Got Mirror

Blob

Date:: Mon Mar 21 15:52:15 2022 UTC
Message:: fix a bug where 'gotadmin pack' packed too many objects unless -a was used
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 <endian.h>
26 #include <stdint.h>
27 #include <imsg.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <sha1.h>
32 #include <time.h>
33 #include <unistd.h>
34 #include <limits.h>
35 #include <zlib.h>
36 
37 #include "got_error.h"
38 #include "got_cancel.h"
39 #include "got_object.h"
40 #include "got_path.h"
41 #include "got_reference.h"
42 #include "got_repository_admin.h"
43 #include "got_opentemp.h"
44 
45 #include "got_lib_deltify.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_idset.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_deflate.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_repository.h"
54 #include "got_lib_ratelimit.h"
55 
56 #ifndef MIN
57 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
58 #endif
59 
60 #ifndef MAX
61 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
62 #endif
63 
64 struct got_pack_meta {
65 	struct got_object_id id;
66 	char	*path;
67 	int	obj_type;
68 	off_t	size;
69 	time_t	mtime;
70 
71 	/* The best delta we picked */
72 	struct got_pack_meta *head;
73 	struct got_pack_meta *prev;
74 	unsigned char *delta_buf; /* if not encoded in delta cache file */
75 	off_t	delta_offset;	/* offset in delta cache file */
76 	off_t	delta_len;	/* encoded delta length */
77 	int	nchain;
78 
79 	int	have_reused_delta;
80 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
81 	struct got_object_id *base_obj_id;
82 
83 	/* Only used for delta window */
84 	struct got_delta_table *dtab;
85 
86 	/* Only used for writing offset deltas */
87 	off_t	off;
88 };
89 
90 struct got_pack_metavec {
91 	struct got_pack_meta **meta;
92 	int nmeta;
93 	int metasz;
94 };
95 
96 static const struct got_error *
97 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
98     const char *path, int obj_type, time_t mtime)
99 {
100 	const struct got_error *err = NULL;
101 	struct got_pack_meta *m;
102 
103 	*new = NULL;
104 
105 	m = calloc(1, sizeof(*m));
106 	if (m == NULL)
107 		return got_error_from_errno("calloc");
108 
109 	memcpy(&m->id, id, sizeof(m->id));
110 
111 	m->path = strdup(path);
112 	if (m->path == NULL) {
113 		err = got_error_from_errno("strdup");
114 		free(m);
115 		return err;
116 	}
117 
118 	m->obj_type = obj_type;
119 	m->mtime = mtime;
120 	*new = m;
121 	return NULL;
122 }
123 
124 static void
125 clear_meta(struct got_pack_meta *meta)
126 {
127 	if (meta == NULL)
128 		return;
129 	free(meta->path);
130 	meta->path = NULL;
131 	free(meta->delta_buf);
132 	meta->delta_buf = NULL;
133 	free(meta->base_obj_id);
134 	meta->base_obj_id = NULL;
135 }
136 
137 static void
138 free_nmeta(struct got_pack_meta **meta, int nmeta)
139 {
140 	int i;
141 
142 	for (i = 0; i < nmeta; i++)
143 		clear_meta(meta[i]);
144 	free(meta);
145 }
146 
147 static int
148 delta_order_cmp(const void *pa, const void *pb)
149 {
150 	struct got_pack_meta *a, *b;
151 	int cmp;
152 
153 	a = *(struct got_pack_meta **)pa;
154 	b = *(struct got_pack_meta **)pb;
155 
156 	if (a->obj_type != b->obj_type)
157 		return a->obj_type - b->obj_type;
158 	cmp = strcmp(a->path, b->path);
159 	if (cmp != 0)
160 		return cmp;
161 	if (a->mtime != b->mtime)
162 		return a->mtime - b->mtime;
163 	return got_object_id_cmp(&a->id, &b->id);
164 }
165 
166 static off_t
167 delta_size(struct got_delta_instruction *deltas, int ndeltas)
168 {
169 	int i;
170 	off_t size = 32;
171 	for (i = 0; i < ndeltas; i++) {
172 		if (deltas[i].copy)
173 			size += GOT_DELTA_SIZE_SHIFT;
174 		else
175 			size += deltas[i].len + 1;
176 	}
177 	return size;
178 }
179 
180 static const struct got_error *
181 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
182 {
183 	char *n;
184 
185 	if (*len + nseg >= *sz) {
186 		while (*len + nseg >= *sz)
187 			*sz += *sz / 2;
188 		n = realloc(*p, *sz);
189 		if (n == NULL)
190 			return got_error_from_errno("realloc");
191 		*p = n;
192 	}
193 	memcpy(*p + *len, seg, nseg);
194 	*len += nseg;
195 	return NULL;
196 }
197 
198 static const struct got_error *
199 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
200     struct got_delta_instruction *deltas, int ndeltas,
201     off_t delta_size, off_t base_size)
202 {
203 	const struct got_error *err;
204 	unsigned char buf[16], *bp;
205 	int i, j;
206 	size_t len = 0;
207 	off_t n;
208 	struct got_delta_instruction *d;
209 
210 	m->delta_buf = malloc(delta_size);
211 	if (m->delta_buf == NULL)
212 		return got_error_from_errno("calloc");
213 
214 	/* base object size */
215 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
216 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
217 	for (i = 1; n > 0; i++) {
218 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
219 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
220 		n >>= GOT_DELTA_SIZE_SHIFT;
221 	}
222 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
223 	if (err)
224 		return err;
225 
226 	/* target object size */
227 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
228 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
229 	for (i = 1; n > 0; i++) {
230 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
231 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
232 		n >>= GOT_DELTA_SIZE_SHIFT;
233 	}
234 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
235 	if (err)
236 		return err;
237 
238 	for (j = 0; j < ndeltas; j++) {
239 		d = &deltas[j];
240 		if (d->copy) {
241 			n = d->offset;
242 			bp = &buf[1];
243 			buf[0] = GOT_DELTA_BASE_COPY;
244 			for (i = 0; i < 4; i++) {
245 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
246 				buf[0] |= 1 << i;
247 				*bp++ = n & 0xff;
248 				n >>= 8;
249 				if (n == 0)
250 					break;
251 			}
252 
253 			n = d->len;
254 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
255 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
256 				for (i = 0; i < 3 && n > 0; i++) {
257 					buf[0] |= 1 << (i + 4);
258 					*bp++ = n & 0xff;
259 					n >>= 8;
260 				}
261 			}
262 			err = append(&m->delta_buf, &len, &delta_size,
263 			    buf, bp - buf);
264 			if (err)
265 				return err;
266 		} else if (o->f == NULL) {
267 			n = 0;
268 			while (n != d->len) {
269 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
270 				err = append(&m->delta_buf, &len, &delta_size,
271 				    buf, 1);
272 				if (err)
273 					return err;
274 				err = append(&m->delta_buf, &len, &delta_size,
275 				    o->data + o->hdrlen + d->offset + n,
276 				    buf[0]);
277 				if (err)
278 					return err;
279 				n += buf[0];
280 			}
281 		} else {
282 			char content[128];
283 			size_t r;
284 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
285 				return got_error_from_errno("fseeko");
286 			n = 0;
287 			while (n != d->len) {
288 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
289 				err = append(&m->delta_buf, &len, &delta_size,
290 				    buf, 1);
291 				if (err)
292 					return err;
293 				r = fread(content, 1, buf[0], o->f);
294 				if (r != buf[0])
295 					return got_ferror(o->f, GOT_ERR_IO);
296 				err = append(&m->delta_buf, &len, &delta_size,
297 				    content, buf[0]);
298 				if (err)
299 					return err;
300 				n += buf[0];
301 			}
302 		}
303 	}
304 
305 	m->delta_len = len;
306 	return NULL;
307 }
308 
309 static const struct got_error *
310 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
311     struct got_delta_instruction *deltas, int ndeltas,
312     off_t base_size, FILE *f)
313 {
314 	unsigned char buf[16], *bp;
315 	int i, j;
316 	off_t n;
317 	size_t w;
318 	struct got_delta_instruction *d;
319 
320 	/* base object size */
321 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
322 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
323 	for (i = 1; n > 0; i++) {
324 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
325 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
326 		n >>= GOT_DELTA_SIZE_SHIFT;
327 	}
328 	w = fwrite(buf, 1, i, f);
329 	if (w != i)
330 		return got_ferror(f, GOT_ERR_IO);
331 
332 	/* target object size */
333 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
334 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
335 	for (i = 1; n > 0; i++) {
336 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
337 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
338 		n >>= GOT_DELTA_SIZE_SHIFT;
339 	}
340 	w = fwrite(buf, 1, i, f);
341 	if (w != i)
342 		return got_ferror(f, GOT_ERR_IO);
343 
344 	for (j = 0; j < ndeltas; j++) {
345 		d = &deltas[j];
346 		if (d->copy) {
347 			n = d->offset;
348 			bp = &buf[1];
349 			buf[0] = GOT_DELTA_BASE_COPY;
350 			for (i = 0; i < 4; i++) {
351 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
352 				buf[0] |= 1 << i;
353 				*bp++ = n & 0xff;
354 				n >>= 8;
355 				if (n == 0)
356 					break;
357 			}
358 
359 			n = d->len;
360 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
361 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
362 				for (i = 0; i < 3 && n > 0; i++) {
363 					buf[0] |= 1 << (i + 4);
364 					*bp++ = n & 0xff;
365 					n >>= 8;
366 				}
367 			}
368 			w = fwrite(buf, 1, bp - buf, f);
369 			if (w != bp - buf)
370 				return got_ferror(f, GOT_ERR_IO);
371 		} else if (o->f == NULL) {
372 			n = 0;
373 			while (n != d->len) {
374 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
375 				w = fwrite(buf, 1, 1, f);
376 				if (w != 1)
377 					return got_ferror(f, GOT_ERR_IO);
378 				w = fwrite(o->data + o->hdrlen + d->offset + n,
379 				    1, buf[0], f);
380 				if (w != buf[0])
381 					return got_ferror(f, GOT_ERR_IO);
382 				n += buf[0];
383 			}
384 		} else {
385 			char content[128];
386 			size_t r;
387 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
388 				return got_error_from_errno("fseeko");
389 			n = 0;
390 			while (n != d->len) {
391 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
392 				w = fwrite(buf, 1, 1, f);
393 				if (w != 1)
394 					return got_ferror(f, GOT_ERR_IO);
395 				r = fread(content, 1, buf[0], o->f);
396 				if (r != buf[0])
397 					return got_ferror(o->f, GOT_ERR_IO);
398 				w = fwrite(content, 1, buf[0], f);
399 				if (w != buf[0])
400 					return got_ferror(f, GOT_ERR_IO);
401 				n += buf[0];
402 			}
403 		}
404 	}
405 
406 	m->delta_len = ftello(f) - m->delta_offset;
407 	return NULL;
408 }
409 
410 static const struct got_error *
411 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
412     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
413     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
414     int nobj_written)
415 {
416 	const struct got_error *err;
417 	int elapsed;
418 
419 	if (progress_cb == NULL)
420 		return NULL;
421 
422 	err = got_ratelimit_check(&elapsed, rl);
423 	if (err || !elapsed)
424 		return err;
425 
426 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
427 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
428 }
429 
430 static const struct got_error *
431 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
432 {
433 	if (v->nmeta == v->metasz){
434 		size_t newsize = 2 * v->metasz;
435 		struct got_pack_meta **new;
436 		new = reallocarray(v->meta, newsize, sizeof(*new));
437 		if (new == NULL)
438 			return got_error_from_errno("reallocarray");
439 		v->meta = new;
440 		v->metasz = newsize; 
441 	}
442 
443 	v->meta[v->nmeta++] = m;
444 	return NULL;
445 }
446 
447 static const struct got_error *
448 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
449     struct got_object_idset *idset, struct got_pack *pack,
450     struct got_packidx *packidx, int delta_cache_fd,
451     struct got_repository *repo)
452 {
453 	const struct got_error *err = NULL;
454 	struct got_pack_meta *base = NULL;
455 	struct got_object_id *base_obj_id = NULL;
456 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
457 	uint64_t base_size, result_size;
458 
459 	if (m->have_reused_delta)
460 		return NULL;
461 
462 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
463 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
464 	    packidx, idx, &m->id, repo);
465 	if (err)
466 		return err;
467 
468 	if (delta_offset + delta_len < delta_offset)
469 		return got_error(GOT_ERR_BAD_PACKFILE);
470 
471 	base = got_object_idset_get(idset, base_obj_id);
472 	if (base == NULL)
473 		goto done;
474 
475 	m->delta_len = delta_len;
476 	m->delta_offset = delta_cache_offset;
477 	m->prev = base;
478 	m->size = result_size;
479 	m->have_reused_delta = 1;
480 	m->reused_delta_offset = delta_offset;
481 	m->base_obj_id = base_obj_id;
482 	base_obj_id = NULL;
483 	err = add_meta(m, v);
484 done:
485 	free(base_obj_id);
486 	return err;
487 }
488 
489 static const struct got_error *
490 find_pack_for_reuse(struct got_packidx **best_packidx,
491     struct got_repository *repo)
492 {
493 	const struct got_error *err = NULL;
494 	struct got_pathlist_entry *pe;
495 	const char *best_packidx_path = NULL;
496 	int nobj_max = 0;
497 
498 	*best_packidx = NULL;
499 
500 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
501 		const char *path_packidx = pe->path;
502 		struct got_packidx *packidx;
503 		int nobj;
504 
505 		err = got_repo_get_packidx(&packidx, path_packidx, repo);
506 		if (err)
507 			break;
508 
509 		nobj = be32toh(packidx->hdr.fanout_table[0xff]);
510 		if (nobj > nobj_max) {
511 			best_packidx_path = path_packidx;
512 			nobj_max = nobj;
513 		}
514 	}
515 
516 	if (best_packidx_path) {
517 		err = got_repo_get_packidx(best_packidx, best_packidx_path,
518 		    repo);
519 	}
520 
521 	return err;
522 }
523 
524 struct search_deltas_arg {
525 	struct got_packidx *packidx;
526 	struct got_pack *pack;
527 	struct got_object_idset *idset;
528 	struct got_pack_metavec *v;
529 	int delta_cache_fd;
530 	struct got_repository *repo;
531 	got_pack_progress_cb progress_cb;
532 	void *progress_arg;
533 	struct got_ratelimit *rl;
534 	got_cancel_cb cancel_cb;
535 	void *cancel_arg;
536 	int ncolored;
537 	int nfound;
538 	int ntrees;
539 	int ncommits;
540 };
541 
542 static const struct got_error *
543 search_delta_for_object(struct got_object_id *id, void *data, void *arg)
544 {
545 	const struct got_error *err;
546 	struct got_pack_meta *m = data;
547 	struct search_deltas_arg *a = arg;
548 	int obj_idx;
549 	struct got_object *obj = NULL;
550 
551 	if (a->cancel_cb) {
552 		err = (*a->cancel_cb)(a->cancel_arg);
553 		if (err)
554 			return err;
555 	}
556 
557 	if (!got_repo_check_packidx_bloom_filter(a->repo,
558 	    a->packidx->path_packidx, id))
559 		return NULL;
560 
561 	obj_idx = got_packidx_get_object_idx(a->packidx, id);
562 	if (obj_idx == -1)
563 		return NULL;
564 
565 	/* TODO:
566 	 * Opening and closing an object just to check its flags
567 	 * is a bit expensive. We could have an imsg which requests
568 	 * plain type/size information for an object without doing
569 	 * work such as traversing the object's entire delta chain
570 	 * to find the base object type, and other such info which
571 	 * we don't really need here.
572 	 */
573 	err = got_object_open_from_packfile(&obj, &m->id, a->pack,
574 	    a->packidx, obj_idx, a->repo);
575 	if (err)
576 		return err;
577 
578 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
579 		reuse_delta(obj_idx, m, a->v, a->idset, a->pack, a->packidx,
580 		    a->delta_cache_fd, a->repo);
581 		if (err)
582 			goto done;
583 		err = report_progress(a->progress_cb, a->progress_arg, a->rl,
584 		    a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
585 		    got_object_idset_num_elements(a->idset), a->v->nmeta, 0);
586 	}
587 done:
588 	got_object_close(obj);
589 	return err;
590 }
591 
592 static const struct got_error *
593 search_deltas(struct got_pack_metavec *v, struct got_object_idset *idset,
594     int delta_cache_fd, int ncolored, int nfound, int ntrees, int ncommits,
595     struct got_repository *repo,
596     got_pack_progress_cb progress_cb, void *progress_arg,
597     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
598 {
599 	const struct got_error *err = NULL;
600 	char *path_packfile = NULL;
601 	struct got_packidx *packidx;
602 	struct got_pack *pack;
603 	struct search_deltas_arg sda;
604 
605 	err = find_pack_for_reuse(&packidx, repo);
606 	if (err)
607 		return err;
608 
609 	if (packidx == NULL)
610 		return NULL;
611 
612 	err = got_packidx_get_packfile_path(&path_packfile,
613 	    packidx->path_packidx);
614 	if (err)
615 		return err;
616 
617 	pack = got_repo_get_cached_pack(repo, path_packfile);
618 	if (pack == NULL) {
619 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
620 		if (err)
621 			goto done;
622 	}
623 
624 	sda.packidx = packidx;
625 	sda.pack = pack;
626 	sda.idset = idset;
627 	sda.v = v;
628 	sda.delta_cache_fd = delta_cache_fd;
629 	sda.repo = repo;
630 	sda.progress_cb = progress_cb;
631 	sda.progress_arg = progress_arg;
632 	sda.rl = rl;
633 	sda.cancel_cb = cancel_cb;
634 	sda.cancel_arg = cancel_arg;
635 	sda.ncolored = ncolored;
636 	sda.nfound = nfound;
637 	sda.ntrees = ntrees;
638 	sda.ncommits = ncommits;
639 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
640 done:
641 	free(path_packfile);
642 	return err;
643 }
644 
645 static const struct got_error *
646 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
647     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
648     struct got_repository *repo,
649     got_pack_progress_cb progress_cb, void *progress_arg,
650     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
651 {
652 	const struct got_error *err = NULL;
653 	struct got_pack_meta *m = NULL, *base = NULL;
654 	struct got_raw_object *raw = NULL, *base_raw = NULL;
655 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
656 	int i, j, ndeltas, best_ndeltas;
657 	off_t size, best_size;
658 	const int max_base_candidates = 3;
659 	size_t delta_memsize = 0;
660 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
661 	int outfd = -1;
662 
663 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
664 	for (i = 0; i < nmeta; i++) {
665 		if (cancel_cb) {
666 			err = (*cancel_cb)(cancel_arg);
667 			if (err)
668 				break;
669 		}
670 		err = report_progress(progress_cb, progress_arg, rl,
671 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
672 		    nreused + i, 0);
673 		if (err)
674 			goto done;
675 		m = meta[i];
676 
677 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
678 		    m->obj_type == GOT_OBJ_TYPE_TAG)
679 			continue;
680 
681 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
682 		if (err)
683 			goto done;
684 		m->size = raw->size;
685 
686 		if (raw->f == NULL) {
687 			err = got_deltify_init_mem(&m->dtab, raw->data,
688 			    raw->hdrlen, raw->size + raw->hdrlen);
689 		} else {
690 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
691 			    raw->size + raw->hdrlen);
692 		}
693 		if (err)
694 			goto done;
695 
696 		if (i > max_base_candidates) {
697 			struct got_pack_meta *n = NULL;
698 			n = meta[i - (max_base_candidates + 1)];
699 			got_deltify_free(n->dtab);
700 			n->dtab = NULL;
701 		}
702 
703 		best_size = raw->size;
704 		best_ndeltas = 0;
705 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
706 			if (cancel_cb) {
707 				err = (*cancel_cb)(cancel_arg);
708 				if (err)
709 					goto done;
710 			}
711 			base = meta[j];
712 			/* long chains make unpacking slow, avoid such bases */
713 			if (base->nchain >= 128 ||
714 			    base->obj_type != m->obj_type)
715 				continue;
716 
717 			err = got_object_raw_open(&base_raw, &outfd, repo,
718 			    &base->id);
719 			if (err)
720 				goto done;
721 
722 			if (raw->f == NULL && base_raw->f == NULL) {
723 				err = got_deltify_mem_mem(&deltas, &ndeltas,
724 				    raw->data, raw->hdrlen,
725 				    raw->size + raw->hdrlen,
726 				    base->dtab, base_raw->data,
727 				    base_raw->hdrlen,
728 				    base_raw->size + base_raw->hdrlen);
729 			} else if (raw->f == NULL) {
730 				err = got_deltify_mem_file(&deltas, &ndeltas,
731 				    raw->data, raw->hdrlen,
732 				    raw->size + raw->hdrlen,
733 				    base->dtab, base_raw->f,
734 				    base_raw->hdrlen,
735 				    base_raw->size + base_raw->hdrlen);
736 			} else if (base_raw->f == NULL) {
737 				err = got_deltify_file_mem(&deltas, &ndeltas,
738 				    raw->f, raw->hdrlen,
739 				    raw->size + raw->hdrlen,
740 				    base->dtab, base_raw->data,
741 				    base_raw->hdrlen,
742 				    base_raw->size + base_raw->hdrlen);
743 			} else {
744 				err = got_deltify(&deltas, &ndeltas,
745 				    raw->f, raw->hdrlen,
746 				    raw->size + raw->hdrlen,
747 				    base->dtab, base_raw->f, base_raw->hdrlen,
748 				    base_raw->size + base_raw->hdrlen);
749 			}
750 			got_object_raw_close(base_raw);
751 			base_raw = NULL;
752 			if (err)
753 				goto done;
754 
755 			size = delta_size(deltas, ndeltas);
756 			if (size + 32 < best_size){
757 				/*
758 				 * if we already picked a best delta,
759 				 * replace it.
760 				 */
761 				best_size = size;
762 				free(best_deltas);
763 				best_deltas = deltas;
764 				best_ndeltas = ndeltas;
765 				deltas = NULL;
766 				m->nchain = base->nchain + 1;
767 				m->prev = base;
768 				m->head = base->head;
769 				if (m->head == NULL)
770 					m->head = base;
771 			} else {
772 				free(deltas);
773 				deltas = NULL;
774 				ndeltas = 0;
775 			}
776 		}
777 
778 		if (best_ndeltas > 0) {
779 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
780 			    delta_memsize + best_size <= max_delta_memsize) {
781 				delta_memsize += best_size;
782 				err = encode_delta_in_mem(m, raw, best_deltas,
783 				    best_ndeltas, best_size, m->prev->size);
784 			} else {
785 				m->delta_offset = ftello(delta_cache);
786 				/*
787 				 * TODO:
788 				 * Storing compressed delta data in the delta
789 				 * cache file would probably be more efficient
790 				 * than writing uncompressed delta data here
791 				 * and compressing it while writing the pack
792 				 * file. This would also allow for reusing
793 				 * deltas in their compressed form.
794 				 */
795 				err = encode_delta(m, raw, best_deltas,
796 				    best_ndeltas, m->prev->size, delta_cache);
797 			}
798 			free(best_deltas);
799 			best_deltas = NULL;
800 			best_ndeltas = 0;
801 			if (err)
802 				goto done;
803 		}
804 
805 		got_object_raw_close(raw);
806 		raw = NULL;
807 	}
808 done:
809 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
810 		got_deltify_free(meta[i]->dtab);
811 		meta[i]->dtab = NULL;
812 	}
813 	if (raw)
814 		got_object_raw_close(raw);
815 	if (base_raw)
816 		got_object_raw_close(base_raw);
817 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
818 		err = got_error_from_errno("close");
819 	free(deltas);
820 	free(best_deltas);
821 	return err;
822 }
823 
824 static const struct got_error *
825 search_packidx(int *found, struct got_object_id *id,
826     struct got_repository *repo)
827 {
828 	const struct got_error *err = NULL;
829 	struct got_packidx *packidx = NULL;
830 	int idx;
831 
832 	*found = 0;
833 
834 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
835 	if (err == NULL)
836 		*found = 1; /* object is already packed */
837 	else if (err->code == GOT_ERR_NO_OBJ)
838 		err = NULL;
839 	return err;
840 }
841 
842 static const int obj_types[] = {
843 	GOT_OBJ_TYPE_ANY,
844 	GOT_OBJ_TYPE_COMMIT,
845 	GOT_OBJ_TYPE_TREE,
846 	GOT_OBJ_TYPE_BLOB,
847 	GOT_OBJ_TYPE_TAG,
848 	GOT_OBJ_TYPE_OFFSET_DELTA,
849 	GOT_OBJ_TYPE_REF_DELTA
850 };
851 
852 static const struct got_error *
853 add_object(int want_meta, struct got_object_idset *idset,
854     struct got_object_id *id, const char *path, int obj_type,
855     time_t mtime, int loose_obj_only, struct got_repository *repo)
856 {
857 	const struct got_error *err;
858 	struct got_pack_meta *m = NULL;
859 
860 	if (loose_obj_only) {
861 		int is_packed;
862 		err = search_packidx(&is_packed, id, repo);
863 		if (err)
864 			return err;
865 		if (is_packed && want_meta)
866 			return NULL;
867 	}
868 
869 	if (want_meta) {
870 		err = alloc_meta(&m, id, path, obj_type, mtime);
871 		if (err)
872 			return err;
873 	}
874 
875 	return got_object_idset_add(idset, id, m);
876 }
877 
878 static const struct got_error *
879 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
880     struct got_object_idset *idset, struct got_object_id *tree_id,
881     const char *dpath, time_t mtime, struct got_repository *repo,
882     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
883     got_pack_progress_cb progress_cb, void *progress_arg,
884     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
885 {
886 	const struct got_error *err;
887 	struct got_tree_object *tree;
888 	char *p = NULL;
889 	int i;
890 
891 	err = got_object_open_as_tree(&tree, repo, tree_id);
892 	if (err)
893 		return err;
894 
895 	(*ntrees)++;
896 	err = report_progress(progress_cb, progress_arg, rl,
897 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
898 	if (err)
899 		return err;
900 
901 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
902 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
903 		struct got_object_id *id = got_tree_entry_get_id(e);
904 		mode_t mode = got_tree_entry_get_mode(e);
905 
906 		if (cancel_cb) {
907 			err = (*cancel_cb)(cancel_arg);
908 			if (err)
909 				break;
910 		}
911 
912 		if (got_object_tree_entry_is_submodule(e) ||
913 		    got_object_idset_contains(idset, id))
914 			continue;
915 		
916 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
917 		    got_tree_entry_get_name(e)) == -1) {
918 			err = got_error_from_errno("asprintf");
919 			break;
920 		}
921 
922 		if (S_ISDIR(mode)) {
923 			struct got_object_qid *qid;
924 			err = got_object_qid_alloc(&qid, id);
925 			if (err)
926 				break;
927 			STAILQ_INSERT_TAIL(ids, qid, entry);
928 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
929 			err = add_object(want_meta, idset, id, p,
930 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo);
931 			if (err)
932 				break;
933 			(*nfound)++;
934 			err = report_progress(progress_cb, progress_arg, rl,
935 			    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
936 			if (err)
937 				break;
938 		}
939 		free(p);
940 		p = NULL;
941 	}
942 
943 	got_object_tree_close(tree);
944 	free(p);
945 	return err;
946 }
947 
948 static const struct got_error *
949 load_tree(int want_meta, struct got_object_idset *idset,
950     struct got_object_id *tree_id, const char *dpath, time_t mtime,
951     struct got_repository *repo, int loose_obj_only,
952     int *ncolored, int *nfound, int *ntrees,
953     got_pack_progress_cb progress_cb, void *progress_arg,
954     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
955 {
956 	const struct got_error *err = NULL;
957 	struct got_object_id_queue tree_ids;
958 	struct got_object_qid *qid;
959 
960 	if (got_object_idset_contains(idset, tree_id))
961 		return NULL;
962 
963 	err = got_object_qid_alloc(&qid, tree_id);
964 	if (err)
965 		return err;
966 
967 	STAILQ_INIT(&tree_ids);
968 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
969 
970 	while (!STAILQ_EMPTY(&tree_ids)) {
971 		if (cancel_cb) {
972 			err = (*cancel_cb)(cancel_arg);
973 			if (err)
974 				break;
975 		}
976 
977 		qid = STAILQ_FIRST(&tree_ids);
978 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
979 
980 		if (got_object_idset_contains(idset, qid->id)) {
981 			got_object_qid_free(qid);
982 			continue;
983 		}
984 
985 		err = add_object(want_meta, idset, qid->id, dpath,
986 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo);
987 		if (err) {
988 			got_object_qid_free(qid);
989 			break;
990 		}
991 
992 		(*nfound)++;
993 		err = report_progress(progress_cb, progress_arg, rl,
994 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
995 		if (err)
996 			break;
997 
998 		err = load_tree_entries(&tree_ids, want_meta, idset, qid->id,
999 		    dpath, mtime, repo, loose_obj_only, ncolored, nfound,
1000 		    ntrees, progress_cb, progress_arg, rl,
1001 		    cancel_cb, cancel_arg);
1002 		got_object_qid_free(qid);
1003 		if (err)
1004 			break;
1005 	}
1006 
1007 	got_object_id_queue_free(&tree_ids);
1008 	return err;
1009 }
1010 
1011 static const struct got_error *
1012 load_commit(int want_meta, struct got_object_idset *idset,
1013     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1014     int *ncolored, int *nfound, int *ntrees,
1015     got_pack_progress_cb progress_cb, void *progress_arg,
1016     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1017 {
1018 	const struct got_error *err;
1019 	struct got_commit_object *commit;
1020 
1021 	if (got_object_idset_contains(idset, id))
1022 		return NULL;
1023 
1024 	if (loose_obj_only) {
1025 		int is_packed;
1026 		err = search_packidx(&is_packed, id, repo);
1027 		if (err)
1028 			return err;
1029 		if (is_packed && want_meta)
1030 			return NULL;
1031 	}
1032 
1033 	err = got_object_open_as_commit(&commit, repo, id);
1034 	if (err)
1035 		return err;
1036 
1037 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1038 	    got_object_commit_get_committer_time(commit),
1039 	    loose_obj_only, repo);
1040 	if (err)
1041 		goto done;
1042 
1043 	(*nfound)++;
1044 	err = report_progress(progress_cb, progress_arg, rl,
1045 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1046 	if (err)
1047 		goto done;
1048 
1049 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1050 	    "", got_object_commit_get_committer_time(commit),
1051 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1052 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1053 done:
1054 	got_object_commit_close(commit);
1055 	return err;
1056 }
1057 
1058 static const struct got_error *
1059 load_tag(int want_meta, struct got_object_idset *idset,
1060     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1061     int *ncolored, int *nfound, int *ntrees,
1062     got_pack_progress_cb progress_cb, void *progress_arg,
1063     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1064 {
1065 	const struct got_error *err;
1066 	struct got_tag_object *tag = NULL;
1067 
1068 	if (got_object_idset_contains(idset, id))
1069 		return NULL;
1070 
1071 	if (loose_obj_only) {
1072 		int is_packed;
1073 		err = search_packidx(&is_packed, id, repo);
1074 		if (err)
1075 			return err;
1076 		if (is_packed && want_meta)
1077 			return NULL;
1078 	}
1079 
1080 	err = got_object_open_as_tag(&tag, repo, id);
1081 	if (err)
1082 		return err;
1083 
1084 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1085 	    got_object_tag_get_tagger_time(tag),
1086 	    loose_obj_only, repo);
1087 	if (err)
1088 		goto done;
1089 
1090 	(*nfound)++;
1091 	err = report_progress(progress_cb, progress_arg, rl,
1092 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
1093 	if (err)
1094 		goto done;
1095 
1096 	switch (got_object_tag_get_object_type(tag)) {
1097 	case GOT_OBJ_TYPE_COMMIT:
1098 		err = load_commit(want_meta, idset,
1099 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1100 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1101 		    cancel_cb, cancel_arg);
1102 		break;
1103 	case GOT_OBJ_TYPE_TREE:
1104 		err = load_tree(want_meta, idset,
1105 		    got_object_tag_get_object_id(tag), "",
1106 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1107 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1108 		    cancel_cb, cancel_arg);
1109 		break;
1110 	default:
1111 		break;
1112 	}
1113 
1114 done:
1115 	got_object_tag_close(tag);
1116 	return err;
1117 }
1118 
1119 enum findtwixt_color {
1120 	COLOR_KEEP = 0,
1121 	COLOR_DROP,
1122 	COLOR_BLANK,
1123 };
1124 static const int findtwixt_colors[] = {
1125 	COLOR_KEEP,
1126 	COLOR_DROP,
1127 	COLOR_BLANK
1128 };
1129 
1130 static const struct got_error *
1131 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1132     int color, struct got_repository *repo)
1133 {
1134 	const struct got_error *err;
1135 	struct got_object_qid *qid;
1136 
1137 	err = got_object_qid_alloc(&qid, id);
1138 	if (err)
1139 		return err;
1140 
1141 	STAILQ_INSERT_TAIL(ids, qid, entry);
1142 	qid->data = (void *)&findtwixt_colors[color];
1143 	return NULL;
1144 }
1145 
1146 static const struct got_error *
1147 drop_commit(struct got_object_idset *keep, struct got_object_idset *drop,
1148     struct got_object_id *id, struct got_repository *repo,
1149     got_cancel_cb cancel_cb, void *cancel_arg)
1150 {
1151 	const struct got_error *err = NULL;
1152 	struct got_commit_object *commit;
1153 	const struct got_object_id_queue *parents;
1154 	struct got_object_id_queue ids;
1155 	struct got_object_qid *qid;
1156 
1157 	STAILQ_INIT(&ids);
1158 
1159 	err = got_object_qid_alloc(&qid, id);
1160 	if (err)
1161 		return err;
1162 	STAILQ_INSERT_HEAD(&ids, qid, entry);
1163 
1164 	while (!STAILQ_EMPTY(&ids)) {
1165 		if (cancel_cb) {
1166 			err = (*cancel_cb)(cancel_arg);
1167 			if (err)
1168 				break;
1169 		}
1170 
1171 		qid = STAILQ_FIRST(&ids);
1172 		STAILQ_REMOVE_HEAD(&ids, entry);
1173 
1174 		if (got_object_idset_contains(drop, qid->id)) {
1175 			got_object_qid_free(qid);
1176 			continue;
1177 		}
1178 
1179 		err = got_object_idset_add(drop, qid->id,
1180 		    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1181 		if (err) {
1182 			got_object_qid_free(qid);
1183 			break;
1184 		}
1185 
1186 		if (!got_object_idset_contains(keep, qid->id)) {
1187 			got_object_qid_free(qid);
1188 			continue;
1189 		}
1190 
1191 		err = got_object_open_as_commit(&commit, repo, qid->id);
1192 		got_object_qid_free(qid);
1193 		if (err)
1194 			break;
1195 
1196 		parents = got_object_commit_get_parent_ids(commit);
1197 		if (parents) {
1198 			err = got_object_id_queue_copy(parents, &ids);
1199 			if (err) {
1200 				got_object_commit_close(commit);
1201 				break;
1202 			}
1203 		}
1204 		got_object_commit_close(commit);
1205 	}
1206 
1207 	got_object_id_queue_free(&ids);
1208 	return err;
1209 }
1210 
1211 struct append_id_arg {
1212 	struct got_object_id **array;
1213 	int idx;
1214 };
1215 
1216 static const struct got_error *
1217 append_id(struct got_object_id *id, void *data, void *arg)
1218 {
1219 	struct append_id_arg *a = arg;
1220 
1221 	a->array[a->idx] = got_object_id_dup(id);
1222 	if (a->array[a->idx] == NULL)
1223 		return got_error_from_errno("got_object_id_dup");
1224 
1225 	a->idx++;
1226 	return NULL;
1227 }
1228 
1229 static const struct got_error *
1230 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1231     struct got_object_id **head, int nhead,
1232     struct got_object_id **tail, int ntail,
1233     struct got_repository *repo,
1234     got_pack_progress_cb progress_cb, void *progress_arg,
1235     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1236 {
1237 	const struct got_error *err = NULL;
1238 	struct got_object_id_queue ids;
1239 	struct got_object_idset *keep, *drop;
1240 	struct got_object_qid *qid;
1241 	int i, ncolor, nkeep, obj_type;
1242 
1243 	STAILQ_INIT(&ids);
1244 	*res = NULL;
1245 	*nres = 0;
1246 	*ncolored = 0;
1247 
1248 	keep = got_object_idset_alloc();
1249 	if (keep == NULL)
1250 		return got_error_from_errno("got_object_idset_alloc");
1251 
1252 	drop = got_object_idset_alloc();
1253 	if (drop == NULL) {
1254 		err = got_error_from_errno("got_object_idset_alloc");
1255 		goto done;
1256 	}
1257 
1258 	for (i = 0; i < nhead; i++) {
1259 		struct got_object_id *id = head[i];
1260 		if (id == NULL)
1261 			continue;
1262 		err = got_object_get_type(&obj_type, repo, id);
1263 		if (err)
1264 			return err;
1265 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1266 			continue;
1267 		err = queue_commit_id(&ids, id, COLOR_KEEP, repo);
1268 		if (err)
1269 			goto done;
1270 	}		
1271 	for (i = 0; i < ntail; i++) {
1272 		struct got_object_id *id = tail[i];
1273 		if (id == NULL)
1274 			continue;
1275 		err = got_object_get_type(&obj_type, repo, id);
1276 		if (err)
1277 			return err;
1278 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1279 			continue;
1280 		err = queue_commit_id(&ids, id, COLOR_DROP, repo);
1281 		if (err)
1282 			goto done;
1283 	}
1284 
1285 	while (!STAILQ_EMPTY(&ids)) {
1286 		int qcolor;
1287 		qid = STAILQ_FIRST(&ids);
1288 		qcolor = *((int *)qid->data);
1289 
1290 		if (got_object_idset_contains(drop, qid->id))
1291 			ncolor = COLOR_DROP;
1292 		else if (got_object_idset_contains(keep, qid->id))
1293 			ncolor = COLOR_KEEP;
1294 		else
1295 			ncolor = COLOR_BLANK;
1296 
1297 		(*ncolored)++;
1298 		err = report_progress(progress_cb, progress_arg, rl,
1299 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1300 		if (err)
1301 			goto done;
1302 
1303 		if (ncolor == COLOR_DROP || (ncolor == COLOR_KEEP &&
1304 		    qcolor == COLOR_KEEP)) {
1305 			STAILQ_REMOVE_HEAD(&ids, entry);
1306 			got_object_qid_free(qid);
1307 			continue;
1308 		}
1309 
1310 		if (ncolor == COLOR_KEEP && qcolor == COLOR_DROP) {
1311 			err = drop_commit(keep, drop, qid->id, repo,
1312 			    cancel_cb, cancel_arg);
1313 			if (err)
1314 				goto done;
1315 		} else if (ncolor == COLOR_BLANK) {
1316 			struct got_commit_object *commit;
1317 			struct got_object_id *id;
1318 			const struct got_object_id_queue *parents;
1319 			struct got_object_qid *pid;
1320 
1321 			id = got_object_id_dup(qid->id);
1322 			if (id == NULL) {
1323 				err = got_error_from_errno("got_object_id_dup");
1324 				goto done;
1325 			}
1326 			if (qcolor == COLOR_KEEP)
1327 				err = got_object_idset_add(keep, id,
1328 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1329 			else
1330 				err = got_object_idset_add(drop, id,
1331 				    (void *)&obj_types[GOT_OBJ_TYPE_COMMIT]);
1332 			if (err) {
1333 				free(id);
1334 				goto done;
1335 			}
1336 
1337 			err = got_object_open_as_commit(&commit, repo, id);
1338 			if (err) {
1339 				free(id);
1340 				goto done;
1341 			}
1342 			parents = got_object_commit_get_parent_ids(commit);
1343 			if (parents) {
1344 				STAILQ_FOREACH(pid, parents, entry) {
1345 					err = queue_commit_id(&ids, pid->id,
1346 					    qcolor, repo);
1347 					if (err) {
1348 						free(id);
1349 						goto done;
1350 					}
1351 				}
1352 			}
1353 			got_object_commit_close(commit);
1354 			commit = NULL;
1355 		} else {
1356 			/* should not happen */
1357 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1358 			    "%s ncolor=%d qcolor=%d", __func__, ncolor, qcolor);
1359 			goto done;
1360 		}
1361 
1362 		STAILQ_REMOVE_HEAD(&ids, entry);
1363 		got_object_qid_free(qid);
1364 	}
1365 
1366 	nkeep = got_object_idset_num_elements(keep);
1367 	if (nkeep > 0) {
1368 		struct append_id_arg arg;
1369 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1370 		if (arg.array == NULL) {
1371 			err = got_error_from_errno("calloc");
1372 			goto done;
1373 		}
1374 		arg.idx = 0;
1375 		err = got_object_idset_for_each(keep, append_id, &arg);
1376 		if (err) {
1377 			free(arg.array);
1378 			goto done;
1379 		}
1380 		*res = arg.array;
1381 		*nres = nkeep;
1382 	}
1383 done:
1384 	got_object_idset_free(keep);
1385 	got_object_idset_free(drop);
1386 	got_object_id_queue_free(&ids);
1387 	return err;
1388 }
1389 
1390 static const struct got_error *
1391 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1392     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1393     struct got_object_id **ours, int nours, struct got_repository *repo,
1394     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1395     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1396 {
1397 	const struct got_error *err = NULL;
1398 	struct got_object_id **ids = NULL;
1399 	int i, nobj = 0, obj_type;
1400 
1401 	*ncolored = 0;
1402 	*nfound = 0;
1403 	*ntrees = 0;
1404 
1405 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1406 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1407 	if (err || nobj == 0)
1408 		goto done;
1409 
1410 	for (i = 0; i < ntheirs; i++) {
1411 		struct got_object_id *id = theirs[i];
1412 		if (id == NULL)
1413 			continue;
1414 		err = got_object_get_type(&obj_type, repo, id);
1415 		if (err)
1416 			return err;
1417 		if (obj_type != GOT_OBJ_TYPE_COMMIT)
1418 			continue;
1419 		err = load_commit(0, idset, id, repo, loose_obj_only,
1420 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1421 		    cancel_cb, cancel_arg);
1422 		if (err)
1423 			goto done;
1424 	}
1425 
1426 	for (i = 0; i < ntheirs; i++) {
1427 		struct got_object_id *id = theirs[i];
1428 		struct got_pack_meta *m;
1429 		if (id == NULL)
1430 			continue;
1431 		m = got_object_idset_get(idset, id);
1432 		if (m == NULL) {
1433 			err = got_object_get_type(&obj_type, repo, id);
1434 			if (err)
1435 				goto done;
1436 		} else
1437 			obj_type = m->obj_type;
1438 		if (obj_type != GOT_OBJ_TYPE_TAG)
1439 			continue;
1440 		err = load_tag(0, idset, id, repo, loose_obj_only,
1441 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1442 		    cancel_cb, cancel_arg);
1443 		if (err)
1444 			goto done;
1445 	}
1446 
1447 	for (i = 0; i < nobj; i++) {
1448 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1449 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1450 		    cancel_cb, cancel_arg);
1451 		if (err)
1452 			goto done;
1453 	}
1454 
1455 	for (i = 0; i < nours; i++) {
1456 		struct got_object_id *id = ours[i];
1457 		struct got_pack_meta *m;
1458 		if (id == NULL)
1459 			continue;
1460 		m = got_object_idset_get(idset, id);
1461 		if (m == NULL) {
1462 			err = got_object_get_type(&obj_type, repo, id);
1463 			if (err)
1464 				goto done;
1465 		} else
1466 			obj_type = m->obj_type;
1467 		if (obj_type != GOT_OBJ_TYPE_TAG)
1468 			continue;
1469 		err = load_tag(1, idset, id, repo, loose_obj_only,
1470 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1471 		    cancel_cb, cancel_arg);
1472 		if (err)
1473 			goto done;
1474 	}
1475 done:
1476 	for (i = 0; i < nobj; i++) {
1477 		free(ids[i]);
1478 	}
1479 	free(ids);
1480 	return err;
1481 }
1482 
1483 const struct got_error *
1484 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1485 {
1486 	size_t n;
1487 
1488 	SHA1Update(ctx, buf, len);
1489 	n = fwrite(buf, 1, len, f);
1490 	if (n != len)
1491 		return got_ferror(f, GOT_ERR_IO);
1492 	return NULL;
1493 }
1494 
1495 static void
1496 putbe32(char *b, uint32_t n)
1497 {
1498 	b[0] = n >> 24;
1499 	b[1] = n >> 16;
1500 	b[2] = n >> 8;
1501 	b[3] = n >> 0;
1502 }
1503 
1504 static int
1505 write_order_cmp(const void *pa, const void *pb)
1506 {
1507 	struct got_pack_meta *a, *b, *ahd, *bhd;
1508 
1509 	a = *(struct got_pack_meta **)pa;
1510 	b = *(struct got_pack_meta **)pb;
1511 	ahd = (a->head == NULL) ? a : a->head;
1512 	bhd = (b->head == NULL) ? b : b->head;
1513 	if (ahd->mtime != bhd->mtime)
1514 		return bhd->mtime - ahd->mtime;
1515 	if (ahd != bhd)
1516 		return (uintptr_t)bhd - (uintptr_t)ahd;
1517 	if (a->nchain != b->nchain)
1518 		return a->nchain - b->nchain;
1519 	return a->mtime - b->mtime;
1520 }
1521 
1522 static int
1523 reuse_write_order_cmp(const void *pa, const void *pb)
1524 {
1525 	struct got_pack_meta *a, *b;
1526 
1527 	a = *(struct got_pack_meta **)pa;
1528 	b = *(struct got_pack_meta **)pb;
1529 
1530 	if (a->reused_delta_offset < b->reused_delta_offset)
1531 		return -1;
1532 	if (a->reused_delta_offset > b->reused_delta_offset)
1533 		return 1;
1534 	return 0;
1535 }
1536 
1537 static const struct got_error *
1538 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1539 {
1540 	size_t i;
1541 
1542 	*hdrlen = 0;
1543 
1544 	hdr[0] = obj_type << 4;
1545 	hdr[0] |= len & 0xf;
1546 	len >>= 4;
1547 	for (i = 1; len != 0; i++){
1548 		if (i >= bufsize)
1549 			return got_error(GOT_ERR_NO_SPACE);
1550 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1551 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1552 		len >>= GOT_DELTA_SIZE_SHIFT;
1553 	}
1554 
1555 	*hdrlen = i;
1556 	return NULL;
1557 }
1558 
1559 static int
1560 packoff(char *hdr, off_t off)
1561 {
1562 	int i, j;
1563 	char rbuf[8];
1564 
1565 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1566 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1567 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1568 		    GOT_DELTA_SIZE_MORE;
1569 	}
1570 
1571 	j = 0;
1572 	while (i > 0)
1573 		hdr[j++] = rbuf[--i];
1574 	return j;
1575 }
1576 
1577 static const struct got_error *
1578 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1579     struct got_pack_meta *m)
1580 {
1581 	const struct got_error *err;
1582 	char buf[32];
1583 	int nh;
1584 
1585 	if (m->prev->off != 0) {
1586 		err = packhdr(&nh, buf, sizeof(buf),
1587 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1588 		if (err)
1589 			return err;
1590 		nh += packoff(buf + nh, m->off - m->prev->off);
1591 		err = hwrite(packfile, buf, nh, ctx);
1592 		if (err)
1593 			return err;
1594 		*packfile_size += nh;
1595 	} else {
1596 		err = packhdr(&nh, buf, sizeof(buf),
1597 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1598 		if (err)
1599 			return err;
1600 		err = hwrite(packfile, buf, nh, ctx);
1601 		if (err)
1602 			return err;
1603 		*packfile_size += nh;
1604 		err = hwrite(packfile, m->prev->id.sha1,
1605 		    sizeof(m->prev->id.sha1), ctx);
1606 		if (err)
1607 			return err;
1608 		*packfile_size += sizeof(m->prev->id.sha1);
1609 	}
1610 
1611 	return NULL;
1612 }
1613 
1614 static const struct got_error *
1615 write_packed_object(off_t *packfile_size, FILE *packfile,
1616     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1617     SHA1_CTX *ctx, struct got_repository *repo)
1618 {
1619 	const struct got_error *err = NULL;
1620 	struct got_deflate_checksum csum;
1621 	char buf[32];
1622 	int nh;
1623 	struct got_raw_object *raw = NULL;
1624 	off_t outlen;
1625 
1626 	csum.output_sha1 = ctx;
1627 	csum.output_crc = NULL;
1628 
1629 	m->off = ftello(packfile);
1630 	if (m->delta_len == 0) {
1631 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1632 		if (err)
1633 			goto done;
1634 		err = packhdr(&nh, buf, sizeof(buf),
1635 		    m->obj_type, raw->size);
1636 		if (err)
1637 			goto done;
1638 		err = hwrite(packfile, buf, nh, ctx);
1639 		if (err)
1640 			goto done;
1641 		*packfile_size += nh;
1642 		if (raw->f == NULL) {
1643 			err = got_deflate_to_file_mmap(&outlen,
1644 			    raw->data + raw->hdrlen, 0, raw->size,
1645 			    packfile, &csum);
1646 			if (err)
1647 				goto done;
1648 		} else {
1649 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1650 			    == -1) {
1651 				err = got_error_from_errno("fseeko");
1652 				goto done;
1653 			}
1654 			err = got_deflate_to_file(&outlen, raw->f,
1655 			    raw->size, packfile, &csum);
1656 			if (err)
1657 				goto done;
1658 		}
1659 		*packfile_size += outlen;
1660 		got_object_raw_close(raw);
1661 		raw = NULL;
1662 	} else if (m->delta_buf) {
1663 		err = deltahdr(packfile_size, ctx, packfile, m);
1664 		if (err)
1665 			goto done;
1666 		err = got_deflate_to_file_mmap(&outlen,
1667 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1668 		if (err)
1669 			goto done;
1670 		*packfile_size += outlen;
1671 		free(m->delta_buf);
1672 		m->delta_buf = NULL;
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, packfile, m);
1680 		if (err)
1681 			goto done;
1682 		err = got_deflate_to_file(&outlen, delta_cache,
1683 		    m->delta_len, packfile, &csum);
1684 		if (err)
1685 			goto done;
1686 		*packfile_size += outlen;
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, FILE *packfile, 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 	size_t n;
1710 	off_t packfile_size = 0;
1711 	int outfd = -1;
1712 
1713 	SHA1Init(&ctx);
1714 
1715 	err = hwrite(packfile, "PACK", 4, &ctx);
1716 	if (err)
1717 		return err;
1718 	putbe32(buf, GOT_PACKFILE_VERSION);
1719 	err = hwrite(packfile, buf, 4, &ctx);
1720 	if (err)
1721 		goto done;
1722 	putbe32(buf, ndeltify + nreuse);
1723 	err = hwrite(packfile, buf, 4, &ctx);
1724 	if (err)
1725 		goto done;
1726 
1727 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1728 	    write_order_cmp);
1729 	for (i = 0; i < ndeltify; i++) {
1730 		err = report_progress(progress_cb, progress_arg, rl,
1731 		    ncolored, nfound, ntrees, packfile_size, nours,
1732 		    ndeltify + nreuse, ndeltify + nreuse, i);
1733 		if (err)
1734 			goto done;
1735 		m = deltify[i];
1736 		err = write_packed_object(&packfile_size, packfile,
1737 		    delta_cache, m, &outfd, &ctx, repo);
1738 		if (err)
1739 			goto done;
1740 	}
1741 
1742 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1743 	    reuse_write_order_cmp);
1744 	for (i = 0; i < nreuse; i++) {
1745 		err = report_progress(progress_cb, progress_arg, rl,
1746 		    ncolored, nfound, ntrees, packfile_size, nours,
1747 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1748 		if (err)
1749 			goto done;
1750 		m = reuse[i];
1751 		err = write_packed_object(&packfile_size, packfile,
1752 		    delta_cache, m, &outfd, &ctx, repo);
1753 		if (err)
1754 			goto done;
1755 	}
1756 
1757 	SHA1Final(pack_sha1, &ctx);
1758 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1759 	if (n != SHA1_DIGEST_LENGTH)
1760 		err = got_ferror(packfile, GOT_ERR_IO);
1761 	packfile_size += SHA1_DIGEST_LENGTH;
1762 	packfile_size += sizeof(struct got_packfile_hdr);
1763 	if (progress_cb) {
1764 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1765 		    packfile_size, nours, ndeltify + nreuse,
1766 		    ndeltify + nreuse, ndeltify + nreuse);
1767 		if (err)
1768 			goto done;
1769 	}
1770 done:
1771 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1772 		err = got_error_from_errno("close");
1773 	return err;
1774 }
1775 
1776 static const struct got_error *
1777 remove_unused_object(struct got_object_idset_element *entry, void *arg)
1778 {
1779 	struct got_object_idset *idset = arg;
1780 
1781 	if (got_object_idset_get_element_data(entry) == NULL)
1782 		got_object_idset_remove_element(idset, entry);
1783 
1784 	return NULL;
1785 }
1786 
1787 static const struct got_error *
1788 remove_reused_object(struct got_object_idset_element *entry, void *arg)
1789 {
1790 	struct got_object_idset *idset = arg;
1791 	struct got_pack_meta *m;
1792 
1793 	m = got_object_idset_get_element_data(entry);
1794 	if (m->have_reused_delta)
1795 		got_object_idset_remove_element(idset, entry);
1796 
1797 	return NULL;
1798 }
1799 
1800 static const struct got_error *
1801 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1802 {
1803 	struct got_pack_meta *m = data;
1804 	struct got_pack_metavec *v = arg;
1805 
1806 	return add_meta(m, v);
1807 }
1808 
1809 const struct got_error *
1810 got_pack_create(uint8_t *packsha1, FILE *packfile,
1811     struct got_object_id **theirs, int ntheirs,
1812     struct got_object_id **ours, int nours,
1813     struct got_repository *repo, int loose_obj_only, int allow_empty,
1814     got_pack_progress_cb progress_cb, void *progress_arg,
1815     got_cancel_cb cancel_cb, void *cancel_arg)
1816 {
1817 	const struct got_error *err;
1818 	int delta_cache_fd = -1;
1819 	FILE *delta_cache = NULL;
1820 	struct got_object_idset *idset;
1821 	struct got_ratelimit rl;
1822 	struct got_pack_metavec deltify, reuse;
1823 	int ncolored = 0, nfound = 0, ntrees = 0;
1824 
1825 	memset(&deltify, 0, sizeof(deltify));
1826 	memset(&reuse, 0, sizeof(reuse));
1827 
1828 	got_ratelimit_init(&rl, 0, 500);
1829 
1830 	idset = got_object_idset_alloc();
1831 	if (idset == NULL)
1832 		return got_error_from_errno("got_object_idset_alloc");
1833 
1834 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1835 	    ntheirs, ours, nours, repo, loose_obj_only,
1836 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1837 	if (err)
1838 		return err;
1839 
1840 	err = got_object_idset_for_each_element(idset,
1841 	    remove_unused_object, idset);
1842 	if (err)
1843 		goto done;
1844 
1845 	if (progress_cb) {
1846 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1847 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1848 		if (err)
1849 			goto done;
1850 	}
1851 
1852 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1853 		err = got_error(GOT_ERR_CANNOT_PACK);
1854 		goto done;
1855 	}
1856 
1857 	delta_cache_fd = got_opentempfd();
1858 	if (delta_cache_fd == -1) {
1859 		err = got_error_from_errno("got_opentemp");
1860 		goto done;
1861 	}
1862 
1863 	reuse.metasz = 64;
1864 	reuse.meta = calloc(reuse.metasz,
1865 	    sizeof(struct got_pack_meta *));
1866 	if (reuse.meta == NULL) {
1867 		err = got_error_from_errno("calloc");
1868 		goto done;
1869 	}
1870 
1871 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1872 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1873 	    cancel_cb, cancel_arg);
1874 	if (err)
1875 		goto done;
1876 	if (reuse.nmeta > 0) {
1877 		err = got_object_idset_for_each_element(idset,
1878 		    remove_reused_object, idset);
1879 		if (err)
1880 			goto done;
1881 	}
1882 
1883 	delta_cache = fdopen(delta_cache_fd, "a+");
1884 	if (delta_cache == NULL) {
1885 		err = got_error_from_errno("fdopen");
1886 		goto done;
1887 	}
1888 	delta_cache_fd = -1;
1889 
1890 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1891 		err = got_error_from_errno("fseeko");
1892 		goto done;
1893 	}
1894 
1895 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1896 	    sizeof(struct got_pack_meta *));
1897 	if (deltify.meta == NULL) {
1898 		err = got_error_from_errno("calloc");
1899 		goto done;
1900 	}
1901 	deltify.metasz = got_object_idset_num_elements(idset);
1902 
1903 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1904 	if (err)
1905 		goto done;
1906 	if (deltify.nmeta > 0) {
1907 		err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1908 		    nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1909 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1910 		if (err)
1911 			goto done;
1912 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1913 			err = got_error_from_errno("fseeko");
1914 			goto done;
1915 		}
1916 	}
1917 
1918 	err = genpack(packsha1, packfile, 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 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1931 		err = got_error_from_errno("fclose");
1932 	return err;
1933 }