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 <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <sha1.h>
35 #include <time.h>
36 #include <unistd.h>
37 #include <limits.h>
38 #include <zlib.h>
39 
40 #include "got_error.h"
41 #include "got_cancel.h"
42 #include "got_object.h"
43 #include "got_path.h"
44 #include "got_reference.h"
45 #include "got_repository_admin.h"
46 
47 #include "got_lib_deltify.h"
48 #include "got_lib_delta.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_idset.h"
51 #include "got_lib_object_cache.h"
52 #include "got_lib_deflate.h"
53 #include "got_lib_ratelimit.h"
54 #include "got_lib_pack.h"
55 #include "got_lib_pack_create.h"
56 #include "got_lib_repository.h"
57 #include "got_lib_inflate.h"
58 
59 #include "murmurhash2.h"
60 
61 #ifndef MIN
62 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
63 #endif
64 
65 #ifndef MAX
66 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
67 #endif
68 
69 #ifndef nitems
70 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
71 #endif
72 
73 static const struct got_error *
74 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
75     const char *path, int obj_type, time_t mtime, uint32_t seed)
76 {
77 	struct got_pack_meta *m;
78 
79 	*new = NULL;
80 
81 	m = calloc(1, sizeof(*m));
82 	if (m == NULL)
83 		return got_error_from_errno("calloc");
84 
85 	memcpy(&m->id, id, sizeof(m->id));
86 
87 	m->path_hash = murmurhash2(path, strlen(path), seed);
88 	m->obj_type = obj_type;
89 	m->mtime = mtime;
90 	*new = m;
91 	return NULL;
92 }
93 
94 static void
95 clear_meta(struct got_pack_meta *meta)
96 {
97 	if (meta == NULL)
98 		return;
99 	meta->path_hash = 0;
100 	free(meta->delta_buf);
101 	meta->delta_buf = NULL;
102 	free(meta->base_obj_id);
103 	meta->base_obj_id = NULL;
104 	meta->reused_delta_offset = 0;
105 }
106 
107 static void
108 free_nmeta(struct got_pack_meta **meta, int nmeta)
109 {
110 	int i;
111 
112 	for (i = 0; i < nmeta; i++)
113 		clear_meta(meta[i]);
114 	free(meta);
115 }
116 
117 static int
118 delta_order_cmp(const void *pa, const void *pb)
119 {
120 	struct got_pack_meta *a, *b;
121 
122 	a = *(struct got_pack_meta **)pa;
123 	b = *(struct got_pack_meta **)pb;
124 
125 	if (a->obj_type != b->obj_type)
126 		return a->obj_type - b->obj_type;
127 	if (a->path_hash < b->path_hash)
128 		return -1;
129 	if (a->path_hash > b->path_hash)
130 		return 1;
131 	if (a->mtime < b->mtime)
132 		return -1;
133 	if (a->mtime > b->mtime)
134 		return 1;
135 	return got_object_id_cmp(&a->id, &b->id);
136 }
137 
138 static off_t
139 delta_size(struct got_delta_instruction *deltas, int ndeltas)
140 {
141 	int i;
142 	off_t size = 32;
143 	for (i = 0; i < ndeltas; i++) {
144 		if (deltas[i].copy)
145 			size += GOT_DELTA_SIZE_SHIFT;
146 		else
147 			size += deltas[i].len + 1;
148 	}
149 	return size;
150 }
151 
152 static const struct got_error *
153 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
154 {
155 	char *n;
156 
157 	if (*len + nseg >= *sz) {
158 		while (*len + nseg >= *sz)
159 			*sz += *sz / 2;
160 		n = realloc(*p, *sz);
161 		if (n == NULL)
162 			return got_error_from_errno("realloc");
163 		*p = n;
164 	}
165 	memcpy(*p + *len, seg, nseg);
166 	*len += nseg;
167 	return NULL;
168 }
169 
170 static const struct got_error *
171 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
172     struct got_delta_instruction *deltas, int ndeltas,
173     off_t delta_size, off_t base_size)
174 {
175 	const struct got_error *err;
176 	unsigned char buf[16], *bp;
177 	int i, j;
178 	size_t len = 0, compressed_len;
179 	off_t bufsize = delta_size;
180 	off_t n;
181 	struct got_delta_instruction *d;
182 	uint8_t *delta_buf;
183 
184 	delta_buf = malloc(bufsize);
185 	if (delta_buf == NULL)
186 		return got_error_from_errno("malloc");
187 
188 	/* base object size */
189 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
190 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
191 	for (i = 1; n > 0; i++) {
192 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
193 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
194 		n >>= GOT_DELTA_SIZE_SHIFT;
195 	}
196 	err = append(&delta_buf, &len, &bufsize, buf, i);
197 	if (err)
198 		goto done;
199 
200 	/* target object size */
201 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
202 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
203 	for (i = 1; n > 0; i++) {
204 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
205 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
206 		n >>= GOT_DELTA_SIZE_SHIFT;
207 	}
208 	err = append(&delta_buf, &len, &bufsize, buf, i);
209 	if (err)
210 		goto done;
211 
212 	for (j = 0; j < ndeltas; j++) {
213 		d = &deltas[j];
214 		if (d->copy) {
215 			n = d->offset;
216 			bp = &buf[1];
217 			buf[0] = GOT_DELTA_BASE_COPY;
218 			for (i = 0; i < 4; i++) {
219 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
220 				buf[0] |= 1 << i;
221 				*bp++ = n & 0xff;
222 				n >>= 8;
223 				if (n == 0)
224 					break;
225 			}
226 
227 			n = d->len;
228 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
229 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
230 				for (i = 0; i < 3 && n > 0; i++) {
231 					buf[0] |= 1 << (i + 4);
232 					*bp++ = n & 0xff;
233 					n >>= 8;
234 				}
235 			}
236 			err = append(&delta_buf, &len, &bufsize,
237 			    buf, bp - buf);
238 			if (err)
239 				goto done;
240 		} else if (o->f == NULL) {
241 			n = 0;
242 			while (n != d->len) {
243 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
244 				err = append(&delta_buf, &len, &bufsize,
245 				    buf, 1);
246 				if (err)
247 					goto done;
248 				err = append(&delta_buf, &len, &bufsize,
249 				    o->data + o->hdrlen + d->offset + n,
250 				    buf[0]);
251 				if (err)
252 					goto done;
253 				n += buf[0];
254 			}
255 		} else {
256 			char content[128];
257 			size_t r;
258 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
259 				err = got_error_from_errno("fseeko");
260 				goto done;
261 			}
262 			n = 0;
263 			while (n != d->len) {
264 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
265 				err = append(&delta_buf, &len, &bufsize,
266 				    buf, 1);
267 				if (err)
268 					goto done;
269 				r = fread(content, 1, buf[0], o->f);
270 				if (r != buf[0]) {
271 					err = got_ferror(o->f, GOT_ERR_IO);
272 					goto done;
273 				}
274 				err = append(&delta_buf, &len, &bufsize,
275 				    content, buf[0]);
276 				if (err)
277 					goto done;
278 				n += buf[0];
279 			}
280 		}
281 	}
282 
283 	err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
284 	    NULL, NULL, delta_buf, 0, len);
285 	if (err)
286 		goto done;
287 
288 	m->delta_len = len;
289 	m->delta_compressed_len = compressed_len;
290 done:
291 	free(delta_buf);
292 	return err;
293 }
294 
295 static const struct got_error *
296 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
297     struct got_delta_instruction *deltas, int ndeltas,
298     off_t base_size, FILE *f)
299 {
300 	const struct got_error *err;
301 	unsigned char buf[16], *bp;
302 	int i, j;
303 	off_t n;
304 	struct got_deflate_buf zb;
305 	struct got_delta_instruction *d;
306 	off_t delta_len = 0, compressed_len = 0;
307 
308 	err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
309 	if (err)
310 		return err;
311 
312 	/* base object size */
313 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
314 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
315 	for (i = 1; n > 0; i++) {
316 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
317 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
318 		n >>= GOT_DELTA_SIZE_SHIFT;
319 	}
320 
321 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
322 	    buf, 0, i, f, NULL);
323 	if (err)
324 		goto done;
325 	delta_len += i;
326 
327 	/* target object size */
328 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
329 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
330 	for (i = 1; n > 0; i++) {
331 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
332 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
333 		n >>= GOT_DELTA_SIZE_SHIFT;
334 	}
335 
336 	err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
337 	    buf, 0, i, f, NULL);
338 	if (err)
339 		goto done;
340 	delta_len += i;
341 
342 	for (j = 0; j < ndeltas; j++) {
343 		d = &deltas[j];
344 		if (d->copy) {
345 			n = d->offset;
346 			bp = &buf[1];
347 			buf[0] = GOT_DELTA_BASE_COPY;
348 			for (i = 0; i < 4; i++) {
349 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
350 				buf[0] |= 1 << i;
351 				*bp++ = n & 0xff;
352 				n >>= 8;
353 				if (n == 0)
354 					break;
355 			}
356 			n = d->len;
357 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
358 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
359 				for (i = 0; i < 3 && n > 0; i++) {
360 					buf[0] |= 1 << (i + 4);
361 					*bp++ = n & 0xff;
362 					n >>= 8;
363 				}
364 			}
365 			err = got_deflate_append_to_file_mmap(&zb,
366 			    &compressed_len, buf, 0, bp - buf, f, NULL);
367 			if (err)
368 				goto done;
369 			delta_len += (bp - buf);
370 		} else if (o->f == NULL) {
371 			n = 0;
372 			while (n != d->len) {
373 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
374 				err = got_deflate_append_to_file_mmap(&zb,
375 				    &compressed_len, buf, 0, 1, f, NULL);
376 				if (err)
377 					goto done;
378 				delta_len++;
379 				err = got_deflate_append_to_file_mmap(&zb,
380 				    &compressed_len,
381 				    o->data + o->hdrlen + d->offset + n, 0,
382 				    buf[0], f, NULL);
383 				if (err)
384 					goto done;
385 				delta_len += buf[0];
386 				n += buf[0];
387 			}
388 		} else {
389 			char content[128];
390 			size_t r;
391 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
392 				err = got_error_from_errno("fseeko");
393 				goto done;
394 			}
395 			n = 0;
396 			while (n != d->len) {
397 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
398 				err = got_deflate_append_to_file_mmap(&zb,
399 				    &compressed_len, buf, 0, 1, f, NULL);
400 				if (err)
401 					goto done;
402 				delta_len++;
403 				r = fread(content, 1, buf[0], o->f);
404 				if (r != buf[0]) {
405 					err = got_ferror(o->f, GOT_ERR_IO);
406 					goto done;
407 				}
408 				err = got_deflate_append_to_file_mmap(&zb,
409 				    &compressed_len, content, 0, buf[0], f,
410 				    NULL);
411 				if (err)
412 					goto done;
413 				delta_len += buf[0];
414 				n += buf[0];
415 			}
416 		}
417 	}
418 
419 	err = got_deflate_flush(&zb, f, NULL, &compressed_len);
420 	if (err)
421 		goto done;
422 
423 	/* sanity check */
424 	if (compressed_len != ftello(f) - m->delta_offset) {
425 		err = got_error(GOT_ERR_COMPRESSION);
426 		goto done;
427 	}
428 
429 	m->delta_len = delta_len;
430 	m->delta_compressed_len = compressed_len;
431 done:
432 	got_deflate_end(&zb);
433 	return err;
434 }
435 
436 const struct got_error *
437 got_pack_report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
438     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
439     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
440     int nobj_written)
441 {
442 	const struct got_error *err;
443 	int elapsed;
444 
445 	if (progress_cb == NULL)
446 		return NULL;
447 
448 	err = got_ratelimit_check(&elapsed, rl);
449 	if (err || !elapsed)
450 		return err;
451 
452 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
453 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
454 }
455 
456 const struct got_error *
457 got_pack_add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
458 {
459 	if (v->nmeta == v->metasz){
460 		size_t newsize = 2 * v->metasz;
461 		struct got_pack_meta **new;
462 		new = reallocarray(v->meta, newsize, sizeof(*new));
463 		if (new == NULL)
464 			return got_error_from_errno("reallocarray");
465 		v->meta = new;
466 		v->metasz = newsize;
467 	}
468 
469 	v->meta[v->nmeta++] = m;
470 	return NULL;
471 }
472 
473 const struct got_error *
474 got_pack_find_pack_for_reuse(struct got_packidx **best_packidx,
475     struct got_repository *repo)
476 {
477 	const struct got_error *err = NULL;
478 	struct got_pathlist_entry *pe;
479 	const char *best_packidx_path = NULL;
480 	int nobj_max = 0;
481 
482 	*best_packidx = NULL;
483 
484 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
485 		const char *path_packidx = pe->path;
486 		struct got_packidx *packidx;
487 		int nobj;
488 
489 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
490 		if (err)
491 			break;
492 
493 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
494 		if (nobj > nobj_max) {
495 			best_packidx_path = path_packidx;
496 			nobj_max = nobj;
497 		}
498 	}
499 
500 	if (best_packidx_path) {
501 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
502 		    repo);
503 	}
504 
505 	return err;
506 }
507 
508 const struct got_error *
509 got_pack_cache_pack_for_packidx(struct got_pack **pack,
510     struct got_packidx *packidx, struct got_repository *repo)
511 {
512 	const struct got_error *err;
513 	char *path_packfile = NULL;
514 
515 	err = got_packidx_get_packfile_path(&path_packfile,
516 	    packidx->path_packidx);
517 	if (err)
518 		return err;
519 
520 	*pack = got_repo_get_cached_pack(repo, path_packfile);
521 	if (*pack == NULL) {
522 		err = got_repo_cache_pack(pack, repo, path_packfile, packidx);
523 		if (err)
524 			goto done;
525 	}
526 done:
527 	free(path_packfile);
528 	return err;
529 }
530 
531 static const struct got_error *
532 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
533     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
534     struct got_repository *repo,
535     got_pack_progress_cb progress_cb, void *progress_arg,
536     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
537 {
538 	const struct got_error *err = NULL;
539 	struct got_pack_meta *m = NULL, *base = NULL;
540 	struct got_raw_object *raw = NULL, *base_raw = NULL;
541 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
542 	int i, j, ndeltas, best_ndeltas;
543 	off_t size, best_size;
544 	const int max_base_candidates = 3;
545 	size_t delta_memsize = 0;
546 	const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
547 	int outfd = -1;
548 	uint32_t delta_seed;
549 
550 	delta_seed = arc4random();
551 
552 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
553 	for (i = 0; i < nmeta; i++) {
554 		if (cancel_cb) {
555 			err = (*cancel_cb)(cancel_arg);
556 			if (err)
557 				break;
558 		}
559 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
560 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
561 		    nreused + i, 0);
562 		if (err)
563 			goto done;
564 		m = meta[i];
565 
566 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
567 		    m->obj_type == GOT_OBJ_TYPE_TAG)
568 			continue;
569 
570 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
571 		if (err)
572 			goto done;
573 		m->size = raw->size;
574 
575 		if (raw->f == NULL) {
576 			err = got_deltify_init_mem(&m->dtab, raw->data,
577 			    raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
578 		} else {
579 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
580 			    raw->size + raw->hdrlen, delta_seed);
581 		}
582 		if (err)
583 			goto done;
584 
585 		if (i > max_base_candidates) {
586 			struct got_pack_meta *n = NULL;
587 			n = meta[i - (max_base_candidates + 1)];
588 			got_deltify_free(n->dtab);
589 			n->dtab = NULL;
590 		}
591 
592 		best_size = raw->size;
593 		best_ndeltas = 0;
594 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
595 			if (cancel_cb) {
596 				err = (*cancel_cb)(cancel_arg);
597 				if (err)
598 					goto done;
599 			}
600 			base = meta[j];
601 			/* long chains make unpacking slow, avoid such bases */
602 			if (base->nchain >= 128 ||
603 			    base->obj_type != m->obj_type)
604 				continue;
605 
606 			err = got_object_raw_open(&base_raw, &outfd, repo,
607 			    &base->id);
608 			if (err)
609 				goto done;
610 
611 			if (raw->f == NULL && base_raw->f == NULL) {
612 				err = got_deltify_mem_mem(&deltas, &ndeltas,
613 				    raw->data, raw->hdrlen,
614 				    raw->size + raw->hdrlen, delta_seed,
615 				    base->dtab, base_raw->data,
616 				    base_raw->hdrlen,
617 				    base_raw->size + base_raw->hdrlen);
618 			} else if (raw->f == NULL) {
619 				err = got_deltify_mem_file(&deltas, &ndeltas,
620 				    raw->data, raw->hdrlen,
621 				    raw->size + raw->hdrlen, delta_seed,
622 				    base->dtab, base_raw->f,
623 				    base_raw->hdrlen,
624 				    base_raw->size + base_raw->hdrlen);
625 			} else if (base_raw->f == NULL) {
626 				err = got_deltify_file_mem(&deltas, &ndeltas,
627 				    raw->f, raw->hdrlen,
628 				    raw->size + raw->hdrlen, delta_seed,
629 				    base->dtab, base_raw->data,
630 				    base_raw->hdrlen,
631 				    base_raw->size + base_raw->hdrlen);
632 			} else {
633 				err = got_deltify(&deltas, &ndeltas,
634 				    raw->f, raw->hdrlen,
635 				    raw->size + raw->hdrlen, delta_seed,
636 				    base->dtab, base_raw->f, base_raw->hdrlen,
637 				    base_raw->size + base_raw->hdrlen);
638 			}
639 			got_object_raw_close(base_raw);
640 			base_raw = NULL;
641 			if (err)
642 				goto done;
643 
644 			size = delta_size(deltas, ndeltas);
645 			if (size + 32 < best_size){
646 				/*
647 				 * if we already picked a best delta,
648 				 * replace it.
649 				 */
650 				best_size = size;
651 				free(best_deltas);
652 				best_deltas = deltas;
653 				best_ndeltas = ndeltas;
654 				deltas = NULL;
655 				m->nchain = base->nchain + 1;
656 				m->prev = base;
657 				m->head = base->head;
658 				if (m->head == NULL)
659 					m->head = base;
660 			} else {
661 				free(deltas);
662 				deltas = NULL;
663 				ndeltas = 0;
664 			}
665 		}
666 
667 		if (best_ndeltas > 0) {
668 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
669 			    delta_memsize + best_size <= max_delta_memsize) {
670 				delta_memsize += best_size;
671 				err = encode_delta_in_mem(m, raw, best_deltas,
672 				    best_ndeltas, best_size, m->prev->size);
673 			} else {
674 				m->delta_offset = ftello(delta_cache);
675 				err = encode_delta(m, raw, best_deltas,
676 				    best_ndeltas, m->prev->size, delta_cache);
677 			}
678 			free(best_deltas);
679 			best_deltas = NULL;
680 			best_ndeltas = 0;
681 			if (err)
682 				goto done;
683 		}
684 
685 		got_object_raw_close(raw);
686 		raw = NULL;
687 	}
688 done:
689 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
690 		got_deltify_free(meta[i]->dtab);
691 		meta[i]->dtab = NULL;
692 	}
693 	if (raw)
694 		got_object_raw_close(raw);
695 	if (base_raw)
696 		got_object_raw_close(base_raw);
697 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
698 		err = got_error_from_errno("close");
699 	free(deltas);
700 	free(best_deltas);
701 	return err;
702 }
703 
704 static const struct got_error *
705 search_packidx(int *found, struct got_object_id *id,
706     struct got_repository *repo)
707 {
708 	const struct got_error *err = NULL;
709 	struct got_packidx *packidx = NULL;
710 	int idx;
711 
712 	*found = 0;
713 
714 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
715 	if (err == NULL)
716 		*found = 1; /* object is already packed */
717 	else if (err->code == GOT_ERR_NO_OBJ)
718 		err = NULL;
719 	return err;
720 }
721 
722 const struct got_error *
723 got_pack_add_object(int want_meta, struct got_object_idset *idset,
724     struct got_object_id *id, const char *path, int obj_type,
725     time_t mtime, uint32_t seed, int loose_obj_only,
726     struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
727     got_pack_progress_cb progress_cb, void *progress_arg,
728     struct got_ratelimit *rl)
729 {
730 	const struct got_error *err;
731 	struct got_pack_meta *m = NULL;
732 
733 	if (loose_obj_only) {
734 		int is_packed;
735 		err = search_packidx(&is_packed, id, repo);
736 		if (err)
737 			return err;
738 		if (is_packed && want_meta)
739 			return NULL;
740 	}
741 
742 	if (want_meta) {
743 		err = alloc_meta(&m, id, path, obj_type, mtime, seed);
744 		if (err)
745 			return err;
746 
747 		(*nfound)++;
748 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
749 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
750 		if (err) {
751 			clear_meta(m);
752 			free(m);
753 			return err;
754 		}
755 	}
756 
757 	err = got_object_idset_add(idset, id, m);
758 	if (err) {
759 		clear_meta(m);
760 		free(m);
761 	}
762 	return err;
763 }
764 
765 const struct got_error *
766 got_pack_load_tree_entries(struct got_object_id_queue *ids, int want_meta,
767     struct got_object_idset *idset, struct got_object_idset *idset_exclude,
768     struct got_tree_object *tree,
769     const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
770     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
771     got_pack_progress_cb progress_cb, void *progress_arg,
772     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
773 {
774 	const struct got_error *err;
775 	char *p = NULL;
776 	int i;
777 
778 	(*ntrees)++;
779 	err = got_pack_report_progress(progress_cb, progress_arg, rl,
780 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
781 	if (err)
782 		return err;
783 
784 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
785 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
786 		struct got_object_id *id = got_tree_entry_get_id(e);
787 		mode_t mode = got_tree_entry_get_mode(e);
788 
789 		if (cancel_cb) {
790 			err = (*cancel_cb)(cancel_arg);
791 			if (err)
792 				break;
793 		}
794 
795 		if (got_object_tree_entry_is_submodule(e) ||
796 		    got_object_idset_contains(idset, id) ||
797 		    got_object_idset_contains(idset_exclude, id))
798 			continue;
799 
800 		/*
801 		 * If got-read-pack is crawling trees for us then
802 		 * we are only here to collect blob IDs.
803 		 */
804 		if (ids == NULL && S_ISDIR(mode))
805 			continue;
806 
807 		if (asprintf(&p, "%s%s%s", dpath,
808 		    got_path_is_root_dir(dpath) ? "" : "/",
809 		    got_tree_entry_get_name(e)) == -1) {
810 			err = got_error_from_errno("asprintf");
811 			break;
812 		}
813 
814 		if (S_ISDIR(mode)) {
815 			struct got_object_qid *qid;
816 			err = got_object_qid_alloc(&qid, id);
817 			if (err)
818 				break;
819 			qid->data = p;
820 			p = NULL;
821 			STAILQ_INSERT_TAIL(ids, qid, entry);
822 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
823 			err = got_pack_add_object(want_meta,
824 			    want_meta ? idset : idset_exclude, id, p,
825 			    GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
826 			    repo, ncolored, nfound, ntrees,
827 			    progress_cb, progress_arg, rl);
828 			if (err)
829 				break;
830 			free(p);
831 			p = NULL;
832 		} else {
833 			free(p);
834 			p = NULL;
835 		}
836 	}
837 
838 	free(p);
839 	return err;
840 }
841 
842 const struct got_error *
843 got_pack_load_tree(int want_meta, struct got_object_idset *idset,
844     struct got_object_idset *idset_exclude,
845     struct got_object_id *tree_id, const char *dpath, time_t mtime,
846     uint32_t seed, struct got_repository *repo, int loose_obj_only,
847     int *ncolored, int *nfound, int *ntrees,
848     got_pack_progress_cb progress_cb, void *progress_arg,
849     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
850 {
851 	const struct got_error *err = NULL;
852 	struct got_object_id_queue tree_ids;
853 	struct got_object_qid *qid;
854 	struct got_tree_object *tree = NULL;
855 
856 	if (got_object_idset_contains(idset, tree_id) ||
857 	    got_object_idset_contains(idset_exclude, tree_id))
858 		return NULL;
859 
860 	err = got_object_qid_alloc(&qid, tree_id);
861 	if (err)
862 		return err;
863 	qid->data = strdup(dpath);
864 	if (qid->data == NULL) {
865 		err = got_error_from_errno("strdup");
866 		got_object_qid_free(qid);
867 		return err;
868 	}
869 
870 	STAILQ_INIT(&tree_ids);
871 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
872 
873 	while (!STAILQ_EMPTY(&tree_ids)) {
874 		const char *path;
875 		if (cancel_cb) {
876 			err = (*cancel_cb)(cancel_arg);
877 			if (err)
878 				break;
879 		}
880 
881 		qid = STAILQ_FIRST(&tree_ids);
882 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
883 		path = qid->data;
884 
885 		if (got_object_idset_contains(idset, &qid->id) ||
886 		    got_object_idset_contains(idset_exclude, &qid->id)) {
887 			free(qid->data);
888 			got_object_qid_free(qid);
889 			continue;
890 		}
891 
892 		err = got_pack_add_object(want_meta,
893 		    want_meta ? idset : idset_exclude,
894 		    &qid->id, path, GOT_OBJ_TYPE_TREE,
895 		    mtime, seed, loose_obj_only, repo,
896 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
897 		if (err) {
898 			free(qid->data);
899 			got_object_qid_free(qid);
900 			break;
901 		}
902 
903 		err = got_object_open_as_tree(&tree, repo, &qid->id);
904 		if (err) {
905 			free(qid->data);
906 			got_object_qid_free(qid);
907 			break;
908 		}
909 
910 		err = got_pack_load_tree_entries(&tree_ids, want_meta, idset,
911 		    idset_exclude, tree, path, mtime, seed, repo,
912 		    loose_obj_only, ncolored, nfound, ntrees,
913 		    progress_cb, progress_arg, rl,
914 		    cancel_cb, cancel_arg);
915 		free(qid->data);
916 		got_object_qid_free(qid);
917 		if (err)
918 			break;
919 
920 		got_object_tree_close(tree);
921 		tree = NULL;
922 	}
923 
924 	STAILQ_FOREACH(qid, &tree_ids, entry)
925 		free(qid->data);
926 	got_object_id_queue_free(&tree_ids);
927 	if (tree)
928 		got_object_tree_close(tree);
929 	return err;
930 }
931 
932 static const struct got_error *
933 load_commit(int want_meta, struct got_object_idset *idset,
934     struct got_object_idset *idset_exclude,
935     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
936     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
937     got_pack_progress_cb progress_cb, void *progress_arg,
938     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
939 {
940 	const struct got_error *err;
941 	struct got_commit_object *commit;
942 
943 	if (got_object_idset_contains(idset, id) ||
944 	    got_object_idset_contains(idset_exclude, id))
945 		return NULL;
946 
947 	if (loose_obj_only) {
948 		int is_packed;
949 		err = search_packidx(&is_packed, id, repo);
950 		if (err)
951 			return err;
952 		if (is_packed && want_meta)
953 			return NULL;
954 	}
955 
956 	err = got_object_open_as_commit(&commit, repo, id);
957 	if (err)
958 		return err;
959 
960 	err = got_pack_add_object(want_meta,
961 	    want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_COMMIT,
962 	    got_object_commit_get_committer_time(commit), seed,
963 	    loose_obj_only, repo,
964 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
965 	if (err)
966 		goto done;
967 
968 	err = got_pack_load_tree(want_meta, idset, idset_exclude,
969 	    got_object_commit_get_tree_id(commit),
970 	    "", got_object_commit_get_committer_time(commit), seed,
971 	    repo, loose_obj_only, ncolored, nfound, ntrees,
972 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
973 done:
974 	got_object_commit_close(commit);
975 	return err;
976 }
977 
978 static const struct got_error *
979 load_tag(int want_meta, struct got_object_idset *idset,
980     struct got_object_idset *idset_exclude,
981     struct got_object_id *id, struct got_repository *repo, uint32_t seed,
982     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
983     got_pack_progress_cb progress_cb, void *progress_arg,
984     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
985 {
986 	const struct got_error *err;
987 	struct got_tag_object *tag = NULL;
988 
989 	if (got_object_idset_contains(idset, id) ||
990 	    got_object_idset_contains(idset_exclude, id))
991 		return NULL;
992 
993 	if (loose_obj_only) {
994 		int is_packed;
995 		err = search_packidx(&is_packed, id, repo);
996 		if (err)
997 			return err;
998 		if (is_packed && want_meta)
999 			return NULL;
1000 	}
1001 
1002 	err = got_object_open_as_tag(&tag, repo, id);
1003 	if (err)
1004 		return err;
1005 
1006 	err = got_pack_add_object(want_meta,
1007 	    want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_TAG,
1008 	    got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1009 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1010 	if (err)
1011 		goto done;
1012 
1013 	switch (got_object_tag_get_object_type(tag)) {
1014 	case GOT_OBJ_TYPE_COMMIT:
1015 		err = load_commit(want_meta, idset, idset_exclude,
1016 		    got_object_tag_get_object_id(tag), repo, seed,
1017 		    loose_obj_only, ncolored, nfound, ntrees,
1018 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1019 		break;
1020 	case GOT_OBJ_TYPE_TREE:
1021 		err = got_pack_load_tree(want_meta, idset, idset_exclude,
1022 		    got_object_tag_get_object_id(tag), "",
1023 		    got_object_tag_get_tagger_time(tag), seed, repo,
1024 		    loose_obj_only, ncolored, nfound, ntrees,
1025 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1026 		break;
1027 	default:
1028 		break;
1029 	}
1030 
1031 done:
1032 	got_object_tag_close(tag);
1033 	return err;
1034 }
1035 
1036 const struct got_error *
1037 got_pack_paint_commit(struct got_object_qid *qid, intptr_t color)
1038 {
1039 	if (color < 0 || color >= COLOR_MAX)
1040 		return got_error(GOT_ERR_RANGE);
1041 
1042 	qid->data = (void *)color;
1043 	return NULL;
1044 }
1045 
1046 const struct got_error *
1047 got_pack_queue_commit_id(struct got_object_id_queue *ids,
1048     struct got_object_id *id, intptr_t color, struct got_repository *repo)
1049 {
1050 	const struct got_error *err;
1051 	struct got_object_qid *qid;
1052 
1053 	err = got_object_qid_alloc(&qid, id);
1054 	if (err)
1055 		return err;
1056 
1057 	STAILQ_INSERT_TAIL(ids, qid, entry);
1058 	return got_pack_paint_commit(qid, color);
1059 }
1060 
1061 struct append_id_arg {
1062 	struct got_object_id **array;
1063 	int idx;
1064 	struct got_object_idset *drop;
1065 	struct got_object_idset *skip;
1066 };
1067 
1068 static const struct got_error *
1069 append_id(struct got_object_id *id, void *data, void *arg)
1070 {
1071 	struct append_id_arg *a = arg;
1072 
1073 	if (got_object_idset_contains(a->skip, id) ||
1074 	    got_object_idset_contains(a->drop, id))
1075 		return NULL;
1076 
1077 	a->array[++a->idx] = got_object_id_dup(id);
1078 	if (a->array[a->idx] == NULL)
1079 		return got_error_from_errno("got_object_id_dup");
1080 
1081 	return NULL;
1082 }
1083 
1084 static const struct got_error *
1085 queue_commit_or_tag_id(struct got_object_id *id, intptr_t color,
1086     struct got_object_id_queue *ids, struct got_repository *repo)
1087 {
1088 	const struct got_error *err;
1089 	struct got_tag_object *tag = NULL;
1090 	int obj_type;
1091 
1092 	err = got_object_get_type(&obj_type, repo, id);
1093 	if (err)
1094 		return err;
1095 
1096 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1097 		err = got_object_open_as_tag(&tag, repo, id);
1098 		if (err)
1099 			return err;
1100 		obj_type = got_object_tag_get_object_type(tag);
1101 		id = got_object_tag_get_object_id(tag);
1102 	}
1103 
1104 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1105 		err = got_pack_queue_commit_id(ids, id, color, repo);
1106 		if (err)
1107 			goto done;
1108 	}
1109 done:
1110 	if (tag)
1111 		got_object_tag_close(tag);
1112 	return err;
1113 }
1114 
1115 const struct got_error *
1116 got_pack_find_pack_for_commit_painting(struct got_packidx **best_packidx,
1117     struct got_object_id_queue *ids, int nids, struct got_repository *repo)
1118 {
1119 	const struct got_error *err = NULL;
1120 	struct got_pathlist_entry *pe;
1121 	const char *best_packidx_path = NULL;
1122 	int nobj_max = 0;
1123 	int ncommits_max = 0;
1124 
1125 	*best_packidx = NULL;
1126 
1127 	/*
1128 	 * Find the largest pack which contains at least some of the
1129 	 * commits we are interested in.
1130 	 */
1131 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1132 		const char *path_packidx = pe->path;
1133 		struct got_packidx *packidx;
1134 		int nobj, idx, ncommits = 0;
1135 		struct got_object_qid *qid;
1136 
1137 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1138 		if (err)
1139 			break;
1140 
1141 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1142 		if (nobj <= nobj_max)
1143 			continue;
1144 
1145 		STAILQ_FOREACH(qid, ids, entry) {
1146 			idx = got_packidx_get_object_idx(packidx, &qid->id);
1147 			if (idx != -1)
1148 				ncommits++;
1149 		}
1150 		if (ncommits > ncommits_max) {
1151 			best_packidx_path = path_packidx;
1152 			nobj_max = nobj;
1153 			ncommits_max = ncommits;
1154 		}
1155 	}
1156 
1157 	if (best_packidx_path && err == NULL) {
1158 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1159 		    repo);
1160 	}
1161 
1162 	return err;
1163 }
1164 
1165 static const struct got_error *
1166 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1167     struct got_object_id **head, int nhead,
1168     struct got_object_id **tail, int ntail,
1169     struct got_repository *repo,
1170     got_pack_progress_cb progress_cb, void *progress_arg,
1171     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1172 {
1173 	const struct got_error *err = NULL;
1174 	struct got_object_id_queue ids;
1175 	struct got_object_idset *keep, *drop, *skip = NULL;
1176 	int i, nkeep;
1177 
1178 	STAILQ_INIT(&ids);
1179 	*res = NULL;
1180 	*nres = 0;
1181 	*ncolored = 0;
1182 
1183 	keep = got_object_idset_alloc();
1184 	if (keep == NULL)
1185 		return got_error_from_errno("got_object_idset_alloc");
1186 
1187 	drop = got_object_idset_alloc();
1188 	if (drop == NULL) {
1189 		err = got_error_from_errno("got_object_idset_alloc");
1190 		goto done;
1191 	}
1192 
1193 	skip = got_object_idset_alloc();
1194 	if (skip == NULL) {
1195 		err = got_error_from_errno("got_object_idset_alloc");
1196 		goto done;
1197 	}
1198 
1199 	for (i = 0; i < nhead; i++) {
1200 		struct got_object_id *id = head[i];
1201 		if (id == NULL)
1202 			continue;
1203 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1204 		if (err)
1205 			goto done;
1206 	}
1207 
1208 	for (i = 0; i < ntail; i++) {
1209 		struct got_object_id *id = tail[i];
1210 		if (id == NULL)
1211 			continue;
1212 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1213 		if (err)
1214 			goto done;
1215 	}
1216 
1217 	err = got_pack_paint_commits(ncolored, &ids, nhead + ntail,
1218 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1219 	    cancel_cb, cancel_arg);
1220 	if (err)
1221 		goto done;
1222 
1223 	nkeep = got_object_idset_num_elements(keep);
1224 	if (nkeep > 0) {
1225 		struct append_id_arg arg;
1226 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1227 		if (arg.array == NULL) {
1228 			err = got_error_from_errno("calloc");
1229 			goto done;
1230 		}
1231 		arg.idx = -1;
1232 		arg.skip = skip;
1233 		arg.drop = drop;
1234 		err = got_object_idset_for_each(keep, append_id, &arg);
1235 		if (err) {
1236 			free(arg.array);
1237 			goto done;
1238 		}
1239 		*res = arg.array;
1240 		*nres = arg.idx + 1;
1241 	}
1242 done:
1243 	got_object_idset_free(keep);
1244 	got_object_idset_free(drop);
1245 	if (skip)
1246 		got_object_idset_free(skip);
1247 	got_object_id_queue_free(&ids);
1248 	return err;
1249 }
1250 
1251 static const struct got_error *
1252 find_pack_for_enumeration(struct got_packidx **best_packidx,
1253     struct got_object_id **ids, int nids, struct got_repository *repo)
1254 {
1255 	const struct got_error *err = NULL;
1256 	struct got_pathlist_entry *pe;
1257 	const char *best_packidx_path = NULL;
1258 	int nobj_max = 0;
1259 	int ncommits_max = 0;
1260 
1261 	*best_packidx = NULL;
1262 
1263 	/*
1264 	 * Find the largest pack which contains at least some of the
1265 	 * commits and tags we are interested in.
1266 	 */
1267 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1268 		const char *path_packidx = pe->path;
1269 		struct got_packidx *packidx;
1270 		int nobj, i, idx, ncommits = 0;
1271 
1272 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
1273 		if (err)
1274 			break;
1275 
1276 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1277 		if (nobj <= nobj_max)
1278 			continue;
1279 
1280 		for (i = 0; i < nids; i++) {
1281 			idx = got_packidx_get_object_idx(packidx, ids[i]);
1282 			if (idx != -1)
1283 				ncommits++;
1284 		}
1285 		if (ncommits > ncommits_max) {
1286 			best_packidx_path = path_packidx;
1287 			nobj_max = nobj;
1288 			ncommits_max = ncommits;
1289 		}
1290 	}
1291 
1292 	if (best_packidx_path && err == NULL) {
1293 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
1294 		    repo);
1295 	}
1296 
1297 	return err;
1298 }
1299 
1300 static const struct got_error *
1301 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1302     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1303     struct got_object_id **ours, int nours, struct got_repository *repo,
1304     uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1305     void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1306     void *cancel_arg)
1307 {
1308 	const struct got_error *err = NULL;
1309 	struct got_object_id **ids = NULL;
1310 	struct got_packidx *packidx = NULL;
1311 	int i, nobj = 0, obj_type, found_all_objects = 0;
1312 	struct got_object_idset *idset_exclude;
1313 
1314 	idset_exclude = got_object_idset_alloc();
1315 	if (idset_exclude == NULL)
1316 		return got_error_from_errno("got_object_idset_alloc");
1317 
1318 	*ncolored = 0;
1319 	*nfound = 0;
1320 	*ntrees = 0;
1321 
1322 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1323 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1324 	if (err)
1325 		goto done;
1326 
1327 	err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
1328 	if (err)
1329 		goto done;
1330 	if (packidx) {
1331 		err = got_pack_load_packed_object_ids(&found_all_objects,
1332 		    theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
1333 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1334 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1335 		if (err)
1336 			goto done;
1337 	}
1338 
1339 	for (i = 0; i < ntheirs; i++) {
1340 		struct got_object_id *id = theirs[i];
1341 		if (id == NULL)
1342 			continue;
1343 		err = got_object_get_type(&obj_type, repo, id);
1344 		if (err)
1345 			return err;
1346 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1347 			if (!found_all_objects) {
1348 				err = load_commit(0, idset, idset_exclude,
1349 				    id, repo, seed, loose_obj_only,
1350 				    ncolored, nfound, ntrees,
1351 				    progress_cb, progress_arg, rl,
1352 				    cancel_cb, cancel_arg);
1353 				if (err)
1354 					goto done;
1355 			}
1356 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1357 			err = load_tag(0, idset, idset_exclude, id, repo,
1358 			    seed, loose_obj_only, ncolored, nfound, ntrees,
1359 			    progress_cb, progress_arg, rl,
1360 			    cancel_cb, cancel_arg);
1361 			if (err)
1362 				goto done;
1363 		}
1364 	}
1365 
1366 	found_all_objects = 0;
1367 	err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
1368 	if (err)
1369 		goto done;
1370 	if (packidx) {
1371 		err = got_pack_load_packed_object_ids(&found_all_objects, ids,
1372 		    nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
1373 		    loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1374 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1375 		if (err)
1376 			goto done;
1377 	}
1378 
1379 	if (!found_all_objects) {
1380 		for (i = 0; i < nobj; i++) {
1381 			err = load_commit(1, idset, idset_exclude, ids[i],
1382 			    repo, seed, loose_obj_only, ncolored, nfound,
1383 			    ntrees, progress_cb, progress_arg, rl,
1384 			    cancel_cb, cancel_arg);
1385 			if (err)
1386 				goto done;
1387 		}
1388 	}
1389 
1390 	for (i = 0; i < nours; i++) {
1391 		struct got_object_id *id = ours[i];
1392 		struct got_pack_meta *m;
1393 		if (id == NULL)
1394 			continue;
1395 		m = got_object_idset_get(idset, id);
1396 		if (m == NULL) {
1397 			err = got_object_get_type(&obj_type, repo, id);
1398 			if (err)
1399 				goto done;
1400 		} else
1401 			obj_type = m->obj_type;
1402 		if (obj_type != GOT_OBJ_TYPE_TAG)
1403 			continue;
1404 		err = load_tag(1, idset, idset_exclude, id, repo,
1405 		    seed, loose_obj_only, ncolored, nfound, ntrees,
1406 		    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1407 		if (err)
1408 			goto done;
1409 	}
1410 done:
1411 	for (i = 0; i < nobj; i++) {
1412 		free(ids[i]);
1413 	}
1414 	free(ids);
1415 	got_object_idset_free(idset_exclude);
1416 	return err;
1417 }
1418 
1419 static const struct got_error *
1420 hwrite(int fd, const void *buf, off_t len, SHA1_CTX *ctx)
1421 {
1422 	ssize_t w;
1423 
1424 	SHA1Update(ctx, buf, len);
1425 	w = write(fd, buf, len);
1426 	if (w == -1)
1427 		return got_error_from_errno("write");
1428 	else if (w != len)
1429 		return got_error(GOT_ERR_IO);
1430 	return NULL;
1431 }
1432 
1433 static const struct got_error *
1434 hcopy(FILE *fsrc, int fd_dst, off_t len, SHA1_CTX *ctx)
1435 {
1436 	unsigned char buf[65536];
1437 	off_t remain = len;
1438 	size_t n;
1439 	ssize_t w;
1440 
1441 	while (remain > 0) {
1442 		size_t copylen = MIN(sizeof(buf), remain);
1443 		n = fread(buf, 1, copylen, fsrc);
1444 		if (n != copylen)
1445 			return got_ferror(fsrc, GOT_ERR_IO);
1446 		SHA1Update(ctx, buf, copylen);
1447 		w = write(fd_dst, buf, copylen);
1448 		if (w == -1)
1449 			return got_error_from_errno("write");
1450 		else if (w != copylen)
1451 			return got_error(GOT_ERR_IO);
1452 		remain -= copylen;
1453 	}
1454 
1455 	return NULL;
1456 }
1457 
1458 static const struct got_error *
1459 hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
1460     int fd, off_t len, SHA1_CTX *ctx)
1461 {
1462 	ssize_t w;
1463 
1464 	if (src_offset + len > src_size)
1465 		return got_error(GOT_ERR_RANGE);
1466 
1467 	SHA1Update(ctx, src + src_offset, len);
1468 	w = write(fd, src + src_offset, len);
1469 	if (w == -1)
1470 		return got_error_from_errno("write");
1471 	else if (w != len)
1472 		return got_error(GOT_ERR_IO);
1473 	return NULL;
1474 }
1475 
1476 static void
1477 putbe32(char *b, uint32_t n)
1478 {
1479 	b[0] = n >> 24;
1480 	b[1] = n >> 16;
1481 	b[2] = n >> 8;
1482 	b[3] = n >> 0;
1483 }
1484 
1485 static int
1486 write_order_cmp(const void *pa, const void *pb)
1487 {
1488 	struct got_pack_meta *a, *b, *ahd, *bhd;
1489 
1490 	a = *(struct got_pack_meta **)pa;
1491 	b = *(struct got_pack_meta **)pb;
1492 	ahd = (a->head == NULL) ? a : a->head;
1493 	bhd = (b->head == NULL) ? b : b->head;
1494 	if (bhd->mtime < ahd->mtime)
1495 		return -1;
1496 	if (bhd->mtime > ahd->mtime)
1497 		return 1;
1498 	if (bhd < ahd)
1499 		return -1;
1500 	if (bhd > ahd)
1501 		return 1;
1502 	if (a->nchain != b->nchain)
1503 		return a->nchain - b->nchain;
1504 	if (a->mtime < b->mtime)
1505 		return -1;
1506 	if (a->mtime > b->mtime)
1507 		return 1;
1508 	return got_object_id_cmp(&a->id, &b->id);
1509 }
1510 
1511 static int
1512 reuse_write_order_cmp(const void *pa, const void *pb)
1513 {
1514 	struct got_pack_meta *a, *b;
1515 
1516 	a = *(struct got_pack_meta **)pa;
1517 	b = *(struct got_pack_meta **)pb;
1518 
1519 	if (a->reused_delta_offset < b->reused_delta_offset)
1520 		return -1;
1521 	if (a->reused_delta_offset > b->reused_delta_offset)
1522 		return 1;
1523 	return 0;
1524 }
1525 
1526 static const struct got_error *
1527 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1528 {
1529 	size_t i;
1530 
1531 	*hdrlen = 0;
1532 
1533 	hdr[0] = obj_type << 4;
1534 	hdr[0] |= len & 0xf;
1535 	len >>= 4;
1536 	for (i = 1; len != 0; i++){
1537 		if (i >= bufsize)
1538 			return got_error(GOT_ERR_NO_SPACE);
1539 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1540 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1541 		len >>= GOT_DELTA_SIZE_SHIFT;
1542 	}
1543 
1544 	*hdrlen = i;
1545 	return NULL;
1546 }
1547 
1548 static int
1549 packoff(char *hdr, off_t off)
1550 {
1551 	int i, j;
1552 	char rbuf[8];
1553 
1554 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1555 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1556 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1557 		    GOT_DELTA_SIZE_MORE;
1558 	}
1559 
1560 	j = 0;
1561 	while (i > 0)
1562 		hdr[j++] = rbuf[--i];
1563 	return j;
1564 }
1565 
1566 static const struct got_error *
1567 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, int packfd,
1568     struct got_pack_meta *m)
1569 {
1570 	const struct got_error *err;
1571 	char buf[32];
1572 	int nh;
1573 
1574 	if (m->prev->off != 0) {
1575 		err = packhdr(&nh, buf, sizeof(buf),
1576 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1577 		if (err)
1578 			return err;
1579 		nh += packoff(buf + nh, m->off - m->prev->off);
1580 		err = hwrite(packfd, buf, nh, ctx);
1581 		if (err)
1582 			return err;
1583 		*packfile_size += nh;
1584 	} else {
1585 		err = packhdr(&nh, buf, sizeof(buf),
1586 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1587 		if (err)
1588 			return err;
1589 		err = hwrite(packfd, buf, nh, ctx);
1590 		if (err)
1591 			return err;
1592 		*packfile_size += nh;
1593 		err = hwrite(packfd, m->prev->id.sha1,
1594 		    sizeof(m->prev->id.sha1), ctx);
1595 		if (err)
1596 			return err;
1597 		*packfile_size += sizeof(m->prev->id.sha1);
1598 	}
1599 
1600 	return NULL;
1601 }
1602 
1603 static const struct got_error *
1604 write_packed_object(off_t *packfile_size, int packfd,
1605     FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
1606     struct got_pack_meta *m, int *outfd, SHA1_CTX *ctx,
1607     struct got_repository *repo)
1608 {
1609 	const struct got_error *err = NULL;
1610 	struct got_deflate_checksum csum;
1611 	char buf[32];
1612 	int nh;
1613 	struct got_raw_object *raw = NULL;
1614 	off_t outlen;
1615 
1616 	csum.output_sha1 = ctx;
1617 	csum.output_crc = NULL;
1618 
1619 	m->off = *packfile_size;
1620 	if (m->delta_len == 0) {
1621 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1622 		if (err)
1623 			goto done;
1624 		err = packhdr(&nh, buf, sizeof(buf),
1625 		    m->obj_type, raw->size);
1626 		if (err)
1627 			goto done;
1628 		err = hwrite(packfd, buf, nh, ctx);
1629 		if (err)
1630 			goto done;
1631 		*packfile_size += nh;
1632 		if (raw->f == NULL) {
1633 			err = got_deflate_to_fd_mmap(&outlen,
1634 			    raw->data + raw->hdrlen, 0, raw->size,
1635 			    packfd, &csum);
1636 			if (err)
1637 				goto done;
1638 		} else {
1639 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1640 			    == -1) {
1641 				err = got_error_from_errno("fseeko");
1642 				goto done;
1643 			}
1644 			err = got_deflate_to_fd(&outlen, raw->f,
1645 			    raw->size, packfd, &csum);
1646 			if (err)
1647 				goto done;
1648 		}
1649 		*packfile_size += outlen;
1650 		got_object_raw_close(raw);
1651 		raw = NULL;
1652 	} else if (m->delta_buf) {
1653 		err = deltahdr(packfile_size, ctx, packfd, m);
1654 		if (err)
1655 			goto done;
1656 		err = hwrite(packfd, m->delta_buf,
1657 		    m->delta_compressed_len, ctx);
1658 		if (err)
1659 			goto done;
1660 		*packfile_size += m->delta_compressed_len;
1661 		free(m->delta_buf);
1662 		m->delta_buf = NULL;
1663 	} else if (delta_cache_map) {
1664 		err = deltahdr(packfile_size, ctx, packfd, m);
1665 		if (err)
1666 			goto done;
1667 		err = hcopy_mmap(delta_cache_map, m->delta_offset,
1668 		    delta_cache_size, packfd, m->delta_compressed_len,
1669 		    ctx);
1670 		if (err)
1671 			goto done;
1672 		*packfile_size += m->delta_compressed_len;
1673 	} else {
1674 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1675 		    == -1) {
1676 			err = got_error_from_errno("fseeko");
1677 			goto done;
1678 		}
1679 		err = deltahdr(packfile_size, ctx, packfd, m);
1680 		if (err)
1681 			goto done;
1682 		err = hcopy(delta_cache, packfd,
1683 		    m->delta_compressed_len, ctx);
1684 		if (err)
1685 			goto done;
1686 		*packfile_size += m->delta_compressed_len;
1687 	}
1688 done:
1689 	if (raw)
1690 		got_object_raw_close(raw);
1691 	return err;
1692 }
1693 
1694 static const struct got_error *
1695 genpack(uint8_t *pack_sha1, int packfd, FILE *delta_cache,
1696     struct got_pack_meta **deltify, int ndeltify,
1697     struct got_pack_meta **reuse, int nreuse,
1698     int ncolored, int nfound, int ntrees, int nours,
1699     struct got_repository *repo,
1700     got_pack_progress_cb progress_cb, void *progress_arg,
1701     struct got_ratelimit *rl,
1702     got_cancel_cb cancel_cb, void *cancel_arg)
1703 {
1704 	const struct got_error *err = NULL;
1705 	int i;
1706 	SHA1_CTX ctx;
1707 	struct got_pack_meta *m;
1708 	char buf[32];
1709 	ssize_t w;
1710 	off_t packfile_size = 0;
1711 	int outfd = -1;
1712 	int delta_cache_fd = -1;
1713 	uint8_t *delta_cache_map = NULL;
1714 	size_t delta_cache_size = 0;
1715 
1716 	SHA1Init(&ctx);
1717 
1718 #ifndef GOT_PACK_NO_MMAP
1719 	delta_cache_fd = dup(fileno(delta_cache));
1720 	if (delta_cache_fd != -1) {
1721 		struct stat sb;
1722 		if (fstat(delta_cache_fd, &sb) == -1) {
1723 			err = got_error_from_errno("fstat");
1724 			goto done;
1725 		}
1726 		if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
1727 			delta_cache_map = mmap(NULL, sb.st_size,
1728 			    PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
1729 			if (delta_cache_map == MAP_FAILED) {
1730 				if (errno != ENOMEM) {
1731 					err = got_error_from_errno("mmap");
1732 					goto done;
1733 				}
1734 				delta_cache_map = NULL; /* fallback on stdio */
1735 			} else
1736 				delta_cache_size = (size_t)sb.st_size;
1737 		}
1738 	}
1739 #endif
1740 	err = hwrite(packfd, "PACK", 4, &ctx);
1741 	if (err)
1742 		goto done;
1743 	putbe32(buf, GOT_PACKFILE_VERSION);
1744 	err = hwrite(packfd, buf, 4, &ctx);
1745 	if (err)
1746 		goto done;
1747 	putbe32(buf, ndeltify + nreuse);
1748 	err = hwrite(packfd, buf, 4, &ctx);
1749 	if (err)
1750 		goto done;
1751 
1752 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1753 	    write_order_cmp);
1754 	for (i = 0; i < ndeltify; i++) {
1755 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
1756 		    ncolored, nfound, ntrees, packfile_size, nours,
1757 		    ndeltify + nreuse, ndeltify + nreuse, i);
1758 		if (err)
1759 			goto done;
1760 		m = deltify[i];
1761 		err = write_packed_object(&packfile_size, packfd,
1762 		    delta_cache, delta_cache_map, delta_cache_size,
1763 		    m, &outfd, &ctx, repo);
1764 		if (err)
1765 			goto done;
1766 	}
1767 
1768 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1769 	    reuse_write_order_cmp);
1770 	for (i = 0; i < nreuse; i++) {
1771 		err = got_pack_report_progress(progress_cb, progress_arg, rl,
1772 		    ncolored, nfound, ntrees, packfile_size, nours,
1773 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1774 		if (err)
1775 			goto done;
1776 		m = reuse[i];
1777 		err = write_packed_object(&packfile_size, packfd,
1778 		    delta_cache, delta_cache_map, delta_cache_size,
1779 		    m, &outfd, &ctx, repo);
1780 		if (err)
1781 			goto done;
1782 	}
1783 
1784 	SHA1Final(pack_sha1, &ctx);
1785 	w = write(packfd, pack_sha1, SHA1_DIGEST_LENGTH);
1786 	if (w == -1)
1787 		err = got_error_from_errno("write");
1788 	else if (w != SHA1_DIGEST_LENGTH)
1789 		err = got_error(GOT_ERR_IO);
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 	return err;
1809 }
1810 
1811 static const struct got_error *
1812 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1813 {
1814 	struct got_pack_meta *m = data;
1815 	struct got_pack_metavec *v = arg;
1816 
1817 	if (m->reused_delta_offset != 0)
1818 		return NULL;
1819 
1820 	return got_pack_add_meta(m, v);
1821 }
1822 
1823 const struct got_error *
1824 got_pack_create(uint8_t *packsha1, int packfd, FILE *delta_cache,
1825     struct got_object_id **theirs, int ntheirs,
1826     struct got_object_id **ours, int nours,
1827     struct got_repository *repo, int loose_obj_only, int allow_empty,
1828     got_pack_progress_cb progress_cb, void *progress_arg,
1829     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1830 {
1831 	const struct got_error *err;
1832 	int delta_cache_fd = -1;
1833 	struct got_object_idset *idset;
1834 	struct got_pack_metavec deltify, reuse;
1835 	int ncolored = 0, nfound = 0, ntrees = 0;
1836 	size_t ndeltify;
1837 	uint32_t seed;
1838 
1839 	seed = arc4random();
1840 
1841 	memset(&deltify, 0, sizeof(deltify));
1842 	memset(&reuse, 0, sizeof(reuse));
1843 
1844 	idset = got_object_idset_alloc();
1845 	if (idset == NULL)
1846 		return got_error_from_errno("got_object_idset_alloc");
1847 
1848 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1849 	    ntheirs, ours, nours, repo, seed, loose_obj_only,
1850 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1851 	if (err)
1852 		goto done;
1853 
1854 	if (progress_cb) {
1855 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1856 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1857 		if (err)
1858 			goto done;
1859 	}
1860 
1861 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1862 		err = got_error(GOT_ERR_CANNOT_PACK);
1863 		goto done;
1864 	}
1865 
1866 	delta_cache_fd = dup(fileno(delta_cache));
1867 	if (delta_cache_fd == -1) {
1868 		err = got_error_from_errno("dup");
1869 		goto done;
1870 	}
1871 
1872 	reuse.metasz = 64;
1873 	reuse.meta = calloc(reuse.metasz,
1874 	    sizeof(struct got_pack_meta *));
1875 	if (reuse.meta == NULL) {
1876 		err = got_error_from_errno("calloc");
1877 		goto done;
1878 	}
1879 
1880 	err = got_pack_search_deltas(&reuse, idset, delta_cache_fd,
1881 	    ncolored, nfound, ntrees, nours,
1882 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1883 	if (err)
1884 		goto done;
1885 
1886 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1887 		err = got_error_from_errno("fseeko");
1888 		goto done;
1889 	}
1890 
1891 	ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
1892 	if (ndeltify > 0) {
1893 		deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
1894 		if (deltify.meta == NULL) {
1895 			err = got_error_from_errno("calloc");
1896 			goto done;
1897 		}
1898 		deltify.metasz = ndeltify;
1899 
1900 		err = got_object_idset_for_each(idset, add_meta_idset_cb,
1901 		    &deltify);
1902 		if (err)
1903 			goto done;
1904 		if (deltify.nmeta > 0) {
1905 			err = pick_deltas(deltify.meta, deltify.nmeta,
1906 			    ncolored, nfound, ntrees, nours, reuse.nmeta,
1907 			    delta_cache, repo, progress_cb, progress_arg, rl,
1908 			    cancel_cb, cancel_arg);
1909 			if (err)
1910 				goto done;
1911 		}
1912 	}
1913 
1914 	if (fflush(delta_cache) == EOF) {
1915 		err = got_error_from_errno("fflush");
1916 		goto done;
1917 	}
1918 	err = genpack(packsha1, packfd, delta_cache, deltify.meta,
1919 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1920 	    nours, repo, progress_cb, progress_arg, rl,
1921 	    cancel_cb, cancel_arg);
1922 	if (err)
1923 		goto done;
1924 done:
1925 	free_nmeta(deltify.meta, deltify.nmeta);
1926 	free_nmeta(reuse.meta, reuse.nmeta);
1927 	got_object_idset_free(idset);
1928 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1929 		err = got_error_from_errno("close");
1930 	return err;
1931 }