_cb, cancel_arg);
1224 	if (err)
1225 		goto done;
1226 
1227 	nkeep = got_object_idset_num_elements(keep);
1228 	if (nkeep > 0) {
1229 		struct append_id_arg arg;
1230 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1231 		if (arg.array == NULL) {
1232 			err = got_error_from_errno("calloc");
1233 			goto done;
1234 		}
1235 		arg.idx = -1;
1236 		arg.skip = skip;
1237 		arg.drop = drop;
1238 		err = got_object_idset_for_each(keep, append_id, &arg);
1239 		if (err) {
1240 			free(arg.array);
1241 			goto done;
1242 		}
1243 		*res = arg.array;
1244 		*nres = arg.idx + 1;
1245 	}
1246 done:
1247 	got_object_idset_free(keep);
1248 	got_object_idset_free(drop);
1249 	if (skip)
1250 		got_object_idset_free(skip);
1251 	got_object_id_queue_free(&ids);
1252 	return err;
1253 }
1254 
1255 static const struct got_error *
1256 find_pack_for_enumeration(struct got_packidx **best_packidx,
1257     struct got_object_id **ids, int nids, struct got_repository *repo)
1258 {
1259 	const struct got_error *err = NULL;
1260 	struct got_pathlist_entry *pe;
1261 	const char *best_packidx_path = NULL;
1262 	int nobj_max = 0;
1263 	int ncommits_max = 0;
1264 
1265 	*best_packidx = NULL;
1266 
1267 	/*
1268 	 * Find the largest pack which contains at least some of the
1269 	 * commits and tags we are interested in.
1270 	 */
1271 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1272 		const char *path_packidx = pe->path;
1273 		struct got_packidx *packidx;
1274 		int nobj, i, idx, ncommits = 0;
1275 
1276 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1277 		if (err)
1278 			break;
1279 
1280 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1281 		if (nobj <= nobj_max)
1282 			continue;
1283 
1284 		for (i = 0; i < nids; i++) {
1285 			idx = got_packidx_get_object_idx(packidx, ids[i]);
1286 			if (idx != -1)
1287 				ncommits++;
1288 		}
1289 		if (ncommits > ncommits_max) {
1290 			best_packidx_path = path_packidx;
1291 			nobj_max = nobj;
1292 			ncommits_max = ncommits;
1293 		}
1294 	}
1295 
1296 	if (best_packidx_path && err == NULL) {
1297 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1298 		    repo);
1299 	}
1300 
1301 	return err;
1302 }
1303 
1304 static const struct got_error *
1305 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1306     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1307     struct got_object_id **ours, int nours, struct got_repository *repo,
1308     uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1309     void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1310     void *cancel_arg)
1311 {
1312 	const struct got_error *err = NULL;
1313 	struct got_object_id **ids = NULL;
1314 	struct got_packidx *packidx = NULL;
1315 	int i, nobj = 0, obj_type, found_all_objects = 0;
1316 	struct got_object_idset *idset_exclude;
1317 
1318 	idset_exclude = got_object_idset_alloc();
1319 	if (idset_exclude == NULL)
1320 		return got_error_from_errno("got_object_idset_alloc");
1321 
1322 	*ncolored = 0;
1323 	*nfound = 0;
1324 	*ntrees = 0;
1325 
1326 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1327 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1328 	if (err)
1329 		goto done;
1330 
1331 	err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
1332 	if (err)
1333 		goto done;
1334 	if (packidx) {
1335 		err = got_pack_load_packed_object_ids(&found_all_objects,
1336 		    theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
1337 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1338 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1339 		if (err)
1340 			goto done;
1341 	}
1342 
1343 	for (i = 0; i < ntheirs; i++) {
1344 		struct got_object_id *id = theirs[i];
1345 		if (id == NULL)
1346 			continue;
1347 		err = got_object_get_type(&obj_type, repo, id);
1348 		if (err)
1349 			return err;
1350 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1351 			if (!found_all_objects) {
1352 				err = load_commit(0, idset, idset_exclude,
1353 				    id, repo, seed, loose_obj_only,
1354 				    ncolored, nfound, ntrees,
1355 				    progress_cb, progress_arg, rl,
1356 				    cancel_cb, cancel_arg);
1357 				if (err)
1358 					goto done;
1359 			}
1360 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1361 			err = load_tag(0, idset, idset_exclude, id, repo,
1362 			    seed, loose_obj_only, ncolored, nfound, ntrees,
1363 			    progress_cb, progress_arg, rl,
1364 			    cancel_cb, cancel_arg);
1365 			if (err)
1366 				goto done;
1367 		}
1368 	}
1369 
1370 	found_all_objects = 0;
1371 	err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
1372 	if (err)
1373 		goto done;
1374 	if (packidx) {
1375 		err = got_pack_load_packed_object_ids(&found_all_objects, ids,
1376 		    nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
1377 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1378 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1379 		if (err)
1380 			goto done;
1381 	}
1382 
1383 	if (!found_all_objects) {
1384 		for (i = 0; i < nobj; i++) {
1385 			err = load_commit(1, idset, idset_exclude, ids[i],
1386 			    repo, seed, loose_obj_only, ncolored, nfound,
1387 			    ntrees, progress_cb, progress_arg, rl,
1388 			    cancel_cb, cancel_arg);
1389 			if (err)
1390 				goto done;
1391 		}
1392 	}
1393 
1394 	for (i = 0; i < nours; i++) {
1395 		struct got_object_id *id = ours[i];
1396 		struct got_pack_meta *m;
1397 		if (id == NULL)
1398 			continue;
1399 		m = got_object_idset_get(idset, id);
1400 		if (m == NULL) {
1401 			err = got_object_get_type(&obj_type, repo, id);
1402 			if (err)
1403 				goto done;
1404 		} else
1405 			obj_type = m->obj_type;
1406 		if (obj_type != GOT_OBJ_TYPE_TAG)
1407 			continue;
1408 		err = load_tag(1, idset, idset_exclude, id, repo,
1409 		    seed, loose_obj_only, ncolored, nfound, ntrees,
1410 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1411 		if (err)
1412 			goto done;
1413 	}
1414 done:
1415 	for (i = 0; i < nobj; i++) {
1416 		free(ids[i]);
1417 	}
1418 	free(ids);
1419 	got_object_idset_free(idset_exclude);
1420 	return err;
1421 }
1422 
1423 static const struct got_error *
1424 hwrite(int fd, const void *buf, off_t len, struct got_hash *ctx)
1425 {
1426 	got_hash_update(ctx, buf, len);
1427 	return got_poll_write_full(fd, buf, len);
1428 }
1429 
1430 static const struct got_error *
1431 hcopy(FILE *fsrc, int fd_dst, off_t len, struct got_hash *ctx)
1432 {
1433 	const struct got_error *err;
1434 	unsigned char buf[65536];
1435 	off_t remain = len;
1436 	size_t n;
1437 
1438 	while (remain > 0) {
1439 		size_t copylen = MIN(sizeof(buf), remain);
1440 		n = fread(buf, 1, copylen, fsrc);
1441 		if (n != copylen)
1442 			return got_ferror(fsrc, GOT_ERR_IO);
1443 		got_hash_update(ctx, buf, copylen);
1444 		err = got_poll_write_full(fd_dst, buf, copylen);
1445 		if (err)
1446 			return err;
1447 		remain -= copylen;
1448 	}
1449 
1450 	return NULL;
1451 }
1452 
1453 static const struct got_error *
1454 hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
1455     int fd, off_t len, struct got_hash *ctx)
1456 {
1457 	if (src_offset + len > src_size)
1458 		return got_error(GOT_ERR_RANGE);
1459 
1460 	got_hash_update(ctx, src + src_offset, len);
1461 	return got_poll_write_full(fd, src + src_offset, len);
1462 }
1463 
1464 static void
1465 putbe32(char *b, uint32_t n)
1466 {
1467 	b[0] = n >> 24;
1468 	b[1] = n >> 16;
1469 	b[2] = n >> 8;
1470 	b[3] = n >> 0;
1471 }
1472 
1473 static int
1474 write_order_cmp(const void *pa, const void *pb)
1475 {
1476 	struct got_pack_meta *a, *b, *ahd, *bhd;
1477 
1478 	a = *(struct got_pack_meta **)pa;
1479 	b = *(struct got_pack_meta **)pb;
1480 	ahd = (a->head == NULL) ? a : a->head;
1481 	bhd = (b->head == NULL) ? b : b->head;
1482 	if (bhd->mtime < ahd->mtime)
1483 		return -1;
1484 	if (bhd->mtime > ahd->mtime)
1485 		return 1;
1486 	if (bhd < ahd)
1487 		return -1;
1488 	if (bhd > ahd)
1489 		return 1;
1490 	if (a->nchain != b->nchain)
1491 		return a->nchain - b->nchain;
1492 	if (a->mtime < b->mtime)
1493 		return -1;
1494 	if (a->mtime > b->mtime)
1495 		return 1;
1496 	return got_object_id_cmp(&a->id, &b->id);
1497 }
1498 
1499 static int
1500 reuse_write_order_cmp(const void *pa, const void *pb)
1501 {
1502 	struct got_pack_meta *a, *b;
1503 
1504 	a = *(struct got_pack_meta **)pa;
1505 	b = *(struct got_pack_meta **)pb;
1506 
1507 	if (a->reused_delta_offset < b->reused_delta_offset)
1508 		return -1;
1509 	if (a->reused_delta_offset > b->reused_delta_offset)
1510 		return 1;
1511 	return 0;
1512 }
1513 
1514 static const struct got_error *
1515 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1516 {
1517 	size_t i;
1518 
1519 	*hdrlen = 0;
1520 
1521 	hdr[0] = obj_type << 4;
1522 	hdr[0] |= len & 0xf;
1523 	len >>= 4;
1524 	for (i = 1; len != 0; i++){
1525 		if (i >= bufsize)
1526 			return got_error(GOT_ERR_NO_SPACE);
1527 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1528 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1529 		len >>= GOT_DELTA_SIZE_SHIFT;
1530 	}
1531 
1532 	*hdrlen = i;
1533 	return NULL;
1534 }
1535 
1536 static int
1537 packoff(char *hdr, off_t off)
1538 {
1539 	int i, j;
1540 	char rbuf[8];
1541 
1542 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1543 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1544 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1545 		    GOT_DELTA_SIZE_MORE;
1546 	}
1547 
1548 	j = 0;
1549 	while (i > 0)
1550 		hdr[j++] = rbuf[--i];
1551 	return j;
1552 }
1553 
1554 static const struct got_error *
1555 deltahdr(off_t *packfile_size, struct got_hash *ctx, int packfd,
1556     int force_refdelta, struct got_pack_meta *m)
1557 {
1558 	const struct got_error *err;
1559 	char buf[32];
1560 	int nh;
1561 
1562 	if (m->prev->off != 0 && !force_refdelta) {
1563 		err = packhdr(&nh, buf, sizeof(buf),
1564 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1565 		if (err)
1566 			return err;
1567 		nh += packoff(buf + nh, m->off - m->prev->off);
1568 		err = hwrite(packfd, buf, nh, ctx);
1569 		if (err)
1570 			return err;
1571 		*packfile_size += nh;
1572 	} else {
1573 		err = packhdr(&nh, buf, sizeof(buf),
1574 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1575 		if (err)
1576 			return err;
1577 		err = hwrite(packfd, buf, nh, ctx);
1578 		if (err)
1579 			return err;
1580 		*packfile_size += nh;
1581 		err = hwrite(packfd, m->prev->id.sha1,
1582 		    sizeof(m->prev->id.sha1), ctx);
1583 		if (err)
1584 			return err;
1585 		*packfile_size += sizeof(m->prev->id.sha1);
1586 	}
1587 
1588 	return NULL;
1589 }
1590 
1591 static const struct got_error *
1592 write_packed_object(off_t *packfile_size, int packfd,
1593     FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
1594     struct got_pack_meta *m, int *outfd, struct got_hash *ctx,
1595     struct got_repository *repo, int force_refdelta)
1596 {
1597 	const struct got_error *err = NULL;
1598 	struct got_deflate_checksum csum;
1599 	char buf[32];
1600 	int nh;
1601 	struct got_raw_object *raw = NULL;
1602 	off_t outlen, delta_offset;
1603 
1604 	memset(&csum, 0, sizeof(csum));
1605 	csum.output_ctx = ctx;
1606 
1607 	if (m->reused_delta_offset)
1608 		delta_offset = m->reused_delta_offset;
1609 	else
1610 		delta_offset = m->delta_offset;
1611 
1612 	m->off = *packfile_size;
1613 	if (m->delta_len == 0) {
1614 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1615 		if (err)
1616 			goto done;
1617 		err = packhdr(&nh, buf, sizeof(buf),
1618 		    m->obj_type, raw->size);
1619 		if (err)
1620 			goto done;
1621 		err = hwrite(packfd, buf, nh, ctx);
1622 		if (err)
1623 			goto done;
1624 		*packfile_size += nh;
1625 		if (raw->f == NULL) {
1626 			err = got_deflate_to_fd_mmap(&outlen,
1627 			    raw->data + raw->hdrlen, 0, raw->size,
1628 			    packfd, &csum);
1629 			if (err)
1630 				goto done;
1631 		} else {
1632 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1633 			    == -1) {
1634 				err = got_error_from_errno("fseeko");
1635 				goto done;
1636 			}
1637 			err = got_deflate_to_fd(&outlen, raw->f,
1638 			    raw->size, packfd, &csum);
1639 			if (err)
1640 				goto done;
1641 		}
1642 		*packfile_size += outlen;
1643 		got_object_raw_close(raw);
1644 		raw = NULL;
1645 	} else if (m->delta_buf) {
1646 		err = deltahdr(packfile_size, ctx, packfd, force_refdelta, m);
1647 		if (err)
1648 			goto done;
1649 		err = hwrite(packfd, m->delta_buf,
1650 		    m->delta_compressed_len, ctx);
1651 		if (err)
1652 			goto done;
1653 		*packfile_size += m->delta_compressed_len;
1654 		free(m->delta_buf);
1655 		m->delta_buf = NULL;
1656 	} else if (delta_cache_map) {
1657 		err = deltahdr(packfile_size, ctx, packfd, force_refdelta, m);
1658 		if (err)
1659 			goto done;
1660 		err = hcopy_mmap(delta_cache_map, delta_offset,
1661 		    delta_cache_size, packfd, m->delta_compressed_len,
1662 		    ctx);
1663 		if (err)
1664 			goto done;
1665 		*packfile_size += m->delta_compressed_len;
1666 	} else {
1667 		if (fseeko(delta_cache, delta_offset, SEEK_SET) == -1) {
1668 			err = got_error_from_errno("fseeko");
1669 			goto done;
1670 		}
1671 		err = deltahdr(packfile_size, ctx, packfd, force_refdelta, m);
1672 		if (err)
1673 			goto done;
1674 		err = hcopy(delta_cache, packfd,
1675 		    m->delta_compressed_len, ctx);
1676 		if (err)
1677 			goto done;
1678 		*packfile_size += m->delta_compressed_len;
1679 	}
1680 done:
1681 	if (raw)
1682 		got_object_raw_close(raw);
1683 	return err;
1684 }
1685 
1686 static const struct got_error *
1687 genpack(uint8_t *pack_sha1, int packfd, struct got_pack *reuse_pack,
1688     FILE *delta_cache, struct got_pack_meta **deltify, int ndeltify,
1689     struct got_pack_meta **reuse, int nreuse,
1690     int ncolored, int nfound, int ntrees, int nours,
1691     struct got_repository *repo, int force_refdelta,
1692     got_pack_progress_cb progress_cb, void *progress_arg,
1693     struct got_ratelimit *rl,
1694     got_cancel_cb cancel_cb, void *cancel_arg)
1695 {
1696 	const struct got_error *err = NULL;
1697 	int i;
1698 	struct got_hash ctx;
1699 	struct got_pack_meta *m;
1700 	char buf[32];
1701 	off_t packfile_size = 0;
1702 	int outfd = -1;
1703 	int delta_cache_fd = -1;
1704 	uint8_t *delta_cache_map = NULL;
1705 	size_t delta_cache_size = 0;
1706 	FILE *packfile = NULL;
1707 
1708 	got_hash_init(&ctx, GOT_HASH_SHA1);
1709 
1710 #ifndef GOT_PACK_NO_MMAP
1711 	delta_cache_fd = dup(fileno(delta_cache));
1712 	if (delta_cache_fd != -1) {
1713 		struct stat sb;
1714 		if (fstat(delta_cache_fd, &sb) == -1) {
1715 			err = got_error_from_errno("fstat");
1716 			goto done;
1717 		}
1718 		if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
1719 			delta_cache_map = mmap(NULL, sb.st_size,
1720 			    PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
1721 			if (delta_cache_map == MAP_FAILED) {
1722 				if (errno != ENOMEM) {
1723 					err = got_error_from_errno("mmap");
1724 					goto done;
1725 				}
1726 				delta_cache_map = NULL; /* fallback on stdio */
1727 			} else
1728 				delta_cache_size = (size_t)sb.st_size;
1729 		}
1730 	}
1731 #endif
1732 	err = hwrite(packfd, "PACK", 4, &ctx);
1733 	if (err)
1734 		goto done;
1735 	putbe32(buf, GOT_PACKFILE_VERSION);
1736 	err = hwrite(packfd, buf, 4, &ctx);
1737 	if (err)
1738 		goto done;
1739 	putbe32(buf, ndeltify + nreuse);
1740 	err = hwrite(packfd, buf, 4, &ctx);
1741 	if (err)
1742 		goto done;
1743 
1744 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1745 	    write_order_cmp);
1746 	for (i = 0; i < ndeltify; i++) {
1747 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
1748 		    ncolored, nfound, ntrees, packfile_size, nours,
1749 		    ndeltify + nreuse, ndeltify + nreuse, i);
1750 		if (err)
1751 			goto done;
1752 		m = deltify[i];
1753 		err = write_packed_object(&packfile_size, packfd,
1754 		    delta_cache, delta_cache_map, delta_cache_size,
1755 		    m, &outfd, &ctx, repo, force_refdelta);
1756 		if (err)
1757 			goto done;
1758 	}
1759 
1760 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1761 	    reuse_write_order_cmp);
1762 	if (nreuse > 0 && reuse_pack->map == NULL) {
1763 		int fd = dup(reuse_pack->fd);
1764 		if (fd == -1) {
1765 			err = got_error_from_errno("dup");
1766 			goto done;
1767 		}
1768 		packfile = fdopen(fd, "r");
1769 		if (packfile == NULL) {
1770 			err = got_error_from_errno("fdopen");
1771 			close(fd);
1772 			goto done;
1773 		}
1774 	}
1775 	for (i = 0; i < nreuse; i++) {
1776 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
1777 		    ncolored, nfound, ntrees, packfile_size, nours,
1778 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1779 		if (err)
1780 			goto done;
1781 		m = reuse[i];
1782 		err = write_packed_object(&packfile_size, packfd,
1783 		    packfile, reuse_pack->map, reuse_pack->filesize,
1784 		    m, &outfd, &ctx, repo, force_refdelta);
1785 		if (err)
1786 			goto done;
1787 	}
1788 
1789 	got_hash_final(&ctx, pack_sha1);
1790 	err = got_poll_write_full(packfd, pack_sha1, SHA1_DIGEST_LENGTH);
1791 	if (err)
1792 		goto done;
1793 	packfile_size += SHA1_DIGEST_LENGTH;
1794 	packfile_size += sizeof(struct got_packfile_hdr);
1795 	if (progress_cb) {
1796 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1797 		    packfile_size, nours, ndeltify + nreuse,
1798 		    ndeltify + nreuse, ndeltify + nreuse);
1799 		if (err)
1800 			goto done;
1801 	}
1802 done:
1803 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1804 		err = got_error_from_errno("close");
1805 	if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
1806 		err = got_error_from_errno("munmap");
1807 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1808 		err = got_error_from_errno("close");
1809 	if (packfile && fclose(packfile) == EOF && err == NULL)
1810 		err = got_error_from_errno("fclose");
1811 	return err;
1812 }
1813 
1814 static const struct got_error *
1815 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1816 {
1817 	struct got_pack_meta *m = data;
1818 	struct got_pack_metavec *v = arg;
1819 
1820 	if (m->reused_delta_offset != 0)
1821 		return NULL;
1822 
1823 	return got_pack_add_meta(m, v);
1824 }
1825 
1826 const struct got_error *
1827 got_pack_create(uint8_t *packsha1, int packfd, FILE *delta_cache,
1828     struct got_object_id **theirs, int ntheirs,
1829     struct got_object_id **ours, int nours,
1830     struct got_repository *repo, int loose_obj_only, int allow_empty,
1831     int force_refdelta, got_pack_progress_cb progress_cb, void *progress_arg,
1832     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1833 {
1834 	const struct got_error *err;
1835 	struct got_object_idset *idset;
1836 	struct got_packidx *reuse_packidx = NULL;
1837 	struct got_pack *reuse_pack = NULL;
1838 	struct got_pack_metavec deltify, reuse;
1839 	int ncolored = 0, nfound = 0, ntrees = 0;
1840 	size_t ndeltify;
1841 	uint32_t seed;
1842 
1843 	seed = arc4random();
1844 
1845 	memset(&deltify, 0, sizeof(deltify));
1846 	memset(&reuse, 0, sizeof(reuse));
1847 
1848 	idset = got_object_idset_alloc();
1849 	if (idset == NULL)
1850 		return got_error_from_errno("got_object_idset_alloc");
1851 
1852 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1853 	    ntheirs, ours, nours, repo, seed, loose_obj_only,
1854 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1855 	if (err)
1856 		goto done;
1857 
1858 	if (progress_cb) {
1859 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1860 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1861 		if (err)
1862 			goto done;
1863 	}
1864 
1865 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1866 		err = got_error(GOT_ERR_CANNOT_PACK);
1867 		goto done;
1868 	}
1869 
1870 	reuse.metasz = 64;
1871 	reuse.meta = calloc(reuse.metasz,
1872 	    sizeof(struct got_pack_meta *));
1873 	if (reuse.meta == NULL) {
1874 		err = got_error_from_errno("calloc");
1875 		goto done;
1876 	}
1877 
1878 	err = got_pack_search_deltas(&reuse_packidx, &reuse_pack,
1879 	    &reuse, idset, ncolored, nfound, ntrees, nours,
1880 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1881 	if (err)
1882 		goto done;
1883 
1884 	if (reuse_packidx && reuse_pack) {
1885 		err = got_repo_pin_pack(repo, reuse_packidx, reuse_pack);
1886 		if (err)
1887 			goto done;
1888 	}
1889 
1890 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1891 		err = got_error_from_errno("fseeko");
1892 		goto done;
1893 	}
1894 
1895 	ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
1896 	if (ndeltify > 0) {
1897 		deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
1898 		if (deltify.meta == NULL) {
1899 			err = got_error_from_errno("calloc");
1900 			goto done;
1901 		}
1902 		deltify.metasz = ndeltify;
1903 
1904 		err = got_object_idset_for_each(idset, add_meta_idset_cb,
1905 		    &deltify);
1906 		if (err)
1907 			goto done;
1908 		if (deltify.nmeta > 0) {
1909 			err = pick_deltas(deltify.meta, deltify.nmeta,
1910 			    ncolored, nfound, ntrees, nours, reuse.nmeta,
1911 			    delta_cache, repo, progress_cb, progress_arg, rl,
1912 			    cancel_cb, cancel_arg);
1913 			if (err)
1914 				goto done;
1915 		}
1916 	}
1917 
1918 	if (fflush(delta_cache) == EOF) {
1919 		err = got_error_from_errno("fflush");
1920 		goto done;
1921 	}
1922 
1923 	if (progress_cb) {
1924 		/*
1925 		 * Report a 1-byte packfile write to indicate we are about
1926 		 * to start sending packfile data. gotd(8) needs this.
1927 		 */
1928 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1929 		    1 /* packfile_size */, nours,
1930 		    got_object_idset_num_elements(idset),
1931 		    deltify.nmeta + reuse.nmeta, 0);
1932 		if (err)
1933 			goto done;
1934 	}
1935 
1936 	/* Pinned pack may have moved to different cache slot. */
1937 	reuse_pack = got_repo_get_pinned_pack(repo);
1938 
1939 	err = genpack(packsha1, packfd, reuse_pack, delta_cache, deltify.meta,
1940 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1941 	    nours, repo, force_refdelta, progress_cb, progress_arg, rl,
1942 	    cancel_cb, cancel_arg);
1943 	if (err)
1944 		goto done;
1945 done:
1946 	free_nmeta(deltify.meta, deltify.nmeta);
1947 	free_nmeta(reuse.meta, reuse.nmeta);
1948 	got_object_idset_free(idset);
1949 	got_repo_unpin_pack(repo);
1950 	return err;
1951 }