Got Mirror

Blob

Date:: Sun Feb 13 00:12:04 2022 UTC
Message:: fix pack.sh test failure from reuse-deltas patch by tweaking progress output
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2020 Ori Bernstein
3  * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include <sys/types.h>
19 #include <sys/queue.h>
20 #include <sys/tree.h>
21 #include <sys/uio.h>
22 #include <sys/stat.h>
23 #include <sys/time.h>
24 
25 #include <stdint.h>
26 #include <imsg.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <sha1.h>
31 #include <time.h>
32 #include <limits.h>
33 #include <zlib.h>
34 
35 #include "got_error.h"
36 #include "got_cancel.h"
37 #include "got_object.h"
38 #include "got_path.h"
39 #include "got_reference.h"
40 #include "got_repository_admin.h"
41 #include "got_opentemp.h"
42 
43 #include "got_lib_deltify.h"
44 #include "got_lib_delta.h"
45 #include "got_lib_object.h"
46 #include "got_lib_object_idset.h"
47 #include "got_lib_object_cache.h"
48 #include "got_lib_deflate.h"
49 #include "got_lib_pack.h"
50 #include "got_lib_privsep.h"
51 #include "got_lib_repository.h"
52 #include "got_lib_ratelimit.h"
53 
54 #ifndef MIN
55 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
56 #endif
57 
58 #ifndef MAX
59 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
60 #endif
61 
62 struct got_pack_meta {
63 	struct got_object_id id;
64 	char	*path;
65 	int	obj_type;
66 	off_t	size;
67 	time_t	mtime;
68 
69 	/* The best delta we picked */
70 	struct got_pack_meta *head;
71 	struct got_pack_meta *prev;
72 	unsigned char *delta_buf; /* if not encoded in delta cache file */
73 	off_t	delta_offset;	/* offset in delta cache file */
74 	off_t	delta_len;	/* encoded delta length */
75 	int	nchain;
76 
77 	int	have_reused_delta;
78 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
79 	struct got_object_id *base_obj_id;
80 
81 	/* Only used for delta window */
82 	struct got_delta_table *dtab;
83 
84 	/* Only used for writing offset deltas */
85 	off_t	off;
86 };
87 
88 struct got_pack_metavec {
89 	struct got_pack_meta **meta;
90 	int nmeta;
91 	int metasz;
92 };
93 
94 static const struct got_error *
95 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
96     const char *path, int obj_type, time_t mtime)
97 {
98 	const struct got_error *err = NULL;
99 	struct got_pack_meta *m;
100 
101 	*new = NULL;
102 
103 	m = calloc(1, sizeof(*m));
104 	if (m == NULL)
105 		return got_error_from_errno("calloc");
106 
107 	memcpy(&m->id, id, sizeof(m->id));
108 
109 	m->path = strdup(path);
110 	if (m->path == NULL) {
111 		err = got_error_from_errno("strdup");
112 		free(m);
113 		return err;
114 	}
115 
116 	m->obj_type = obj_type;
117 	m->mtime = mtime;
118 	*new = m;
119 	return NULL;
120 }
121 
122 static void
123 clear_meta(struct got_pack_meta *meta)
124 {
125 	if (meta == NULL)
126 		return;
127 	free(meta->path);
128 	meta->path = NULL;
129 	free(meta->delta_buf);
130 	meta->delta_buf = NULL;
131 	free(meta->base_obj_id);
132 	meta->base_obj_id = NULL;
133 }
134 
135 static void
136 free_nmeta(struct got_pack_meta **meta, int nmeta)
137 {
138 	int i;
139 
140 	for (i = 0; i < nmeta; i++)
141 		clear_meta(meta[i]);
142 	free(meta);
143 }
144 
145 static int
146 delta_order_cmp(const void *pa, const void *pb)
147 {
148 	struct got_pack_meta *a, *b;
149 	int cmp;
150 
151 	a = *(struct got_pack_meta **)pa;
152 	b = *(struct got_pack_meta **)pb;
153 
154 	if (a->obj_type != b->obj_type)
155 		return a->obj_type - b->obj_type;
156 	cmp = strcmp(a->path, b->path);
157 	if (cmp != 0)
158 		return cmp;
159 	if (a->mtime != b->mtime)
160 		return a->mtime - b->mtime;
161 	return got_object_id_cmp(&a->id, &b->id);
162 }
163 
164 static off_t
165 delta_size(struct got_delta_instruction *deltas, int ndeltas)
166 {
167 	int i;
168 	off_t size = 32;
169 	for (i = 0; i < ndeltas; i++) {
170 		if (deltas[i].copy)
171 			size += GOT_DELTA_SIZE_SHIFT;
172 		else
173 			size += deltas[i].len + 1;
174 	}
175 	return size;
176 }
177 
178 static const struct got_error *
179 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
180 {
181 	char *n;
182 
183 	if (*len + nseg >= *sz) {
184 		while (*len + nseg >= *sz)
185 			*sz += *sz / 2;
186 		n = realloc(*p, *sz);
187 		if (n == NULL)
188 			return got_error_from_errno("realloc");
189 		*p = n;
190 	}
191 	memcpy(*p + *len, seg, nseg);
192 	*len += nseg;
193 	return NULL;
194 }
195 
196 static const struct got_error *
197 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
198     struct got_delta_instruction *deltas, int ndeltas,
199     off_t delta_size, off_t base_size)
200 {
201 	const struct got_error *err;
202 	unsigned char buf[16], *bp;
203 	int i, j;
204 	size_t len = 0;
205 	off_t n;
206 	struct got_delta_instruction *d;
207 
208 	m->delta_buf = malloc(delta_size);
209 	if (m->delta_buf == NULL)
210 		return got_error_from_errno("calloc");
211 
212 	/* base object size */
213 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
214 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
215 	for (i = 1; n > 0; i++) {
216 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
217 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
218 		n >>= GOT_DELTA_SIZE_SHIFT;
219 	}
220 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
221 	if (err)
222 		return err;
223 
224 	/* target object size */
225 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
226 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
227 	for (i = 1; n > 0; i++) {
228 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
229 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
230 		n >>= GOT_DELTA_SIZE_SHIFT;
231 	}
232 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
233 	if (err)
234 		return err;
235 
236 	for (j = 0; j < ndeltas; j++) {
237 		d = &deltas[j];
238 		if (d->copy) {
239 			n = d->offset;
240 			bp = &buf[1];
241 			buf[0] = GOT_DELTA_BASE_COPY;
242 			for (i = 0; i < 4; i++) {
243 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
244 				buf[0] |= 1 << i;
245 				*bp++ = n & 0xff;
246 				n >>= 8;
247 				if (n == 0)
248 					break;
249 			}
250 
251 			n = d->len;
252 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
253 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
254 				for (i = 0; i < 3 && n > 0; i++) {
255 					buf[0] |= 1 << (i + 4);
256 					*bp++ = n & 0xff;
257 					n >>= 8;
258 				}
259 			}
260 			err = append(&m->delta_buf, &len, &delta_size,
261 			    buf, bp - buf);
262 			if (err)
263 				return err;
264 		} else if (o->f == NULL) {
265 			n = 0;
266 			while (n != d->len) {
267 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
268 				err = append(&m->delta_buf, &len, &delta_size,
269 				    buf, 1);
270 				if (err)
271 					return err;
272 				err = append(&m->delta_buf, &len, &delta_size,
273 				    o->data + o->hdrlen + d->offset + n,
274 				    buf[0]);
275 				if (err)
276 					return err;
277 				n += buf[0];
278 			}
279 		} else {
280 			char content[128];
281 			size_t r;
282 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
283 				return got_error_from_errno("fseeko");
284 			n = 0;
285 			while (n != d->len) {
286 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
287 				err = append(&m->delta_buf, &len, &delta_size,
288 				    buf, 1);
289 				if (err)
290 					return err;
291 				r = fread(content, 1, buf[0], o->f);
292 				if (r != buf[0])
293 					return got_ferror(o->f, GOT_ERR_IO);
294 				err = append(&m->delta_buf, &len, &delta_size,
295 				    content, buf[0]);
296 				if (err)
297 					return err;
298 				n += buf[0];
299 			}
300 		}
301 	}
302 
303 	m->delta_len = len;
304 	return NULL;
305 }
306 
307 static const struct got_error *
308 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
309     struct got_delta_instruction *deltas, int ndeltas,
310     off_t base_size, FILE *f)
311 {
312 	unsigned char buf[16], *bp;
313 	int i, j;
314 	off_t n;
315 	size_t w;
316 	struct got_delta_instruction *d;
317 
318 	/* base object size */
319 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
320 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
321 	for (i = 1; n > 0; i++) {
322 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
323 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
324 		n >>= GOT_DELTA_SIZE_SHIFT;
325 	}
326 	w = fwrite(buf, 1, i, f);
327 	if (w != i)
328 		return got_ferror(f, GOT_ERR_IO);
329 
330 	/* target object size */
331 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
332 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
333 	for (i = 1; n > 0; i++) {
334 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
335 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
336 		n >>= GOT_DELTA_SIZE_SHIFT;
337 	}
338 	w = fwrite(buf, 1, i, f);
339 	if (w != i)
340 		return got_ferror(f, GOT_ERR_IO);
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 
357 			n = d->len;
358 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
359 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
360 				for (i = 0; i < 3 && n > 0; i++) {
361 					buf[0] |= 1 << (i + 4);
362 					*bp++ = n & 0xff;
363 					n >>= 8;
364 				}
365 			}
366 			w = fwrite(buf, 1, bp - buf, f);
367 			if (w != bp - buf)
368 				return got_ferror(f, GOT_ERR_IO);
369 		} else if (o->f == NULL) {
370 			n = 0;
371 			while (n != d->len) {
372 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
373 				w = fwrite(buf, 1, 1, f);
374 				if (w != 1)
375 					return got_ferror(f, GOT_ERR_IO);
376 				w = fwrite(o->data + o->hdrlen + d->offset + n,
377 				    1, buf[0], f);
378 				if (w != buf[0])
379 					return got_ferror(f, GOT_ERR_IO);
380 				n += buf[0];
381 			}
382 		} else {
383 			char content[128];
384 			size_t r;
385 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
386 				return got_error_from_errno("fseeko");
387 			n = 0;
388 			while (n != d->len) {
389 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
390 				w = fwrite(buf, 1, 1, f);
391 				if (w != 1)
392 					return got_ferror(f, GOT_ERR_IO);
393 				r = fread(content, 1, buf[0], o->f);
394 				if (r != buf[0])
395 					return got_ferror(o->f, GOT_ERR_IO);
396 				w = fwrite(content, 1, buf[0], f);
397 				if (w != buf[0])
398 					return got_ferror(f, GOT_ERR_IO);
399 				n += buf[0];
400 			}
401 		}
402 	}
403 
404 	m->delta_len = ftello(f) - m->delta_offset;
405 	return NULL;
406 }
407 
408 static const struct got_error *
409 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
410     struct got_ratelimit *rl, off_t packfile_size, int ncommits,
411     int nobj_total, int obj_deltify, int nobj_written)
412 {
413 	const struct got_error *err;
414 	int elapsed;
415 
416 	if (progress_cb == NULL)
417 		return NULL;
418 
419 	err = got_ratelimit_check(&elapsed, rl);
420 	if (err || !elapsed)
421 		return err;
422 
423 	return progress_cb(progress_arg, packfile_size, ncommits,
424 	    nobj_total, obj_deltify, nobj_written);
425 }
426 
427 static const struct got_error *
428 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
429 {
430 	if (v->nmeta == v->metasz){
431 		size_t newsize = 2 * v->metasz;
432 		struct got_pack_meta **new;
433 		new = reallocarray(v->meta, newsize, sizeof(*new));
434 		if (new == NULL)
435 			return got_error_from_errno("reallocarray");
436 		v->meta = new;
437 		v->metasz = newsize; 
438 	}
439 
440 	v->meta[v->nmeta++] = m;
441 	return NULL;
442 }
443 
444 static const struct got_error *
445 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
446     struct got_object_idset *idset, struct got_pack *pack,
447     struct got_packidx *packidx, int delta_cache_fd,
448     struct got_repository *repo)
449 {
450 	const struct got_error *err = NULL;
451 	struct got_pack_meta *base = NULL;
452 	struct got_object_id *base_obj_id = NULL;
453 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
454 	uint64_t base_size, result_size;
455 
456 	if (m->have_reused_delta)
457 		return NULL;
458 
459 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
460 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
461 	    packidx, idx, &m->id, repo);
462 	if (err)
463 		return err;
464 
465 	if (delta_offset + delta_len < delta_offset)
466 		return got_error(GOT_ERR_BAD_PACKFILE);
467 
468 	base = got_object_idset_get(idset, base_obj_id);
469 	if (base == NULL)
470 		goto done;
471 
472 	m->delta_len = delta_len;
473 	m->delta_offset = delta_cache_offset;
474 	m->prev = base;
475 	m->size = result_size;
476 	m->have_reused_delta = 1;
477 	m->reused_delta_offset = delta_offset;
478 	m->base_obj_id = base_obj_id;
479 	base_obj_id = NULL;
480 	err = add_meta(m, v);
481 done:
482 	free(base_obj_id);
483 	return err;
484 }
485 
486 static const struct got_error *
487 find_pack_for_reuse(struct got_packidx **best_packidx,
488     struct got_repository *repo)
489 {
490 	const struct got_error *err;
491 	struct got_pathlist_head packidx_paths;
492 	struct got_pathlist_entry *pe;
493 	const char *best_packidx_path = NULL;
494 	int nobj_max = 0;
495 
496 	TAILQ_INIT(&packidx_paths);
497 	*best_packidx = NULL;
498 
499 	err = got_repo_list_packidx(&packidx_paths, repo);
500 	if (err)
501 		return err;
502 
503 	TAILQ_FOREACH(pe, &packidx_paths, entry) {
504 		const char *path_packidx = pe->path;
505 		struct got_packidx *packidx;
506 		int nobj;
507 
508 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
509 		if (err)
510 			break;
511 
512 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
513 		if (nobj > nobj_max) {
514 			best_packidx_path = path_packidx;
515 			nobj_max = nobj;
516 		}
517 	}
518 
519 	if (best_packidx_path) {
520 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
521 		    repo);
522 	}
523 
524 	TAILQ_FOREACH(pe, &packidx_paths, entry)
525 		free((void *)pe->path);
526 	got_pathlist_free(&packidx_paths);
527 	return err;
528 }
529 
530 struct search_deltas_arg {
531 	struct got_packidx *packidx;
532 	struct got_pack *pack;
533 	struct got_object_idset *idset;
534 	struct got_pack_metavec *v;
535 	int delta_cache_fd;
536 	struct got_repository *repo;
537 	got_pack_progress_cb progress_cb;
538 	void *progress_arg;
539 	struct got_ratelimit *rl;
540 	got_cancel_cb cancel_cb;
541 	void *cancel_arg;
542 	int ncommits;
543 };
544 
545 static const struct got_error *
546 search_delta_for_object(struct got_object_id *id, void *data, void *arg)
547 {
548 	const struct got_error *err;
549 	struct got_pack_meta *m = data;
550 	struct search_deltas_arg *a = arg;
551 	int obj_idx;
552 	struct got_object *obj = NULL;
553 
554 	if (a->cancel_cb) {
555 		err = (*a->cancel_cb)(a->cancel_arg);
556 		if (err)
557 			return err;
558 	}
559 
560 	if (!got_repo_check_packidx_bloom_filter(a->repo,
561 	    a->packidx->path_packidx, id))
562 		return NULL;
563 
564 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
565 	if (obj_idx == -1)
566 		return NULL;
567 
568 	/* TODO:
569 	 * Opening and closing an object just to check its flags
570 	 * is a bit expensive. We could have an imsg which requests
571 	 * plain type/size information for an object without doing
572 	 * work such as traversing the object's entire delta chain
573 	 * to find the base object type, and other such info which
574 	 * we don't really need here.
575 	 */
576 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
577 	    a->packidx, obj_idx, a->repo);
578 	if (err)
579 		return err;
580 
581 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
582 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
583 		    a->delta_cache_fd, a->repo);
584 		if (err)
585 			goto done;
586 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
587 		    0L, a->ncommits, got_object_idset_num_elements(a->idset),
588 		    a->v->nmeta, 0);
589 	}
590 done:
591 	got_object_close(obj);
592 	return err;
593 }
594 
595 static const struct got_error *
596 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
597     int delta_cache_fd, int ncommits, struct got_repository *repo,
598     got_pack_progress_cb progress_cb, void *progress_arg,
599     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
600 {
601 	const struct got_error *err = NULL;
602 	char *path_packfile = NULL;
603 	struct got_packidx *packidx;
604 	struct got_pack *pack;
605 	struct search_deltas_arg sda;
606 
607 	err = find_pack_for_reuse(&packidx, repo);
608 	if (err)
609 		return err;
610 
611 	if (packidx == NULL)
612 		return NULL;
613 
614 	err = got_packidx_get_packfile_path(&path_packfile,
615 	    packidx->path_packidx);
616 	if (err)
617 		return err;
618 
619 	pack = got_repo_get_cached_pack(repo, path_packfile);
620 	if (pack == NULL) {
621 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
622 		if (err)
623 			goto done;
624 	}
625 
626 	sda.packidx = packidx;
627 	sda.pack = pack;
628 	sda.idset = idset;
629 	sda.v = v;
630 	sda.delta_cache_fd = delta_cache_fd;
631 	sda.repo = repo;
632 	sda.progress_cb = progress_cb;
633 	sda.progress_arg = progress_arg;
634 	sda.rl = rl;
635 	sda.cancel_cb = cancel_cb;
636 	sda.cancel_arg = cancel_arg;
637 	sda.ncommits = ncommits;
638 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
639 done:
640 	free(path_packfile);
641 	return err;
642 }
643 
644 static const struct got_error *
645 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncommits,
646     int nreused, FILE *delta_cache, struct got_repository *repo,
647     got_pack_progress_cb progress_cb, void *progress_arg,
648     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
649 {
650 	const struct got_error *err = NULL;
651 	struct got_pack_meta *m = NULL, *base = NULL;
652 	struct got_raw_object *raw = NULL, *base_raw = NULL;
653 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
654 	int i, j, ndeltas, best_ndeltas;
655 	off_t size, best_size;
656 	const int max_base_candidates = 3;
657 	size_t delta_memsize = 0;
658 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
659 	int outfd = -1;
660 
661 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
662 	for (i = 0; i < nmeta; i++) {
663 		if (cancel_cb) {
664 			err = (*cancel_cb)(cancel_arg);
665 			if (err)
666 				break;
667 		}
668 		err = report_progress(progress_cb, progress_arg, rl,
669 		    0L, ncommits, nreused + nmeta, nreused + i, 0);
670 		if (err)
671 			goto done;
672 		m = meta[i];
673 
674 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
675 		    m->obj_type == GOT_OBJ_TYPE_TAG)
676 			continue;
677 
678 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
679 		if (err)
680 			goto done;
681 		m->size = raw->size;
682 
683 		if (raw->f == NULL) {
684 			err = got_deltify_init_mem(&m->dtab, raw->data,
685 			    raw->hdrlen, raw->size + raw->hdrlen);
686 		} else {
687 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
688 			    raw->size + raw->hdrlen);
689 		}
690 		if (err)
691 			goto done;
692 
693 		if (i > max_base_candidates) {
694 			struct got_pack_meta *n = NULL;
695 			n = meta[i - (max_base_candidates + 1)];
696 			got_deltify_free(n->dtab);
697 			n->dtab = NULL;
698 		}
699 
700 		best_size = raw->size;
701 		best_ndeltas = 0;
702 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
703 			if (cancel_cb) {
704 				err = (*cancel_cb)(cancel_arg);
705 				if (err)
706 					goto done;
707 			}
708 			base = meta[j];
709 			/* long chains make unpacking slow, avoid such bases */
710 			if (base->nchain >= 128 ||
711 			    base->obj_type != m->obj_type)
712 				continue;
713 
714 			err = got_object_raw_open(&base_raw, &outfd, repo,
715 			    &base->id);
716 			if (err)
717 				goto done;
718 
719 			if (raw->f == NULL && base_raw->f == NULL) {
720 				err = got_deltify_mem_mem(&deltas, &ndeltas,
721 				    raw->data, raw->hdrlen,
722 				    raw->size + raw->hdrlen,
723 				    base->dtab, base_raw->data,
724 				    base_raw->hdrlen,
725 				    base_raw->size + base_raw->hdrlen);
726 			} else if (raw->f == NULL) {
727 				err = got_deltify_mem_file(&deltas, &ndeltas,
728 				    raw->data, raw->hdrlen,
729 				    raw->size + raw->hdrlen,
730 				    base->dtab, base_raw->f,
731 				    base_raw->hdrlen,
732 				    base_raw->size + base_raw->hdrlen);
733 			} else if (base_raw->f == NULL) {
734 				err = got_deltify_file_mem(&deltas, &ndeltas,
735 				    raw->f, raw->hdrlen,
736 				    raw->size + raw->hdrlen,
737 				    base->dtab, base_raw->data,
738 				    base_raw->hdrlen,
739 				    base_raw->size + base_raw->hdrlen);
740 			} else {
741 				err = got_deltify(&deltas, &ndeltas,
742 				    raw->f, raw->hdrlen,
743 				    raw->size + raw->hdrlen,
744 				    base->dtab, base_raw->f, base_raw->hdrlen,
745 				    base_raw->size + base_raw->hdrlen);
746 			}
747 			got_object_raw_close(base_raw);
748 			base_raw = NULL;
749 			if (err)
750 				goto done;
751 
752 			size = delta_size(deltas, ndeltas);
753 			if (size + 32 < best_size){
754 				/*
755 				 * if we already picked a best delta,
756 				 * replace it.
757 				 */
758 				best_size = size;
759 				free(best_deltas);
760 				best_deltas = deltas;
761 				best_ndeltas = ndeltas;
762 				deltas = NULL;
763 				m->nchain = base->nchain + 1;
764 				m->prev = base;
765 				m->head = base->head;
766 				if (m->head == NULL)
767 					m->head = base;
768 			} else {
769 				free(deltas);
770 				deltas = NULL;
771 				ndeltas = 0;
772 			}
773 		}
774 
775 		if (best_ndeltas > 0) {
776 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
777 			    delta_memsize + best_size <= max_delta_memsize) {
778 				delta_memsize += best_size;
779 				err = encode_delta_in_mem(m, raw, best_deltas,
780 				    best_ndeltas, best_size, m->prev->size);
781 			} else {
782 				m->delta_offset = ftello(delta_cache);
783 				/*
784 				 * TODO:
785 				 * Storing compressed delta data in the delta
786 				 * cache file would probably be more efficient
787 				 * than writing uncompressed delta data here
788 				 * and compressing it while writing the pack
789 				 * file. This would also allow for reusing
790 				 * deltas in their compressed form.
791 				 */
792 				err = encode_delta(m, raw, best_deltas,
793 				    best_ndeltas, m->prev->size, delta_cache);
794 			}
795 			free(best_deltas);
796 			best_deltas = NULL;
797 			best_ndeltas = 0;
798 			if (err)
799 				goto done;
800 		}
801 
802 		got_object_raw_close(raw);
803 		raw = NULL;
804 	}
805 done:
806 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
807 		got_deltify_free(meta[i]->dtab);
808 		meta[i]->dtab = NULL;
809 	}
810 	if (raw)
811 		got_object_raw_close(raw);
812 	if (base_raw)
813 		got_object_raw_close(base_raw);
814 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
815 		err = got_error_from_errno("close");
816 	free(deltas);
817 	free(best_deltas);
818 	return err;
819 }
820 
821 static const struct got_error *
822 search_packidx(int *found, struct got_object_id *id,
823     struct got_repository *repo)
824 {
825 	const struct got_error *err = NULL;
826 	struct got_packidx *packidx = NULL;
827 	int idx;
828 
829 	*found = 0;
830 
831 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
832 	if (err == NULL)
833 		*found = 1; /* object is already packed */
834 	else if (err->code == GOT_ERR_NO_OBJ)
835 		err = NULL;
836 	return err;
837 }
838 
839 static const int obj_types[] = {
840 	GOT_OBJ_TYPE_ANY,
841 	GOT_OBJ_TYPE_COMMIT,
842 	GOT_OBJ_TYPE_TREE,
843 	GOT_OBJ_TYPE_BLOB,
844 	GOT_OBJ_TYPE_TAG,
845 	GOT_OBJ_TYPE_OFFSET_DELTA,
846 	GOT_OBJ_TYPE_REF_DELTA
847 };
848 
849 static const struct got_error *
850 add_object(int want_meta, struct got_object_idset *idset,
851     struct got_object_id *id, const char *path, int obj_type,
852     time_t mtime, int loose_obj_only, struct got_repository *repo)
853 {
854 	const struct got_error *err;
855 	struct got_pack_meta *m = NULL;
856 
857 	if (loose_obj_only) {
858 		int is_packed;
859 		err = search_packidx(&is_packed, id, repo);
860 		if (err)
861 			return err;
862 		if (is_packed)
863 			return NULL;
864 	}
865 
866 	if (want_meta) {
867 		err = alloc_meta(&m, id, path, obj_type, mtime);
868 		if (err)
869 			return err;
870 	}
871 
872 	return got_object_idset_add(idset, id, m);
873 }
874 
875 static const struct got_error *
876 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
877     struct got_object_idset *idset, struct got_object_id *tree_id,
878     const char *dpath, time_t mtime, struct got_repository *repo,
879     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
880 {
881 	const struct got_error *err;
882 	struct got_tree_object *tree;
883 	char *p = NULL;
884 	int i;
885 
886 	err = got_object_open_as_tree(&tree, repo, tree_id);
887 	if (err)
888 		return err;
889 
890 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
891 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
892 		struct got_object_id *id = got_tree_entry_get_id(e);
893 		mode_t mode = got_tree_entry_get_mode(e);
894 
895 		if (cancel_cb) {
896 			err = (*cancel_cb)(cancel_arg);
897 			if (err)
898 				break;
899 		}
900 
901 		if (got_object_tree_entry_is_submodule(e) ||
902 		    got_object_idset_contains(idset, id))
903 			continue;
904 		
905 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
906 		    got_tree_entry_get_name(e)) == -1) {
907 			err = got_error_from_errno("asprintf");
908 			break;
909 		}
910 
911 		if (S_ISDIR(mode)) {
912 			struct got_object_qid *qid;
913 			err = got_object_qid_alloc(&qid, id);
914 			if (err)
915 				break;
916 			STAILQ_INSERT_TAIL(ids, qid, entry);
917 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
918 			err = add_object(want_meta, idset, id, p,
919 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo);
920 			if (err)
921 				break;
922 		}
923 		free(p);
924 		p = NULL;
925 	}
926 
927 	got_object_tree_close(tree);
928 	free(p);
929 	return err;
930 }
931 
932 static const struct got_error *
933 load_tree(int want_meta, struct got_object_idset *idset,
934     struct got_object_id *tree_id, const char *dpath, time_t mtime,
935     int loose_obj_only, struct got_repository *repo,
936     got_cancel_cb cancel_cb, void *cancel_arg)
937 {
938 	const struct got_error *err = NULL;
939 	struct got_object_id_queue tree_ids;
940 	struct got_object_qid *qid;
941 
942 	if (got_object_idset_contains(idset, tree_id))
943 		return NULL;
944 
945 	err = got_object_qid_alloc(&qid, tree_id);
946 	if (err)
947 		return err;
948 
949 	STAILQ_INIT(&tree_ids);
950 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
951 
952 	while (!STAILQ_EMPTY(&tree_ids)) {
953 		if (cancel_cb) {
954 			err = (*cancel_cb)(cancel_arg);
955 			if (err)
956 				break;
957 		}
958 
959 		qid = STAILQ_FIRST(&tree_ids);
960 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
961 
962 		if (got_object_idset_contains(idset, qid->id)) {
963 			got_object_qid_free(qid);
964 			continue;
965 		}
966 
967 		err = add_object(want_meta, idset, qid->id, dpath,
968 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo);
969 		if (err) {
970 			got_object_qid_free(qid);
971 			break;
972 		}
973 
974 		err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
975 		    dpath, mtime, repo, loose_obj_only, cancel_cb, cancel_arg);
976 		got_object_qid_free(qid);
977 		if (err)
978 			break;
979 	}
980 
981 	got_object_id_queue_free(&tree_ids);
982 	return err;
983 }
984 
985 static const struct got_error *
986 load_commit(int want_meta, struct got_object_idset *idset,
987     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
988     got_cancel_cb cancel_cb, void *cancel_arg)
989 {
990 	const struct got_error *err;
991 	struct got_commit_object *commit;
992 
993 	if (got_object_idset_contains(idset, id))
994 		return NULL;
995 
996 	if (loose_obj_only) {
997 		int is_packed;
998 		err = search_packidx(&is_packed, id, repo);
999 		if (err)
1000 			return err;
1001 		if (is_packed)
1002 			return NULL;
1003 	}
1004 
1005 	err = got_object_open_as_commit(&commit, repo, id);
1006 	if (err)
1007 		return err;
1008 
1009 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1010 	    got_object_commit_get_committer_time(commit),
1011 	    loose_obj_only, repo);
1012 	if (err)
1013 		goto done;
1014 
1015 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1016 	    "", got_object_commit_get_committer_time(commit),
1017 	    loose_obj_only, repo, cancel_cb, cancel_arg);
1018 done:
1019 	got_object_commit_close(commit);
1020 	return err;
1021 }
1022 
1023 static const struct got_error *
1024 load_tag(int want_meta, struct got_object_idset *idset,
1025     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1026     got_cancel_cb cancel_cb, void *cancel_arg)
1027 {
1028 	const struct got_error *err;
1029 	struct got_tag_object *tag = NULL;
1030 
1031 	if (got_object_idset_contains(idset, id))
1032 		return NULL;
1033 
1034 	if (loose_obj_only) {
1035 		int is_packed;
1036 		err = search_packidx(&is_packed, id, repo);
1037 		if (err)
1038 			return err;
1039 		if (is_packed)
1040 			return NULL;
1041 	}
1042 
1043 	err = got_object_open_as_tag(&tag, repo, id);
1044 	if (err)
1045 		return err;
1046 
1047 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1048 	    got_object_tag_get_tagger_time(tag),
1049 	    loose_obj_only, repo);
1050 	if (err)
1051 		goto done;
1052 
1053 	switch (got_object_tag_get_object_type(tag)) {
1054 	case GOT_OBJ_TYPE_COMMIT:
1055 		err = load_commit(want_meta, idset,
1056 		    got_object_tag_get_object_id(tag), repo,
1057 		    loose_obj_only, cancel_cb, cancel_arg);
1058 		break;
1059 	case GOT_OBJ_TYPE_TREE:
1060 		err = load_tree(want_meta, idset,
1061 		    got_object_tag_get_object_id(tag), "",
1062 		    got_object_tag_get_tagger_time(tag),
1063 		    loose_obj_only, repo, cancel_cb, cancel_arg);
1064 		break;
1065 	default:
1066 		break;
1067 	}
1068 
1069 done:
1070 	got_object_tag_close(tag);
1071 	return err;
1072 }
1073 
1074 enum findtwixt_color {
1075 	COLOR_KEEP = 0,
1076 	COLOR_DROP,
1077 	COLOR_BLANK,
1078 };
1079 static const int findtwixt_colors[] = {
1080 	COLOR_KEEP,
1081 	COLOR_DROP,
1082 	COLOR_BLANK
1083 };
1084 
1085 static const struct got_error *
1086 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1087     int color, struct got_repository *repo)
1088 {
1089 	const struct got_error *err;
1090 	struct got_object_qid *qid;
1091 
1092 	err = got_object_qid_alloc(&qid, id);
1093 	if (err)
1094 		return err;
1095 
1096 	STAILQ_INSERT_TAIL(ids, qid, entry);
1097 	qid->data = (void *)&findtwixt_colors[color];
1098 	return NULL;
1099 }
1100 
1101 static const struct got_error *
1102 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1103     struct got_object_id *id, struct got_repository *repo,
1104     got_cancel_cb cancel_cb, void *cancel_arg)
1105 {
1106 	const struct got_error *err = NULL;
1107 	struct got_commit_object *commit;
1108 	const struct got_object_id_queue *parents;
1109 	struct got_object_id_queue ids;
1110 	struct got_object_qid *qid;
1111 
1112 	STAILQ_INIT(&ids);
1113 
1114 	err = got_object_qid_alloc(&qid, id);
1115 	if (err)
1116 		return err;
1117 	STAILQ_INSERT_HEAD(&ids, qid, entry);
1118 
1119 	while (!STAILQ_EMPTY(&ids)) {
1120 		if (cancel_cb) {
1121 			err = (*cancel_cb)(cancel_arg);
1122 			if (err)
1123 				break;
1124 		}
1125 
1126 		qid = STAILQ_FIRST(&ids);
1127 		STAILQ_REMOVE_HEAD(&ids, entry);
1128 
1129 		if (got_object_idset_contains(drop, qid->id)) {
1130 			got_object_qid_free(qid);
1131 			continue;
1132 		}
1133 
1134 		err = got_object_idset_add(drop, qid->id,
1135 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1136 		if (err) {
1137 			got_object_qid_free(qid);
1138 			break;
1139 		}
1140 
1141 		if (!got_object_idset_contains(keep, qid->id)) {
1142 			got_object_qid_free(qid);
1143 			continue;
1144 		}
1145 
1146 		err = got_object_open_as_commit(&commit, repo, qid->id);
1147 		got_object_qid_free(qid);
1148 		if (err)
1149 			break;
1150 
1151 		parents = got_object_commit_get_parent_ids(commit);
1152 		if (parents) {
1153 			err = got_object_id_queue_copy(parents, &ids);
1154 			if (err) {
1155 				got_object_commit_close(commit);
1156 				break;
1157 			}
1158 		}
1159 		got_object_commit_close(commit);
1160 	}
1161 
1162 	got_object_id_queue_free(&ids);
1163 	return err;
1164 }
1165 
1166 struct append_id_arg {
1167 	struct got_object_id **array;
1168 	int idx;
1169 };
1170 
1171 static const struct got_error *
1172 append_id(struct got_object_id *id, void *data, void *arg)
1173 {
1174 	struct append_id_arg *a = arg;
1175 
1176 	a->array[a->idx] = got_object_id_dup(id);
1177 	if (a->array[a->idx] == NULL)
1178 		return got_error_from_errno("got_object_id_dup");
1179 
1180 	a->idx++;
1181 	return NULL;
1182 }
1183 
1184 static const struct got_error *
1185 findtwixt(struct got_object_id ***res, int *nres,
1186     struct got_object_id **head, int nhead,
1187     struct got_object_id **tail, int ntail,
1188     struct got_repository *repo,
1189     got_cancel_cb cancel_cb, void *cancel_arg)
1190 {
1191 	const struct got_error *err = NULL;
1192 	struct got_object_id_queue ids;
1193 	struct got_object_idset *keep, *drop;
1194 	struct got_object_qid *qid;
1195 	int i, ncolor, nkeep, obj_type;
1196 
1197 	STAILQ_INIT(&ids);
1198 	*res = NULL;
1199 	*nres = 0;
1200 
1201 	keep = got_object_idset_alloc();
1202 	if (keep == NULL)
1203 		return got_error_from_errno("got_object_idset_alloc");
1204 
1205 	drop = got_object_idset_alloc();
1206 	if (drop == NULL) {
1207 		err = got_error_from_errno("got_object_idset_alloc");
1208 		goto done;
1209 	}
1210 
1211 	for (i = 0; i < nhead; i++) {
1212 		struct got_object_id *id = head[i];
1213 		if (id == NULL)
1214 			continue;
1215 		err = got_object_get_type(&obj_type, repo, id);
1216 		if (err)
1217 			return err;
1218 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1219 			continue;
1220 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1221 		if (err)
1222 			goto done;
1223 	}		
1224 	for (i = 0; i < ntail; i++) {
1225 		struct got_object_id *id = tail[i];
1226 		if (id == NULL)
1227 			continue;
1228 		err = got_object_get_type(&obj_type, repo, id);
1229 		if (err)
1230 			return err;
1231 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1232 			continue;
1233 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1234 		if (err)
1235 			goto done;
1236 	}
1237 
1238 	while (!STAILQ_EMPTY(&ids)) {
1239 		int qcolor;
1240 		qid = STAILQ_FIRST(&ids);
1241 		qcolor = *((int *)qid->data);
1242 
1243 		if (got_object_idset_contains(drop, qid->id))
1244 			ncolor = COLOR_DROP;
1245 		else if (got_object_idset_contains(keep, qid->id))
1246 			ncolor = COLOR_KEEP;
1247 		else
1248 			ncolor = COLOR_BLANK;
1249 
1250 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1251 		    qcolor == COLOR_KEEP)) {
1252 			STAILQ_REMOVE_HEAD(&ids, entry);
1253 			got_object_qid_free(qid);
1254 			continue;
1255 		}
1256 
1257 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1258 			err = drop_commit(keep, drop, qid->id, repo,
1259 			    cancel_cb, cancel_arg);
1260 			if (err)
1261 				goto done;
1262 		} else if (ncolor == COLOR_BLANK) {
1263 			struct got_commit_object *commit;
1264 			struct got_object_id *id;
1265 			const struct got_object_id_queue *parents;
1266 			struct got_object_qid *pid;
1267 
1268 			id = got_object_id_dup(qid->id);
1269 			if (id == NULL) {
1270 				err = got_error_from_errno("got_object_id_dup");
1271 				goto done;
1272 			}
1273 			if (qcolor == COLOR_KEEP)
1274 				err = got_object_idset_add(keep, id,
1275 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1276 			else
1277 				err = got_object_idset_add(drop, id,
1278 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1279 			if (err) {
1280 				free(id);
1281 				goto done;
1282 			}
1283 
1284 			err = got_object_open_as_commit(&commit, repo, id);
1285 			if (err) {
1286 				free(id);
1287 				goto done;
1288 			}
1289 			parents = got_object_commit_get_parent_ids(commit);
1290 			if (parents) {
1291 				STAILQ_FOREACH(pid, parents, entry) {
1292 					err = queue_commit_id(&ids, pid->id,
1293 					    qcolor, repo);
1294 					if (err) {
1295 						free(id);
1296 						goto done;
1297 					}
1298 				}
1299 			}
1300 			got_object_commit_close(commit);
1301 			commit = NULL;
1302 		} else {
1303 			/* should not happen */
1304 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1305 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1306 			goto done;
1307 		}
1308 
1309 		STAILQ_REMOVE_HEAD(&ids, entry);
1310 		got_object_qid_free(qid);
1311 	}
1312 
1313 	nkeep = got_object_idset_num_elements(keep);
1314 	if (nkeep > 0) {
1315 		struct append_id_arg arg;
1316 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1317 		if (arg.array == NULL) {
1318 			err = got_error_from_errno("calloc");
1319 			goto done;
1320 		}
1321 		arg.idx = 0;
1322 		err = got_object_idset_for_each(keep, append_id, &arg);
1323 		if (err) {
1324 			free(arg.array);
1325 			goto done;
1326 		}
1327 		*res = arg.array;
1328 		*nres = nkeep;
1329 	}
1330 done:
1331 	got_object_idset_free(keep);
1332 	got_object_idset_free(drop);
1333 	got_object_id_queue_free(&ids);
1334 	return err;
1335 }
1336 
1337 static const struct got_error *
1338 load_object_ids(struct got_object_idset *idset,
1339     struct got_object_id **theirs, int ntheirs,
1340     struct got_object_id **ours, int nours, struct got_repository *repo,
1341     int loose_obj_only, got_cancel_cb cancel_cb, void *cancel_arg)
1342 {
1343 	const struct got_error *err = NULL;
1344 	struct got_object_id **ids = NULL;
1345 	int i, nobj = 0, obj_type;
1346 
1347 	err = findtwixt(&ids, &nobj, ours, nours, theirs, ntheirs, repo,
1348 	    cancel_cb, cancel_arg);
1349 	if (err || nobj == 0)
1350 		goto done;
1351 
1352 	for (i = 0; i < ntheirs; i++) {
1353 		struct got_object_id *id = theirs[i];
1354 		if (id == NULL)
1355 			continue;
1356 		err = got_object_get_type(&obj_type, repo, id);
1357 		if (err)
1358 			return err;
1359 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1360 			continue;
1361 		err = load_commit(0, idset, id, repo,
1362 		    loose_obj_only, cancel_cb, cancel_arg);
1363 		if (err)
1364 			goto done;
1365 	}
1366 
1367 	for (i = 0; i < ntheirs; i++) {
1368 		struct got_object_id *id = theirs[i];
1369 		struct got_pack_meta *m;
1370 		if (id == NULL)
1371 			continue;
1372 		m = got_object_idset_get(idset, id);
1373 		if (m == NULL) {
1374 			err = got_object_get_type(&obj_type, repo, id);
1375 			if (err)
1376 				goto done;
1377 		} else
1378 			obj_type = m->obj_type;
1379 		if (obj_type != GOT_OBJ_TYPE_TAG)
1380 			continue;
1381 		err = load_tag(0, idset, id, repo,
1382 		    loose_obj_only, cancel_cb, cancel_arg);
1383 		if (err)
1384 			goto done;
1385 	}
1386 
1387 	for (i = 0; i < nobj; i++) {
1388 		err = load_commit(1, idset, ids[i], repo,
1389 		    loose_obj_only, cancel_cb, cancel_arg);
1390 		if (err)
1391 			goto done;
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, id, repo,
1409 		    loose_obj_only, cancel_cb, cancel_arg);
1410 		if (err)
1411 			goto done;
1412 	}
1413 done:
1414 	for (i = 0; i < nobj; i++) {
1415 		free(ids[i]);
1416 	}
1417 	free(ids);
1418 	return err;
1419 }
1420 
1421 const struct got_error *
1422 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1423 {
1424 	size_t n;
1425 
1426 	SHA1Update(ctx, buf, len);
1427 	n = fwrite(buf, 1, len, f);
1428 	if (n != len)
1429 		return got_ferror(f, GOT_ERR_IO);
1430 	return NULL;
1431 }
1432 
1433 static void
1434 putbe32(char *b, uint32_t n)
1435 {
1436 	b[0] = n >> 24;
1437 	b[1] = n >> 16;
1438 	b[2] = n >> 8;
1439 	b[3] = n >> 0;
1440 }
1441 
1442 static int
1443 write_order_cmp(const void *pa, const void *pb)
1444 {
1445 	struct got_pack_meta *a, *b, *ahd, *bhd;
1446 
1447 	a = *(struct got_pack_meta **)pa;
1448 	b = *(struct got_pack_meta **)pb;
1449 	ahd = (a->head == NULL) ? a : a->head;
1450 	bhd = (b->head == NULL) ? b : b->head;
1451 	if (ahd->mtime != bhd->mtime)
1452 		return bhd->mtime - ahd->mtime;
1453 	if (ahd != bhd)
1454 		return (uintptr_t)bhd - (uintptr_t)ahd;
1455 	if (a->nchain != b->nchain)
1456 		return a->nchain - b->nchain;
1457 	return a->mtime - b->mtime;
1458 }
1459 
1460 static int
1461 reuse_write_order_cmp(const void *pa, const void *pb)
1462 {
1463 	struct got_pack_meta *a, *b;
1464 
1465 	a = *(struct got_pack_meta **)pa;
1466 	b = *(struct got_pack_meta **)pb;
1467 
1468 	if (a->reused_delta_offset < b->reused_delta_offset)
1469 		return -1;
1470 	if (a->reused_delta_offset > b->reused_delta_offset)
1471 		return 1;
1472 	return 0;
1473 }
1474 
1475 static const struct got_error *
1476 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1477 {
1478 	size_t i;
1479 
1480 	*hdrlen = 0;
1481 
1482 	hdr[0] = obj_type << 4;
1483 	hdr[0] |= len & 0xf;
1484 	len >>= 4;
1485 	for (i = 1; len != 0; i++){
1486 		if (i >= bufsize)
1487 			return got_error(GOT_ERR_NO_SPACE);
1488 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1489 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1490 		len >>= GOT_DELTA_SIZE_SHIFT;
1491 	}
1492 
1493 	*hdrlen = i;
1494 	return NULL;
1495 }
1496 
1497 static int
1498 packoff(char *hdr, off_t off)
1499 {
1500 	int i, j;
1501 	char rbuf[8];
1502 
1503 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1504 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1505 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1506 		    GOT_DELTA_SIZE_MORE;
1507 	}
1508 
1509 	j = 0;
1510 	while (i > 0)
1511 		hdr[j++] = rbuf[--i];
1512 	return j;
1513 }
1514 
1515 static const struct got_error *
1516 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1517     struct got_pack_meta *m)
1518 {
1519 	const struct got_error *err;
1520 	char buf[32];
1521 	int nh;
1522 
1523 	if (m->prev->off != 0) {
1524 		err = packhdr(&nh, buf, sizeof(buf),
1525 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1526 		if (err)
1527 			return err;
1528 		nh += packoff(buf + nh, m->off - m->prev->off);
1529 		err = hwrite(packfile, buf, nh, ctx);
1530 		if (err)
1531 			return err;
1532 		*packfile_size += nh;
1533 	} else {
1534 		err = packhdr(&nh, buf, sizeof(buf),
1535 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1536 		if (err)
1537 			return err;
1538 		err = hwrite(packfile, buf, nh, ctx);
1539 		if (err)
1540 			return err;
1541 		*packfile_size += nh;
1542 		err = hwrite(packfile, m->prev->id.sha1,
1543 		    sizeof(m->prev->id.sha1), ctx);
1544 		if (err)
1545 			return err;
1546 		*packfile_size += sizeof(m->prev->id.sha1);
1547 	}
1548 
1549 	return NULL;
1550 }
1551 
1552 static const struct got_error *
1553 write_packed_object(off_t *packfile_size, FILE *packfile,
1554     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1555     SHA1_CTX *ctx, struct got_repository *repo)
1556 {
1557 	const struct got_error *err = NULL;
1558 	struct got_deflate_checksum csum;
1559 	char buf[32];
1560 	int nh;
1561 	struct got_raw_object *raw = NULL;
1562 	off_t outlen;
1563 
1564 	csum.output_sha1 = ctx;
1565 	csum.output_crc = NULL;
1566 
1567 	m->off = ftello(packfile);
1568 	if (m->delta_len == 0) {
1569 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1570 		if (err)
1571 			goto done;
1572 		err = packhdr(&nh, buf, sizeof(buf),
1573 		    m->obj_type, raw->size);
1574 		if (err)
1575 			goto done;
1576 		err = hwrite(packfile, buf, nh, ctx);
1577 		if (err)
1578 			goto done;
1579 		*packfile_size += nh;
1580 		if (raw->f == NULL) {
1581 			err = got_deflate_to_file_mmap(&outlen,
1582 			    raw->data + raw->hdrlen, 0, raw->size,
1583 			    packfile, &csum);
1584 			if (err)
1585 				goto done;
1586 		} else {
1587 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1588 			    == -1) {
1589 				err = got_error_from_errno("fseeko");
1590 				goto done;
1591 			}
1592 			err = got_deflate_to_file(&outlen, raw->f,
1593 			    raw->size, packfile, &csum);
1594 			if (err)
1595 				goto done;
1596 		}
1597 		*packfile_size += outlen;
1598 		got_object_raw_close(raw);
1599 		raw = NULL;
1600 	} else if (m->delta_buf) {
1601 		err = deltahdr(packfile_size, ctx, packfile, m);
1602 		if (err)
1603 			goto done;
1604 		err = got_deflate_to_file_mmap(&outlen,
1605 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1606 		if (err)
1607 			goto done;
1608 		*packfile_size += outlen;
1609 		free(m->delta_buf);
1610 		m->delta_buf = NULL;
1611 	} else {
1612 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1613 		    == -1) {
1614 			err = got_error_from_errno("fseeko");
1615 			goto done;
1616 		}
1617 		err = deltahdr(packfile_size, ctx, packfile, m);
1618 		if (err)
1619 			goto done;
1620 		err = got_deflate_to_file(&outlen, delta_cache,
1621 		    m->delta_len, packfile, &csum);
1622 		if (err)
1623 			goto done;
1624 		*packfile_size += outlen;
1625 	}
1626 done:
1627 	if (raw)
1628 		got_object_raw_close(raw);
1629 	return err;
1630 }
1631 
1632 static const struct got_error *
1633 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1634     struct got_pack_meta **deltify, int ndeltify,
1635     struct got_pack_meta **reuse, int nreuse,
1636     int nours, struct got_repository *repo,
1637     got_pack_progress_cb progress_cb, void *progress_arg,
1638     struct got_ratelimit *rl,
1639     got_cancel_cb cancel_cb, void *cancel_arg)
1640 {
1641 	const struct got_error *err = NULL;
1642 	int i;
1643 	SHA1_CTX ctx;
1644 	struct got_pack_meta *m;
1645 	char buf[32];
1646 	size_t n;
1647 	off_t packfile_size = 0;
1648 	int outfd = -1;
1649 
1650 	SHA1Init(&ctx);
1651 
1652 	err = hwrite(packfile, "PACK", 4, &ctx);
1653 	if (err)
1654 		return err;
1655 	putbe32(buf, GOT_PACKFILE_VERSION);
1656 	err = hwrite(packfile, buf, 4, &ctx);
1657 	if (err)
1658 		goto done;
1659 	putbe32(buf, ndeltify + nreuse);
1660 	err = hwrite(packfile, buf, 4, &ctx);
1661 	if (err)
1662 		goto done;
1663 
1664 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1665 	    write_order_cmp);
1666 	for (i = 0; i < ndeltify; i++) {
1667 		err = report_progress(progress_cb, progress_arg, rl,
1668 		    packfile_size, nours, ndeltify + nreuse,
1669 		    ndeltify + nreuse, i);
1670 		if (err)
1671 			goto done;
1672 		m = deltify[i];
1673 		err = write_packed_object(&packfile_size, packfile,
1674 		    delta_cache, m, &outfd, &ctx, repo);
1675 		if (err)
1676 			goto done;
1677 	}
1678 
1679 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1680 	    reuse_write_order_cmp);
1681 	for (i = 0; i < nreuse; i++) {
1682 		err = report_progress(progress_cb, progress_arg, rl,
1683 		    packfile_size, nours, ndeltify + nreuse,
1684 		    ndeltify + nreuse, ndeltify + i);
1685 		if (err)
1686 			goto done;
1687 		m = reuse[i];
1688 		err = write_packed_object(&packfile_size, packfile,
1689 		    delta_cache, m, &outfd, &ctx, repo);
1690 		if (err)
1691 			goto done;
1692 	}
1693 
1694 	SHA1Final(pack_sha1, &ctx);
1695 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1696 	if (n != SHA1_DIGEST_LENGTH)
1697 		err = got_ferror(packfile, GOT_ERR_IO);
1698 	packfile_size += SHA1_DIGEST_LENGTH;
1699 	packfile_size += sizeof(struct got_packfile_hdr);
1700 	if (progress_cb) {
1701 		err = progress_cb(progress_arg, packfile_size, nours,
1702 		    ndeltify + nreuse, ndeltify + nreuse,
1703 		    ndeltify + nreuse);
1704 		if (err)
1705 			goto done;
1706 	}
1707 done:
1708 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1709 		err = got_error_from_errno("close");
1710 	return err;
1711 }
1712 
1713 static const struct got_error *
1714 remove_unused_object(struct got_object_idset_element *entry, void *arg)
1715 {
1716 	struct got_object_idset *idset = arg;
1717 
1718 	if (got_object_idset_get_element_data(entry) == NULL)
1719 		got_object_idset_remove_element(idset, entry);
1720 
1721 	return NULL;
1722 }
1723 
1724 static const struct got_error *
1725 remove_reused_object(struct got_object_idset_element *entry, void *arg)
1726 {
1727 	struct got_object_idset *idset = arg;
1728 	struct got_pack_meta *m;
1729 
1730 	m = got_object_idset_get_element_data(entry);
1731 	if (m->have_reused_delta)
1732 		got_object_idset_remove_element(idset, entry);
1733 
1734 	return NULL;
1735 }
1736 
1737 static const struct got_error *
1738 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1739 {
1740 	struct got_pack_meta *m = data;
1741 	struct got_pack_metavec *v = arg;
1742 
1743 	return add_meta(m, v);
1744 }
1745 
1746 const struct got_error *
1747 got_pack_create(uint8_t *packsha1, FILE *packfile,
1748     struct got_object_id **theirs, int ntheirs,
1749     struct got_object_id **ours, int nours,
1750     struct got_repository *repo, int loose_obj_only, int allow_empty,
1751     got_pack_progress_cb progress_cb, void *progress_arg,
1752     got_cancel_cb cancel_cb, void *cancel_arg)
1753 {
1754 	const struct got_error *err;
1755 	int delta_cache_fd = -1;
1756 	FILE *delta_cache = NULL;
1757 	struct got_object_idset *idset;
1758 	struct got_ratelimit rl;
1759 	struct got_pack_metavec deltify, reuse;
1760 
1761 	memset(&deltify, 0, sizeof(deltify));
1762 	memset(&reuse, 0, sizeof(reuse));
1763 
1764 	got_ratelimit_init(&rl, 0, 500);
1765 
1766 	idset = got_object_idset_alloc();
1767 	if (idset == NULL)
1768 		return got_error_from_errno("got_object_idset_alloc");
1769 
1770 	err = load_object_ids(idset, theirs, ntheirs, ours, nours,
1771 	    repo, loose_obj_only, cancel_cb, cancel_arg);
1772 	if (err)
1773 		return err;
1774 
1775 	err = got_object_idset_for_each_element(idset,
1776 	    remove_unused_object, idset);
1777 	if (err)
1778 		goto done;
1779 
1780 	if (progress_cb) {
1781 		err = progress_cb(progress_arg, 0L, nours,
1782 		    got_object_idset_num_elements(idset), 0, 0);
1783 		if (err)
1784 			goto done;
1785 	}
1786 
1787 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1788 		err = got_error(GOT_ERR_CANNOT_PACK);
1789 		goto done;
1790 	}
1791 
1792 	delta_cache_fd = got_opentempfd();
1793 	if (delta_cache_fd == -1) {
1794 		err = got_error_from_errno("got_opentemp");
1795 		goto done;
1796 	}
1797 
1798 	reuse.metasz = 64;
1799 	reuse.meta = calloc(reuse.metasz,
1800 	    sizeof(struct got_pack_meta *));
1801 	if (reuse.meta == NULL) {
1802 		err = got_error_from_errno("calloc");
1803 		goto done;
1804 	}
1805 
1806 	err = search_deltas(&reuse, idset, delta_cache_fd, nours, repo,
1807 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1808 	if (err)
1809 		goto done;
1810 	if (reuse.nmeta > 0) {
1811 		err = got_object_idset_for_each_element(idset,
1812 		    remove_reused_object, idset);
1813 		if (err)
1814 			goto done;
1815 	}
1816 
1817 	delta_cache = fdopen(delta_cache_fd, "a+");
1818 	if (delta_cache == NULL) {
1819 		err = got_error_from_errno("fdopen");
1820 		goto done;
1821 	}
1822 	delta_cache_fd = -1;
1823 
1824 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1825 		err = got_error_from_errno("fseeko");
1826 		goto done;
1827 	}
1828 
1829 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1830 	    sizeof(struct got_pack_meta *));
1831 	if (deltify.meta == NULL) {
1832 		err = got_error_from_errno("calloc");
1833 		goto done;
1834 	}
1835 	deltify.metasz = got_object_idset_num_elements(idset);
1836 
1837 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1838 	if (err)
1839 		goto done;
1840 	if (deltify.nmeta > 0) {
1841 		err = pick_deltas(deltify.meta, deltify.nmeta, nours,
1842 		    reuse.nmeta, delta_cache, repo,
1843 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1844 		if (err)
1845 			goto done;
1846 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1847 			err = got_error_from_errno("fseeko");
1848 			goto done;
1849 		}
1850 	}
1851 
1852 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1853 	    deltify.nmeta, reuse.meta, reuse.nmeta, nours, repo,
1854 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1855 	if (err)
1856 		goto done;
1857 done:
1858 	free_nmeta(deltify.meta, deltify.nmeta);
1859 	free_nmeta(reuse.meta, reuse.nmeta);
1860 	got_object_idset_free(idset);
1861 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1862 		err = got_error_from_errno("close");
1863 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1864 		err = got_error_from_errno("fclose");
1865 	return err;
1866 }