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/uio.h>
20 #include <sys/stat.h>
21 #include <sys/time.h>
22 
23 #include <stdint.h>
24 #include <stdio.h>
25 #include <stdlib.h>
26 #include <string.h>
27 #include <time.h>
28 #include <unistd.h>
29 #include <limits.h>
30 #include <zlib.h>
31 
32 #if defined(__FreeBSD__)
33 #include <unistd.h>
34 #endif
35 
36 #include "got_error.h"
37 #include "got_cancel.h"
38 #include "got_object.h"
39 #include "got_path.h"
40 #include "got_reference.h"
41 #include "got_repository_admin.h"
42 #include "got_opentemp.h"
43 
44 #include "got_lib_deltify.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_idset.h"
48 #include "got_lib_object_cache.h"
49 #include "got_lib_deflate.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_privsep.h"
52 #include "got_lib_repository.h"
53 #include "got_lib_ratelimit.h"
54 
55 #ifndef MIN
56 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 #endif
58 
59 #ifndef MAX
60 #define	MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
61 #endif
62 
63 #ifndef nitems
64 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
65 #endif
66 
67 struct got_pack_meta {
68 	struct got_object_id id;
69 	char	*path;
70 	int	obj_type;
71 	off_t	size;
72 	time_t	mtime;
73 
74 	/* The best delta we picked */
75 	struct got_pack_meta *head;
76 	struct got_pack_meta *prev;
77 	unsigned char *delta_buf; /* if not encoded in delta cache file */
78 	off_t	delta_offset;	/* offset in delta cache file */
79 	off_t	delta_len;	/* encoded delta length */
80 	int	nchain;
81 
82 	int	have_reused_delta;
83 	off_t   reused_delta_offset; /* offset of delta in reused pack file */
84 	struct got_object_id *base_obj_id;
85 
86 	/* Only used for delta window */
87 	struct got_delta_table *dtab;
88 
89 	/* Only used for writing offset deltas */
90 	off_t	off;
91 };
92 
93 struct got_pack_metavec {
94 	struct got_pack_meta **meta;
95 	int nmeta;
96 	int metasz;
97 };
98 
99 static const struct got_error *
100 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
101     const char *path, int obj_type, time_t mtime)
102 {
103 	const struct got_error *err = NULL;
104 	struct got_pack_meta *m;
105 
106 	*new = NULL;
107 
108 	m = calloc(1, sizeof(*m));
109 	if (m == NULL)
110 		return got_error_from_errno("calloc");
111 
112 	memcpy(&m->id, id, sizeof(m->id));
113 
114 	m->path = strdup(path);
115 	if (m->path == NULL) {
116 		err = got_error_from_errno("strdup");
117 		free(m);
118 		return err;
119 	}
120 
121 	m->obj_type = obj_type;
122 	m->mtime = mtime;
123 	*new = m;
124 	return NULL;
125 }
126 
127 static void
128 clear_meta(struct got_pack_meta *meta)
129 {
130 	if (meta == NULL)
131 		return;
132 	free(meta->path);
133 	meta->path = NULL;
134 	free(meta->delta_buf);
135 	meta->delta_buf = NULL;
136 	free(meta->base_obj_id);
137 	meta->base_obj_id = NULL;
138 }
139 
140 static void
141 free_nmeta(struct got_pack_meta **meta, int nmeta)
142 {
143 	int i;
144 
145 	for (i = 0; i < nmeta; i++)
146 		clear_meta(meta[i]);
147 	free(meta);
148 }
149 
150 static int
151 delta_order_cmp(const void *pa, const void *pb)
152 {
153 	struct got_pack_meta *a, *b;
154 	int cmp;
155 
156 	a = *(struct got_pack_meta **)pa;
157 	b = *(struct got_pack_meta **)pb;
158 
159 	if (a->obj_type != b->obj_type)
160 		return a->obj_type - b->obj_type;
161 	cmp = strcmp(a->path, b->path);
162 	if (cmp != 0)
163 		return cmp;
164 	if (a->mtime != b->mtime)
165 		return a->mtime - b->mtime;
166 	return got_object_id_cmp(&a->id, &b->id);
167 }
168 
169 static off_t
170 delta_size(struct got_delta_instruction *deltas, int ndeltas)
171 {
172 	int i;
173 	off_t size = 32;
174 	for (i = 0; i < ndeltas; i++) {
175 		if (deltas[i].copy)
176 			size += GOT_DELTA_SIZE_SHIFT;
177 		else
178 			size += deltas[i].len + 1;
179 	}
180 	return size;
181 }
182 
183 static const struct got_error *
184 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
185 {
186 	char *n;
187 
188 	if (*len + nseg >= *sz) {
189 		while (*len + nseg >= *sz)
190 			*sz += *sz / 2;
191 		n = realloc(*p, *sz);
192 		if (n == NULL)
193 			return got_error_from_errno("realloc");
194 		*p = n;
195 	}
196 	memcpy(*p + *len, seg, nseg);
197 	*len += nseg;
198 	return NULL;
199 }
200 
201 static const struct got_error *
202 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
203     struct got_delta_instruction *deltas, int ndeltas,
204     off_t delta_size, off_t base_size)
205 {
206 	const struct got_error *err;
207 	unsigned char buf[16], *bp;
208 	int i, j;
209 	size_t len = 0;
210 	off_t n;
211 	struct got_delta_instruction *d;
212 
213 	m->delta_buf = malloc(delta_size);
214 	if (m->delta_buf == NULL)
215 		return got_error_from_errno("calloc");
216 
217 	/* base object size */
218 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
219 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
220 	for (i = 1; n > 0; i++) {
221 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
222 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
223 		n >>= GOT_DELTA_SIZE_SHIFT;
224 	}
225 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
226 	if (err)
227 		return err;
228 
229 	/* target object size */
230 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
231 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
232 	for (i = 1; n > 0; i++) {
233 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
234 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
235 		n >>= GOT_DELTA_SIZE_SHIFT;
236 	}
237 	err = append(&m->delta_buf, &len, &delta_size, buf, i);
238 	if (err)
239 		return err;
240 
241 	for (j = 0; j < ndeltas; j++) {
242 		d = &deltas[j];
243 		if (d->copy) {
244 			n = d->offset;
245 			bp = &buf[1];
246 			buf[0] = GOT_DELTA_BASE_COPY;
247 			for (i = 0; i < 4; i++) {
248 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
249 				buf[0] |= 1 << i;
250 				*bp++ = n & 0xff;
251 				n >>= 8;
252 				if (n == 0)
253 					break;
254 			}
255 
256 			n = d->len;
257 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
258 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
259 				for (i = 0; i < 3 && n > 0; i++) {
260 					buf[0] |= 1 << (i + 4);
261 					*bp++ = n & 0xff;
262 					n >>= 8;
263 				}
264 			}
265 			err = append(&m->delta_buf, &len, &delta_size,
266 			    buf, bp - buf);
267 			if (err)
268 				return err;
269 		} else if (o->f == NULL) {
270 			n = 0;
271 			while (n != d->len) {
272 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
273 				err = append(&m->delta_buf, &len, &delta_size,
274 				    buf, 1);
275 				if (err)
276 					return err;
277 				err = append(&m->delta_buf, &len, &delta_size,
278 				    o->data + o->hdrlen + d->offset + n,
279 				    buf[0]);
280 				if (err)
281 					return err;
282 				n += buf[0];
283 			}
284 		} else {
285 			char content[128];
286 			size_t r;
287 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
288 				return got_error_from_errno("fseeko");
289 			n = 0;
290 			while (n != d->len) {
291 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
292 				err = append(&m->delta_buf, &len, &delta_size,
293 				    buf, 1);
294 				if (err)
295 					return err;
296 				r = fread(content, 1, buf[0], o->f);
297 				if (r != buf[0])
298 					return got_ferror(o->f, GOT_ERR_IO);
299 				err = append(&m->delta_buf, &len, &delta_size,
300 				    content, buf[0]);
301 				if (err)
302 					return err;
303 				n += buf[0];
304 			}
305 		}
306 	}
307 
308 	m->delta_len = len;
309 	return NULL;
310 }
311 
312 static const struct got_error *
313 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
314     struct got_delta_instruction *deltas, int ndeltas,
315     off_t base_size, FILE *f)
316 {
317 	unsigned char buf[16], *bp;
318 	int i, j;
319 	off_t n;
320 	size_t w;
321 	struct got_delta_instruction *d;
322 
323 	/* base object size */
324 	buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
325 	n = base_size >> GOT_DELTA_SIZE_SHIFT;
326 	for (i = 1; n > 0; i++) {
327 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
328 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
329 		n >>= GOT_DELTA_SIZE_SHIFT;
330 	}
331 	w = fwrite(buf, 1, i, f);
332 	if (w != i)
333 		return got_ferror(f, GOT_ERR_IO);
334 
335 	/* target object size */
336 	buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
337 	n = o->size >> GOT_DELTA_SIZE_SHIFT;
338 	for (i = 1; n > 0; i++) {
339 		buf[i - 1] |= GOT_DELTA_SIZE_MORE;
340 		buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
341 		n >>= GOT_DELTA_SIZE_SHIFT;
342 	}
343 	w = fwrite(buf, 1, i, f);
344 	if (w != i)
345 		return got_ferror(f, GOT_ERR_IO);
346 
347 	for (j = 0; j < ndeltas; j++) {
348 		d = &deltas[j];
349 		if (d->copy) {
350 			n = d->offset;
351 			bp = &buf[1];
352 			buf[0] = GOT_DELTA_BASE_COPY;
353 			for (i = 0; i < 4; i++) {
354 				/* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
355 				buf[0] |= 1 << i;
356 				*bp++ = n & 0xff;
357 				n >>= 8;
358 				if (n == 0)
359 					break;
360 			}
361 
362 			n = d->len;
363 			if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
364 				/* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
365 				for (i = 0; i < 3 && n > 0; i++) {
366 					buf[0] |= 1 << (i + 4);
367 					*bp++ = n & 0xff;
368 					n >>= 8;
369 				}
370 			}
371 			w = fwrite(buf, 1, bp - buf, f);
372 			if (w != bp - buf)
373 				return got_ferror(f, GOT_ERR_IO);
374 		} else if (o->f == NULL) {
375 			n = 0;
376 			while (n != d->len) {
377 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
378 				w = fwrite(buf, 1, 1, f);
379 				if (w != 1)
380 					return got_ferror(f, GOT_ERR_IO);
381 				w = fwrite(o->data + o->hdrlen + d->offset + n,
382 				    1, buf[0], f);
383 				if (w != buf[0])
384 					return got_ferror(f, GOT_ERR_IO);
385 				n += buf[0];
386 			}
387 		} else {
388 			char content[128];
389 			size_t r;
390 			if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1)
391 				return got_error_from_errno("fseeko");
392 			n = 0;
393 			while (n != d->len) {
394 				buf[0] = (d->len - n < 127) ? d->len - n : 127;
395 				w = fwrite(buf, 1, 1, f);
396 				if (w != 1)
397 					return got_ferror(f, GOT_ERR_IO);
398 				r = fread(content, 1, buf[0], o->f);
399 				if (r != buf[0])
400 					return got_ferror(o->f, GOT_ERR_IO);
401 				w = fwrite(content, 1, buf[0], f);
402 				if (w != buf[0])
403 					return got_ferror(f, GOT_ERR_IO);
404 				n += buf[0];
405 			}
406 		}
407 	}
408 
409 	m->delta_len = ftello(f) - m->delta_offset;
410 	return NULL;
411 }
412 
413 static const struct got_error *
414 report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
415     struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
416     off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
417     int nobj_written)
418 {
419 	const struct got_error *err;
420 	int elapsed;
421 
422 	if (progress_cb == NULL)
423 		return NULL;
424 
425 	err = got_ratelimit_check(&elapsed, rl);
426 	if (err || !elapsed)
427 		return err;
428 
429 	return progress_cb(progress_arg, ncolored, nfound, ntrees,
430 	    packfile_size, ncommits, nobj_total, obj_deltify, nobj_written);
431 }
432 
433 static const struct got_error *
434 add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
435 {
436 	if (v->nmeta == v->metasz){
437 		size_t newsize = 2 * v->metasz;
438 		struct got_pack_meta **new;
439 		new = reallocarray(v->meta, newsize, sizeof(*new));
440 		if (new == NULL)
441 			return got_error_from_errno("reallocarray");
442 		v->meta = new;
443 		v->metasz = newsize; 
444 	}
445 
446 	v->meta[v->nmeta++] = m;
447 	return NULL;
448 }
449 
450 static const struct got_error *
451 reuse_delta(int idx, struct got_pack_meta *m, struct got_pack_metavec *v,
452     struct got_object_idset *idset, struct got_pack *pack,
453     struct got_packidx *packidx, int delta_cache_fd,
454     struct got_repository *repo)
455 {
456 	const struct got_error *err = NULL;
457 	struct got_pack_meta *base = NULL;
458 	struct got_object_id *base_obj_id = NULL;
459 	off_t delta_len = 0, delta_offset = 0, delta_cache_offset = 0;
460 	uint64_t base_size, result_size;
461 
462 	if (m->have_reused_delta)
463 		return NULL;
464 
465 	err = got_object_read_raw_delta(&base_size, &result_size, &delta_len,
466 	    &delta_offset, &delta_cache_offset, &base_obj_id, delta_cache_fd,
467 	    packidx, idx, &m->id, repo);
468 	if (err)
469 		return err;
470 
471 	if (delta_offset + delta_len < delta_offset)
472 		return got_error(GOT_ERR_BAD_PACKFILE);
473 
474 	base = got_object_idset_get(idset, base_obj_id);
475 	if (base == NULL)
476 		goto done;
477 
478 	m->delta_len = delta_len;
479 	m->delta_offset = delta_cache_offset;
480 	m->prev = base;
481 	m->size = result_size;
482 	m->have_reused_delta = 1;
483 	m->reused_delta_offset = delta_offset;
484 	m->base_obj_id = base_obj_id;
485 	base_obj_id = NULL;
486 	err = add_meta(m, v);
487 done:
488 	free(base_obj_id);
489 	return err;
490 }
491 
492 static const struct got_error *
493 find_pack_for_reuse(struct got_packidx **best_packidx,
494     struct got_repository *repo)
495 {
496 	const struct got_error *err = NULL;
497 	struct got_pathlist_entry *pe;
498 	const char *best_packidx_path = NULL;
499 	int nobj_max = 0;
500 
501 	*best_packidx = NULL;
502 
503 	TAILQ_FOREACH(pe, &repo->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 	return err;
525 }
526 
527 struct search_deltas_arg {
528 	struct got_packidx *packidx;
529 	struct got_pack *pack;
530 	struct got_object_idset *idset;
531 	struct got_pack_metavec *v;
532 	int delta_cache_fd;
533 	struct got_repository *repo;
534 	got_pack_progress_cb progress_cb;
535 	void *progress_arg;
536 	struct got_ratelimit *rl;
537 	got_cancel_cb cancel_cb;
538 	void *cancel_arg;
539 	int ncolored;
540 	int nfound;
541 	int ntrees;
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 		    a->ncolored, a->nfound, a->ntrees, 0L, a->ncommits,
588 		    got_object_idset_num_elements(a->idset), 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 ncolored, int nfound, int ntrees, int ncommits,
598     struct got_repository *repo,
599     got_pack_progress_cb progress_cb, void *progress_arg,
600     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
601 {
602 	const struct got_error *err = NULL;
603 	char *path_packfile = NULL;
604 	struct got_packidx *packidx;
605 	struct got_pack *pack;
606 	struct search_deltas_arg sda;
607 
608 	err = find_pack_for_reuse(&packidx, repo);
609 	if (err)
610 		return err;
611 
612 	if (packidx == NULL)
613 		return NULL;
614 
615 	err = got_packidx_get_packfile_path(&path_packfile,
616 	    packidx->path_packidx);
617 	if (err)
618 		return err;
619 
620 	pack = got_repo_get_cached_pack(repo, path_packfile);
621 	if (pack == NULL) {
622 		err = got_repo_cache_pack(&pack, repo, path_packfile, packidx);
623 		if (err)
624 			goto done;
625 	}
626 
627 	sda.packidx = packidx;
628 	sda.pack = pack;
629 	sda.idset = idset;
630 	sda.v = v;
631 	sda.delta_cache_fd = delta_cache_fd;
632 	sda.repo = repo;
633 	sda.progress_cb = progress_cb;
634 	sda.progress_arg = progress_arg;
635 	sda.rl = rl;
636 	sda.cancel_cb = cancel_cb;
637 	sda.cancel_arg = cancel_arg;
638 	sda.ncolored = ncolored;
639 	sda.nfound = nfound;
640 	sda.ntrees = ntrees;
641 	sda.ncommits = ncommits;
642 	err = got_object_idset_for_each(idset, search_delta_for_object, &sda);
643 done:
644 	free(path_packfile);
645 	return err;
646 }
647 
648 static const struct got_error *
649 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
650     int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
651     struct got_repository *repo,
652     got_pack_progress_cb progress_cb, void *progress_arg,
653     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
654 {
655 	const struct got_error *err = NULL;
656 	struct got_pack_meta *m = NULL, *base = NULL;
657 	struct got_raw_object *raw = NULL, *base_raw = NULL;
658 	struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
659 	int i, j, ndeltas, best_ndeltas;
660 	off_t size, best_size;
661 	const int max_base_candidates = 3;
662 	size_t delta_memsize = 0;
663 	const size_t max_delta_memsize = 25 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
664 	int outfd = -1;
665 
666 	qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
667 	for (i = 0; i < nmeta; i++) {
668 		if (cancel_cb) {
669 			err = (*cancel_cb)(cancel_arg);
670 			if (err)
671 				break;
672 		}
673 		err = report_progress(progress_cb, progress_arg, rl,
674 		    ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
675 		    nreused + i, 0);
676 		if (err)
677 			goto done;
678 		m = meta[i];
679 
680 		if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
681 		    m->obj_type == GOT_OBJ_TYPE_TAG)
682 			continue;
683 
684 		err = got_object_raw_open(&raw, &outfd, repo, &m->id);
685 		if (err)
686 			goto done;
687 		m->size = raw->size;
688 
689 		if (raw->f == NULL) {
690 			err = got_deltify_init_mem(&m->dtab, raw->data,
691 			    raw->hdrlen, raw->size + raw->hdrlen);
692 		} else {
693 			err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
694 			    raw->size + raw->hdrlen);
695 		}
696 		if (err)
697 			goto done;
698 
699 		if (i > max_base_candidates) {
700 			struct got_pack_meta *n = NULL;
701 			n = meta[i - (max_base_candidates + 1)];
702 			got_deltify_free(n->dtab);
703 			n->dtab = NULL;
704 		}
705 
706 		best_size = raw->size;
707 		best_ndeltas = 0;
708 		for (j = MAX(0, i - max_base_candidates); j < i; j++) {
709 			if (cancel_cb) {
710 				err = (*cancel_cb)(cancel_arg);
711 				if (err)
712 					goto done;
713 			}
714 			base = meta[j];
715 			/* long chains make unpacking slow, avoid such bases */
716 			if (base->nchain >= 128 ||
717 			    base->obj_type != m->obj_type)
718 				continue;
719 
720 			err = got_object_raw_open(&base_raw, &outfd, repo,
721 			    &base->id);
722 			if (err)
723 				goto done;
724 
725 			if (raw->f == NULL && base_raw->f == NULL) {
726 				err = got_deltify_mem_mem(&deltas, &ndeltas,
727 				    raw->data, raw->hdrlen,
728 				    raw->size + raw->hdrlen,
729 				    base->dtab, base_raw->data,
730 				    base_raw->hdrlen,
731 				    base_raw->size + base_raw->hdrlen);
732 			} else if (raw->f == NULL) {
733 				err = got_deltify_mem_file(&deltas, &ndeltas,
734 				    raw->data, raw->hdrlen,
735 				    raw->size + raw->hdrlen,
736 				    base->dtab, base_raw->f,
737 				    base_raw->hdrlen,
738 				    base_raw->size + base_raw->hdrlen);
739 			} else if (base_raw->f == NULL) {
740 				err = got_deltify_file_mem(&deltas, &ndeltas,
741 				    raw->f, raw->hdrlen,
742 				    raw->size + raw->hdrlen,
743 				    base->dtab, base_raw->data,
744 				    base_raw->hdrlen,
745 				    base_raw->size + base_raw->hdrlen);
746 			} else {
747 				err = got_deltify(&deltas, &ndeltas,
748 				    raw->f, raw->hdrlen,
749 				    raw->size + raw->hdrlen,
750 				    base->dtab, base_raw->f, base_raw->hdrlen,
751 				    base_raw->size + base_raw->hdrlen);
752 			}
753 			got_object_raw_close(base_raw);
754 			base_raw = NULL;
755 			if (err)
756 				goto done;
757 
758 			size = delta_size(deltas, ndeltas);
759 			if (size + 32 < best_size){
760 				/*
761 				 * if we already picked a best delta,
762 				 * replace it.
763 				 */
764 				best_size = size;
765 				free(best_deltas);
766 				best_deltas = deltas;
767 				best_ndeltas = ndeltas;
768 				deltas = NULL;
769 				m->nchain = base->nchain + 1;
770 				m->prev = base;
771 				m->head = base->head;
772 				if (m->head == NULL)
773 					m->head = base;
774 			} else {
775 				free(deltas);
776 				deltas = NULL;
777 				ndeltas = 0;
778 			}
779 		}
780 
781 		if (best_ndeltas > 0) {
782 			if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
783 			    delta_memsize + best_size <= max_delta_memsize) {
784 				delta_memsize += best_size;
785 				err = encode_delta_in_mem(m, raw, best_deltas,
786 				    best_ndeltas, best_size, m->prev->size);
787 			} else {
788 				m->delta_offset = ftello(delta_cache);
789 				/*
790 				 * TODO:
791 				 * Storing compressed delta data in the delta
792 				 * cache file would probably be more efficient
793 				 * than writing uncompressed delta data here
794 				 * and compressing it while writing the pack
795 				 * file. This would also allow for reusing
796 				 * deltas in their compressed form.
797 				 */
798 				err = encode_delta(m, raw, best_deltas,
799 				    best_ndeltas, m->prev->size, delta_cache);
800 			}
801 			free(best_deltas);
802 			best_deltas = NULL;
803 			best_ndeltas = 0;
804 			if (err)
805 				goto done;
806 		}
807 
808 		got_object_raw_close(raw);
809 		raw = NULL;
810 	}
811 done:
812 	for (i = MAX(0, nmeta - max_base_candidates); i < nmeta; i++) {
813 		got_deltify_free(meta[i]->dtab);
814 		meta[i]->dtab = NULL;
815 	}
816 	if (raw)
817 		got_object_raw_close(raw);
818 	if (base_raw)
819 		got_object_raw_close(base_raw);
820 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
821 		err = got_error_from_errno("close");
822 	free(deltas);
823 	free(best_deltas);
824 	return err;
825 }
826 
827 static const struct got_error *
828 search_packidx(int *found, struct got_object_id *id,
829     struct got_repository *repo)
830 {
831 	const struct got_error *err = NULL;
832 	struct got_packidx *packidx = NULL;
833 	int idx;
834 
835 	*found = 0;
836 
837 	err = got_repo_search_packidx(&packidx, &idx, repo, id);
838 	if (err == NULL)
839 		*found = 1; /* object is already packed */
840 	else if (err->code == GOT_ERR_NO_OBJ)
841 		err = NULL;
842 	return err;
843 }
844 
845 static const struct got_error *
846 add_object(int want_meta, struct got_object_idset *idset,
847     struct got_object_id *id, const char *path, int obj_type,
848     time_t mtime, int loose_obj_only, struct got_repository *repo,
849     int *ncolored, int *nfound, int *ntrees,
850     got_pack_progress_cb progress_cb, void *progress_arg,
851     struct got_ratelimit *rl)
852 {
853 	const struct got_error *err;
854 	struct got_pack_meta *m = NULL;
855 
856 	if (loose_obj_only) {
857 		int is_packed;
858 		err = search_packidx(&is_packed, id, repo);
859 		if (err)
860 			return err;
861 		if (is_packed && want_meta)
862 			return NULL;
863 	}
864 
865 	if (want_meta) {
866 		err = alloc_meta(&m, id, path, obj_type, mtime);
867 		if (err)
868 			return err;
869 
870 		(*nfound)++;
871 		err = report_progress(progress_cb, progress_arg, rl,
872 		    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
873 		if (err) {
874 			clear_meta(m);
875 			free(m);
876 			return err;
877 		}
878 	}
879 
880 	err = got_object_idset_add(idset, id, m);
881 	if (err) {
882 		clear_meta(m);
883 		free(m);
884 	}
885 	return err;
886 }
887 
888 static const struct got_error *
889 load_tree_entries(struct got_object_id_queue *ids, int want_meta,
890     struct got_object_idset *idset, struct got_object_id *tree_id,
891     const char *dpath, time_t mtime, struct got_repository *repo,
892     int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
893     got_pack_progress_cb progress_cb, void *progress_arg,
894     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
895 {
896 	const struct got_error *err;
897 	struct got_tree_object *tree;
898 	char *p = NULL;
899 	int i;
900 
901 	err = got_object_open_as_tree(&tree, repo, tree_id);
902 	if (err)
903 		return err;
904 
905 	(*ntrees)++;
906 	err = report_progress(progress_cb, progress_arg, rl,
907 	    *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0);
908 	if (err)
909 		return err;
910 
911 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
912 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
913 		struct got_object_id *id = got_tree_entry_get_id(e);
914 		mode_t mode = got_tree_entry_get_mode(e);
915 
916 		if (cancel_cb) {
917 			err = (*cancel_cb)(cancel_arg);
918 			if (err)
919 				break;
920 		}
921 
922 		if (got_object_tree_entry_is_submodule(e) ||
923 		    got_object_idset_contains(idset, id))
924 			continue;
925 		
926 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
927 		    got_tree_entry_get_name(e)) == -1) {
928 			err = got_error_from_errno("asprintf");
929 			break;
930 		}
931 
932 		if (S_ISDIR(mode)) {
933 			struct got_object_qid *qid;
934 			err = got_object_qid_alloc(&qid, id);
935 			if (err)
936 				break;
937 			STAILQ_INSERT_TAIL(ids, qid, entry);
938 		} else if (S_ISREG(mode) || S_ISLNK(mode)) {
939 			err = add_object(want_meta, idset, id, p,
940 			    GOT_OBJ_TYPE_BLOB, mtime, loose_obj_only, repo,
941 			    ncolored, nfound, ntrees,
942 			    progress_cb, progress_arg, rl);
943 			if (err)
944 				break;
945 		}
946 		free(p);
947 		p = NULL;
948 	}
949 
950 	got_object_tree_close(tree);
951 	free(p);
952 	return err;
953 }
954 
955 static const struct got_error *
956 load_tree(int want_meta, struct got_object_idset *idset,
957     struct got_object_id *tree_id, const char *dpath, time_t mtime,
958     struct got_repository *repo, int loose_obj_only,
959     int *ncolored, int *nfound, int *ntrees,
960     got_pack_progress_cb progress_cb, void *progress_arg,
961     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
962 {
963 	const struct got_error *err = NULL;
964 	struct got_object_id_queue tree_ids;
965 	struct got_object_qid *qid;
966 
967 	if (got_object_idset_contains(idset, tree_id))
968 		return NULL;
969 
970 	err = got_object_qid_alloc(&qid, tree_id);
971 	if (err)
972 		return err;
973 
974 	STAILQ_INIT(&tree_ids);
975 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
976 
977 	while (!STAILQ_EMPTY(&tree_ids)) {
978 		if (cancel_cb) {
979 			err = (*cancel_cb)(cancel_arg);
980 			if (err)
981 				break;
982 		}
983 
984 		qid = STAILQ_FIRST(&tree_ids);
985 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
986 
987 		if (got_object_idset_contains(idset, &qid->id)) {
988 			got_object_qid_free(qid);
989 			continue;
990 		}
991 
992 		err = add_object(want_meta, idset, &qid->id, dpath,
993 		    GOT_OBJ_TYPE_TREE, mtime, loose_obj_only, repo,
994 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
995 		if (err) {
996 			got_object_qid_free(qid);
997 			break;
998 		}
999 
1000 		err = load_tree_entries(&tree_ids, want_meta, idset, &qid->id,
1001 		    dpath, mtime, repo, loose_obj_only, ncolored, nfound,
1002 		    ntrees, progress_cb, progress_arg, rl,
1003 		    cancel_cb, cancel_arg);
1004 		got_object_qid_free(qid);
1005 		if (err)
1006 			break;
1007 	}
1008 
1009 	got_object_id_queue_free(&tree_ids);
1010 	return err;
1011 }
1012 
1013 static const struct got_error *
1014 load_commit(int want_meta, struct got_object_idset *idset,
1015     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1016     int *ncolored, int *nfound, int *ntrees,
1017     got_pack_progress_cb progress_cb, void *progress_arg,
1018     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1019 {
1020 	const struct got_error *err;
1021 	struct got_commit_object *commit;
1022 
1023 	if (got_object_idset_contains(idset, id))
1024 		return NULL;
1025 
1026 	if (loose_obj_only) {
1027 		int is_packed;
1028 		err = search_packidx(&is_packed, id, repo);
1029 		if (err)
1030 			return err;
1031 		if (is_packed && want_meta)
1032 			return NULL;
1033 	}
1034 
1035 	err = got_object_open_as_commit(&commit, repo, id);
1036 	if (err)
1037 		return err;
1038 
1039 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_COMMIT,
1040 	    got_object_commit_get_committer_time(commit),
1041 	    loose_obj_only, repo,
1042 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1043 	if (err)
1044 		goto done;
1045 
1046 	err = load_tree(want_meta, idset, got_object_commit_get_tree_id(commit),
1047 	    "", got_object_commit_get_committer_time(commit),
1048 	    repo, loose_obj_only, ncolored, nfound, ntrees,
1049 	    progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1050 done:
1051 	got_object_commit_close(commit);
1052 	return err;
1053 }
1054 
1055 static const struct got_error *
1056 load_tag(int want_meta, struct got_object_idset *idset,
1057     struct got_object_id *id, struct got_repository *repo, int loose_obj_only,
1058     int *ncolored, int *nfound, int *ntrees,
1059     got_pack_progress_cb progress_cb, void *progress_arg,
1060     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1061 {
1062 	const struct got_error *err;
1063 	struct got_tag_object *tag = NULL;
1064 
1065 	if (got_object_idset_contains(idset, id))
1066 		return NULL;
1067 
1068 	if (loose_obj_only) {
1069 		int is_packed;
1070 		err = search_packidx(&is_packed, id, repo);
1071 		if (err)
1072 			return err;
1073 		if (is_packed && want_meta)
1074 			return NULL;
1075 	}
1076 
1077 	err = got_object_open_as_tag(&tag, repo, id);
1078 	if (err)
1079 		return err;
1080 
1081 	err = add_object(want_meta, idset, id, "", GOT_OBJ_TYPE_TAG,
1082 	    got_object_tag_get_tagger_time(tag), loose_obj_only, repo,
1083 	    ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1084 	if (err)
1085 		goto done;
1086 
1087 	switch (got_object_tag_get_object_type(tag)) {
1088 	case GOT_OBJ_TYPE_COMMIT:
1089 		err = load_commit(want_meta, idset,
1090 		    got_object_tag_get_object_id(tag), repo, loose_obj_only,
1091 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1092 		    cancel_cb, cancel_arg);
1093 		break;
1094 	case GOT_OBJ_TYPE_TREE:
1095 		err = load_tree(want_meta, idset,
1096 		    got_object_tag_get_object_id(tag), "",
1097 		    got_object_tag_get_tagger_time(tag), repo, loose_obj_only,
1098 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1099 		    cancel_cb, cancel_arg);
1100 		break;
1101 	default:
1102 		break;
1103 	}
1104 
1105 done:
1106 	got_object_tag_close(tag);
1107 	return err;
1108 }
1109 
1110 enum findtwixt_color {
1111 	COLOR_KEEP = 0,
1112 	COLOR_DROP,
1113 	COLOR_BLANK,
1114 	COLOR_SKIP,
1115 };
1116 
1117 static const int findtwixt_colors[] = {
1118 	COLOR_KEEP,
1119 	COLOR_DROP,
1120 	COLOR_BLANK,
1121 	COLOR_SKIP,
1122 };
1123 
1124 static const struct got_error *
1125 paint_commit(struct got_object_qid *qid, int color)
1126 {
1127 	if (color < 0 || color >= nitems(findtwixt_colors))
1128 		return got_error(GOT_ERR_RANGE);
1129 
1130 	qid->data = (void *)&findtwixt_colors[color];
1131 	return NULL;
1132 }
1133 
1134 static const struct got_error *
1135 queue_commit_id(struct got_object_id_queue *ids, struct got_object_id *id,
1136     int color, struct got_repository *repo)
1137 {
1138 	const struct got_error *err;
1139 	struct got_object_qid *qid;
1140 
1141 	err = got_object_qid_alloc(&qid, id);
1142 	if (err)
1143 		return err;
1144 
1145 	STAILQ_INSERT_TAIL(ids, qid, entry);
1146 	return paint_commit(qid, color);
1147 }
1148 
1149 struct append_id_arg {
1150 	struct got_object_id **array;
1151 	int idx;
1152 	struct got_object_idset *drop;
1153 	struct got_object_idset *skip;
1154 };
1155 
1156 static const struct got_error *
1157 append_id(struct got_object_id *id, void *data, void *arg)
1158 {
1159 	struct append_id_arg *a = arg;
1160 
1161 	if (got_object_idset_contains(a->skip, id) ||
1162 	    got_object_idset_contains(a->drop, id))
1163 		return NULL;
1164 
1165 	a->array[++a->idx] = got_object_id_dup(id);
1166 	if (a->array[a->idx] == NULL)
1167 		return got_error_from_errno("got_object_id_dup");
1168 
1169 	return NULL;
1170 }
1171 
1172 static const struct got_error *
1173 queue_commit_or_tag_id(struct got_object_id *id, int color,
1174     struct got_object_id_queue *ids, struct got_repository *repo)
1175 {
1176 	const struct got_error *err;
1177 	struct got_tag_object *tag = NULL;
1178 	int obj_type;
1179 
1180 	err = got_object_get_type(&obj_type, repo, id);
1181 	if (err)
1182 		return err;
1183 
1184 	if (obj_type == GOT_OBJ_TYPE_TAG) {
1185 		err = got_object_open_as_tag(&tag, repo, id);
1186 		if (err)
1187 			return err;
1188 		obj_type = got_object_tag_get_object_type(tag);
1189 		id = got_object_tag_get_object_id(tag);
1190 	}
1191 
1192 	if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1193 		err = queue_commit_id(ids, id, color, repo);
1194 		if (err)
1195 			goto done;
1196 	}
1197 done:
1198 	if (tag)
1199 		got_object_tag_close(tag);
1200 	return err;
1201 }
1202 
1203 static const struct got_error *
1204 paint_commits(int *ncolored, struct got_object_id_queue *ids, int nids,
1205     struct got_object_idset *keep, struct got_object_idset *drop,
1206     struct got_object_idset *skip, struct got_repository *repo,
1207     got_pack_progress_cb progress_cb, void *progress_arg,
1208     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1209 {
1210 	const struct got_error *err = NULL;
1211 	struct got_commit_object *commit = NULL;
1212 	const struct got_object_id_queue *parents;
1213 	struct got_object_qid *qid;
1214 	int nqueued = nids, nskip = 0;
1215 
1216 	while (!STAILQ_EMPTY(ids) && nskip != nqueued) {
1217 		int color;
1218 
1219 		if (cancel_cb) {
1220 			err = cancel_cb(cancel_arg);
1221 			if (err)
1222 				break;
1223 		}
1224 
1225 		qid = STAILQ_FIRST(ids);
1226 		STAILQ_REMOVE_HEAD(ids, entry);
1227 		nqueued--;
1228 		color = *((int *)qid->data);
1229 		if (color == COLOR_SKIP)
1230 			nskip--;
1231 
1232 		if (got_object_idset_contains(skip, &qid->id)) {
1233 			got_object_qid_free(qid);
1234 			continue;
1235 		}
1236 
1237 		switch (color) {
1238 		case COLOR_KEEP:
1239 			if (got_object_idset_contains(keep, &qid->id)) {
1240 				got_object_qid_free(qid);
1241 				continue;
1242 			}
1243 			if (got_object_idset_contains(drop, &qid->id)) {
1244 				err = paint_commit(qid, COLOR_SKIP);
1245 				if (err)
1246 					goto done;
1247 				nskip++;
1248 			} else
1249 				(*ncolored)++;
1250 			err = got_object_idset_add(keep, &qid->id, NULL);
1251 			if (err)
1252 				goto done;
1253 			break;
1254 		case COLOR_DROP:
1255 			if (got_object_idset_contains(drop, &qid->id)) {
1256 				got_object_qid_free(qid);
1257 				continue;
1258 			}
1259 			if (got_object_idset_contains(keep, &qid->id)) {
1260 				err = paint_commit(qid, COLOR_SKIP);
1261 				if (err)
1262 					goto done;
1263 				nskip++;
1264 			} else
1265 				(*ncolored)++;
1266 			err = got_object_idset_add(drop, &qid->id, NULL);
1267 			if (err)
1268 				goto done;
1269 			break;
1270 		case COLOR_SKIP:
1271 			if (!got_object_idset_contains(skip, &qid->id)) {
1272 				err = got_object_idset_add(skip, &qid->id,
1273 				    NULL);
1274 				if (err)
1275 					goto done;
1276 			}
1277 			break;
1278 		default:
1279 			/* should not happen */
1280 			err = got_error_fmt(GOT_ERR_NOT_IMPL,
1281 			    "%s invalid commit color %d", __func__, color);
1282 			goto done;
1283 		}
1284 
1285 		err = report_progress(progress_cb, progress_arg, rl,
1286 		    *ncolored, 0, 0, 0L, 0, 0, 0, 0);
1287 		if (err)
1288 			break;
1289 
1290 
1291 		err = got_object_open_as_commit(&commit, repo, &qid->id);
1292 		if (err)
1293 			break;
1294 
1295 		parents = got_object_commit_get_parent_ids(commit);
1296 		if (parents) {
1297 			struct got_object_qid *pid;
1298 			color = *((int *)qid->data);
1299 			STAILQ_FOREACH(pid, parents, entry) {
1300 				err = queue_commit_id(ids, &pid->id, color,
1301 				    repo);
1302 				if (err)
1303 					break;
1304 				nqueued++;
1305 				if (color == COLOR_SKIP)
1306 					nskip++;
1307 			}
1308 		}
1309 
1310 		got_object_commit_close(commit);
1311 		commit = NULL;
1312 		got_object_qid_free(qid);
1313 	}
1314 done:
1315 	if (commit)
1316 		got_object_commit_close(commit);
1317 	return err;
1318 }
1319 
1320 static const struct got_error *
1321 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1322     struct got_object_id **head, int nhead,
1323     struct got_object_id **tail, int ntail,
1324     struct got_repository *repo,
1325     got_pack_progress_cb progress_cb, void *progress_arg,
1326     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1327 {
1328 	const struct got_error *err = NULL;
1329 	struct got_object_id_queue ids;
1330 	struct got_object_idset *keep, *drop, *skip = NULL;
1331 	int i, nkeep;
1332 
1333 	STAILQ_INIT(&ids);
1334 	*res = NULL;
1335 	*nres = 0;
1336 	*ncolored = 0;
1337 
1338 	keep = got_object_idset_alloc();
1339 	if (keep == NULL)
1340 		return got_error_from_errno("got_object_idset_alloc");
1341 
1342 	drop = got_object_idset_alloc();
1343 	if (drop == NULL) {
1344 		err = got_error_from_errno("got_object_idset_alloc");
1345 		goto done;
1346 	}
1347 
1348 	skip = got_object_idset_alloc();
1349 	if (skip == NULL) {
1350 		err = got_error_from_errno("got_object_idset_alloc");
1351 		goto done;
1352 	}
1353 
1354 	for (i = 0; i < nhead; i++) {
1355 		struct got_object_id *id = head[i];
1356 		if (id == NULL)
1357 			continue;
1358 		err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1359 		if (err)
1360 			goto done;
1361 	}		
1362 
1363 	for (i = 0; i < ntail; i++) {
1364 		struct got_object_id *id = tail[i];
1365 		if (id == NULL)
1366 			continue;
1367 		err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1368 		if (err)
1369 			goto done;
1370 	}
1371 
1372 	err = paint_commits(ncolored, &ids, nhead + ntail,
1373 	    keep, drop, skip, repo, progress_cb, progress_arg, rl,
1374 	    cancel_cb, cancel_arg);
1375 	if (err)
1376 		goto done;
1377 
1378 	nkeep = got_object_idset_num_elements(keep);
1379 	if (nkeep > 0) {
1380 		struct append_id_arg arg;
1381 		arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1382 		if (arg.array == NULL) {
1383 			err = got_error_from_errno("calloc");
1384 			goto done;
1385 		}
1386 		arg.idx = -1;
1387 		arg.skip = skip;
1388 		arg.drop = drop;
1389 		err = got_object_idset_for_each(keep, append_id, &arg);
1390 		if (err) {
1391 			free(arg.array);
1392 			goto done;
1393 		}
1394 		*res = arg.array;
1395 		*nres = arg.idx + 1;
1396 	}
1397 done:
1398 	got_object_idset_free(keep);
1399 	got_object_idset_free(drop);
1400 	if (skip)
1401 		got_object_idset_free(skip);
1402 	got_object_id_queue_free(&ids);
1403 	return err;
1404 }
1405 
1406 static const struct got_error *
1407 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1408     struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1409     struct got_object_id **ours, int nours, struct got_repository *repo,
1410     int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
1411     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1412 {
1413 	const struct got_error *err = NULL;
1414 	struct got_object_id **ids = NULL;
1415 	int i, nobj = 0, obj_type;
1416 
1417 	*ncolored = 0;
1418 	*nfound = 0;
1419 	*ntrees = 0;
1420 
1421 	err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1422 	    repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1423 	if (err || nobj == 0)
1424 		goto done;
1425 
1426 	for (i = 0; i < ntheirs; i++) {
1427 		struct got_object_id *id = theirs[i];
1428 		if (id == NULL)
1429 			continue;
1430 		err = got_object_get_type(&obj_type, repo, id);
1431 		if (err)
1432 			return err;
1433 		if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1434 			err = load_commit(0, idset, id, repo,
1435 			    loose_obj_only, ncolored, nfound, ntrees,
1436 			    progress_cb, progress_arg, rl,
1437 			    cancel_cb, cancel_arg);
1438 			if (err)
1439 				goto done;
1440 		} else if (obj_type == GOT_OBJ_TYPE_TAG) {
1441 			err = load_tag(0, idset, id, repo,
1442 			    loose_obj_only, ncolored, nfound, ntrees,
1443 			    progress_cb, progress_arg, rl,
1444 			    cancel_cb, cancel_arg);
1445 			if (err)
1446 				goto done;
1447 		}
1448 	}
1449 
1450 	for (i = 0; i < nobj; i++) {
1451 		err = load_commit(1, idset, ids[i], repo, loose_obj_only,
1452 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1453 		    cancel_cb, cancel_arg);
1454 		if (err)
1455 			goto done;
1456 	}
1457 
1458 	for (i = 0; i < nours; i++) {
1459 		struct got_object_id *id = ours[i];
1460 		struct got_pack_meta *m;
1461 		if (id == NULL)
1462 			continue;
1463 		m = got_object_idset_get(idset, id);
1464 		if (m == NULL) {
1465 			err = got_object_get_type(&obj_type, repo, id);
1466 			if (err)
1467 				goto done;
1468 		} else
1469 			obj_type = m->obj_type;
1470 		if (obj_type != GOT_OBJ_TYPE_TAG)
1471 			continue;
1472 		err = load_tag(1, idset, id, repo, loose_obj_only,
1473 		    ncolored, nfound, ntrees, progress_cb, progress_arg, rl,
1474 		    cancel_cb, cancel_arg);
1475 		if (err)
1476 			goto done;
1477 	}
1478 done:
1479 	for (i = 0; i < nobj; i++) {
1480 		free(ids[i]);
1481 	}
1482 	free(ids);
1483 	return err;
1484 }
1485 
1486 const struct got_error *
1487 hwrite(FILE *f, void *buf, int len, SHA1_CTX *ctx)
1488 {
1489 	size_t n;
1490 
1491 	SHA1Update(ctx, buf, len);
1492 	n = fwrite(buf, 1, len, f);
1493 	if (n != len)
1494 		return got_ferror(f, GOT_ERR_IO);
1495 	return NULL;
1496 }
1497 
1498 static void
1499 putbe32(char *b, uint32_t n)
1500 {
1501 	b[0] = n >> 24;
1502 	b[1] = n >> 16;
1503 	b[2] = n >> 8;
1504 	b[3] = n >> 0;
1505 }
1506 
1507 static int
1508 write_order_cmp(const void *pa, const void *pb)
1509 {
1510 	struct got_pack_meta *a, *b, *ahd, *bhd;
1511 
1512 	a = *(struct got_pack_meta **)pa;
1513 	b = *(struct got_pack_meta **)pb;
1514 	ahd = (a->head == NULL) ? a : a->head;
1515 	bhd = (b->head == NULL) ? b : b->head;
1516 	if (ahd->mtime != bhd->mtime)
1517 		return bhd->mtime - ahd->mtime;
1518 	if (ahd != bhd)
1519 		return (uintptr_t)bhd - (uintptr_t)ahd;
1520 	if (a->nchain != b->nchain)
1521 		return a->nchain - b->nchain;
1522 	return a->mtime - b->mtime;
1523 }
1524 
1525 static int
1526 reuse_write_order_cmp(const void *pa, const void *pb)
1527 {
1528 	struct got_pack_meta *a, *b;
1529 
1530 	a = *(struct got_pack_meta **)pa;
1531 	b = *(struct got_pack_meta **)pb;
1532 
1533 	if (a->reused_delta_offset < b->reused_delta_offset)
1534 		return -1;
1535 	if (a->reused_delta_offset > b->reused_delta_offset)
1536 		return 1;
1537 	return 0;
1538 }
1539 
1540 static const struct got_error *
1541 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1542 {
1543 	size_t i;
1544 
1545 	*hdrlen = 0;
1546 
1547 	hdr[0] = obj_type << 4;
1548 	hdr[0] |= len & 0xf;
1549 	len >>= 4;
1550 	for (i = 1; len != 0; i++){
1551 		if (i >= bufsize)
1552 			return got_error(GOT_ERR_NO_SPACE);
1553 		hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1554 		hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1555 		len >>= GOT_DELTA_SIZE_SHIFT;
1556 	}
1557 
1558 	*hdrlen = i;
1559 	return NULL;
1560 }
1561 
1562 static int
1563 packoff(char *hdr, off_t off)
1564 {
1565 	int i, j;
1566 	char rbuf[8];
1567 
1568 	rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1569 	for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1570 		rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1571 		    GOT_DELTA_SIZE_MORE;
1572 	}
1573 
1574 	j = 0;
1575 	while (i > 0)
1576 		hdr[j++] = rbuf[--i];
1577 	return j;
1578 }
1579 
1580 static const struct got_error *
1581 deltahdr(off_t *packfile_size, SHA1_CTX *ctx, FILE *packfile,
1582     struct got_pack_meta *m)
1583 {
1584 	const struct got_error *err;
1585 	char buf[32];
1586 	int nh;
1587 
1588 	if (m->prev->off != 0) {
1589 		err = packhdr(&nh, buf, sizeof(buf),
1590 		    GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1591 		if (err)
1592 			return err;
1593 		nh += packoff(buf + nh, m->off - m->prev->off);
1594 		err = hwrite(packfile, buf, nh, ctx);
1595 		if (err)
1596 			return err;
1597 		*packfile_size += nh;
1598 	} else {
1599 		err = packhdr(&nh, buf, sizeof(buf),
1600 		    GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1601 		if (err)
1602 			return err;
1603 		err = hwrite(packfile, buf, nh, ctx);
1604 		if (err)
1605 			return err;
1606 		*packfile_size += nh;
1607 		err = hwrite(packfile, m->prev->id.sha1,
1608 		    sizeof(m->prev->id.sha1), ctx);
1609 		if (err)
1610 			return err;
1611 		*packfile_size += sizeof(m->prev->id.sha1);
1612 	}
1613 
1614 	return NULL;
1615 }
1616 
1617 static const struct got_error *
1618 write_packed_object(off_t *packfile_size, FILE *packfile,
1619     FILE *delta_cache, struct got_pack_meta *m, int *outfd,
1620     SHA1_CTX *ctx, struct got_repository *repo)
1621 {
1622 	const struct got_error *err = NULL;
1623 	struct got_deflate_checksum csum;
1624 	char buf[32];
1625 	int nh;
1626 	struct got_raw_object *raw = NULL;
1627 	off_t outlen;
1628 
1629 	csum.output_sha1 = ctx;
1630 	csum.output_crc = NULL;
1631 
1632 	m->off = ftello(packfile);
1633 	if (m->delta_len == 0) {
1634 		err = got_object_raw_open(&raw, outfd, repo, &m->id);
1635 		if (err)
1636 			goto done;
1637 		err = packhdr(&nh, buf, sizeof(buf),
1638 		    m->obj_type, raw->size);
1639 		if (err)
1640 			goto done;
1641 		err = hwrite(packfile, buf, nh, ctx);
1642 		if (err)
1643 			goto done;
1644 		*packfile_size += nh;
1645 		if (raw->f == NULL) {
1646 			err = got_deflate_to_file_mmap(&outlen,
1647 			    raw->data + raw->hdrlen, 0, raw->size,
1648 			    packfile, &csum);
1649 			if (err)
1650 				goto done;
1651 		} else {
1652 			if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1653 			    == -1) {
1654 				err = got_error_from_errno("fseeko");
1655 				goto done;
1656 			}
1657 			err = got_deflate_to_file(&outlen, raw->f,
1658 			    raw->size, packfile, &csum);
1659 			if (err)
1660 				goto done;
1661 		}
1662 		*packfile_size += outlen;
1663 		got_object_raw_close(raw);
1664 		raw = NULL;
1665 	} else if (m->delta_buf) {
1666 		err = deltahdr(packfile_size, ctx, packfile, m);
1667 		if (err)
1668 			goto done;
1669 		err = got_deflate_to_file_mmap(&outlen,
1670 		    m->delta_buf, 0, m->delta_len, packfile, &csum);
1671 		if (err)
1672 			goto done;
1673 		*packfile_size += outlen;
1674 		free(m->delta_buf);
1675 		m->delta_buf = NULL;
1676 	} else {
1677 		if (fseeko(delta_cache, m->delta_offset, SEEK_SET)
1678 		    == -1) {
1679 			err = got_error_from_errno("fseeko");
1680 			goto done;
1681 		}
1682 		err = deltahdr(packfile_size, ctx, packfile, m);
1683 		if (err)
1684 			goto done;
1685 		err = got_deflate_to_file(&outlen, delta_cache,
1686 		    m->delta_len, packfile, &csum);
1687 		if (err)
1688 			goto done;
1689 		*packfile_size += outlen;
1690 	}
1691 done:
1692 	if (raw)
1693 		got_object_raw_close(raw);
1694 	return err;
1695 }
1696 
1697 static const struct got_error *
1698 genpack(uint8_t *pack_sha1, FILE *packfile, FILE *delta_cache,
1699     struct got_pack_meta **deltify, int ndeltify,
1700     struct got_pack_meta **reuse, int nreuse,
1701     int ncolored, int nfound, int ntrees, int nours,
1702     struct got_repository *repo,
1703     got_pack_progress_cb progress_cb, void *progress_arg,
1704     struct got_ratelimit *rl,
1705     got_cancel_cb cancel_cb, void *cancel_arg)
1706 {
1707 	const struct got_error *err = NULL;
1708 	int i;
1709 	SHA1_CTX ctx;
1710 	struct got_pack_meta *m;
1711 	char buf[32];
1712 	size_t n;
1713 	off_t packfile_size = 0;
1714 	int outfd = -1;
1715 
1716 	SHA1Init(&ctx);
1717 
1718 	err = hwrite(packfile, "PACK", 4, &ctx);
1719 	if (err)
1720 		return err;
1721 	putbe32(buf, GOT_PACKFILE_VERSION);
1722 	err = hwrite(packfile, buf, 4, &ctx);
1723 	if (err)
1724 		goto done;
1725 	putbe32(buf, ndeltify + nreuse);
1726 	err = hwrite(packfile, buf, 4, &ctx);
1727 	if (err)
1728 		goto done;
1729 
1730 	qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1731 	    write_order_cmp);
1732 	for (i = 0; i < ndeltify; i++) {
1733 		err = report_progress(progress_cb, progress_arg, rl,
1734 		    ncolored, nfound, ntrees, packfile_size, nours,
1735 		    ndeltify + nreuse, ndeltify + nreuse, i);
1736 		if (err)
1737 			goto done;
1738 		m = deltify[i];
1739 		err = write_packed_object(&packfile_size, packfile,
1740 		    delta_cache, m, &outfd, &ctx, repo);
1741 		if (err)
1742 			goto done;
1743 	}
1744 
1745 	qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1746 	    reuse_write_order_cmp);
1747 	for (i = 0; i < nreuse; i++) {
1748 		err = report_progress(progress_cb, progress_arg, rl,
1749 		    ncolored, nfound, ntrees, packfile_size, nours,
1750 		    ndeltify + nreuse, ndeltify + nreuse, ndeltify + i);
1751 		if (err)
1752 			goto done;
1753 		m = reuse[i];
1754 		err = write_packed_object(&packfile_size, packfile,
1755 		    delta_cache, m, &outfd, &ctx, repo);
1756 		if (err)
1757 			goto done;
1758 	}
1759 
1760 	SHA1Final(pack_sha1, &ctx);
1761 	n = fwrite(pack_sha1, 1, SHA1_DIGEST_LENGTH, packfile);
1762 	if (n != SHA1_DIGEST_LENGTH)
1763 		err = got_ferror(packfile, GOT_ERR_IO);
1764 	packfile_size += SHA1_DIGEST_LENGTH;
1765 	packfile_size += sizeof(struct got_packfile_hdr);
1766 	if (progress_cb) {
1767 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1768 		    packfile_size, nours, ndeltify + nreuse,
1769 		    ndeltify + nreuse, ndeltify + nreuse);
1770 		if (err)
1771 			goto done;
1772 	}
1773 done:
1774 	if (outfd != -1 && close(outfd) == -1 && err == NULL)
1775 		err = got_error_from_errno("close");
1776 	return err;
1777 }
1778 
1779 static const struct got_error *
1780 remove_unused_object(struct got_object_id *id, void *data, void *arg)
1781 {
1782 	struct got_object_idset *idset = arg;
1783 
1784 	if (data == NULL)
1785 		got_object_idset_remove(NULL, idset, id);
1786 
1787 	return NULL;
1788 }
1789 
1790 static const struct got_error *
1791 remove_reused_object(struct got_object_id *id, void *data, void *arg)
1792 {
1793 	struct got_object_idset *idset = arg;
1794 	struct got_pack_meta *m = data;
1795 
1796 	if (m->have_reused_delta)
1797 		got_object_idset_remove(NULL, idset, id);
1798 
1799 	return NULL;
1800 }
1801 
1802 static const struct got_error *
1803 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1804 {
1805 	struct got_pack_meta *m = data;
1806 	struct got_pack_metavec *v = arg;
1807 
1808 	return add_meta(m, v);
1809 }
1810 
1811 const struct got_error *
1812 got_pack_create(uint8_t *packsha1, FILE *packfile,
1813     struct got_object_id **theirs, int ntheirs,
1814     struct got_object_id **ours, int nours,
1815     struct got_repository *repo, int loose_obj_only, int allow_empty,
1816     got_pack_progress_cb progress_cb, void *progress_arg,
1817     got_cancel_cb cancel_cb, void *cancel_arg)
1818 {
1819 	const struct got_error *err;
1820 	int delta_cache_fd = -1;
1821 	FILE *delta_cache = NULL;
1822 	struct got_object_idset *idset;
1823 	struct got_ratelimit rl;
1824 	struct got_pack_metavec deltify, reuse;
1825 	int ncolored = 0, nfound = 0, ntrees = 0;
1826 
1827 	memset(&deltify, 0, sizeof(deltify));
1828 	memset(&reuse, 0, sizeof(reuse));
1829 
1830 	got_ratelimit_init(&rl, 0, 500);
1831 
1832 	idset = got_object_idset_alloc();
1833 	if (idset == NULL)
1834 		return got_error_from_errno("got_object_idset_alloc");
1835 
1836 	err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1837 	    ntheirs, ours, nours, repo, loose_obj_only,
1838 	    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1839 	if (err)
1840 		return err;
1841 
1842 	err = got_object_idset_for_each(idset, remove_unused_object, idset);
1843 	if (err)
1844 		goto done;
1845 
1846 	if (progress_cb) {
1847 		err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1848 		    0L, nours, got_object_idset_num_elements(idset), 0, 0);
1849 		if (err)
1850 			goto done;
1851 	}
1852 
1853 	if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1854 		err = got_error(GOT_ERR_CANNOT_PACK);
1855 		goto done;
1856 	}
1857 
1858 	delta_cache_fd = got_opentempfd();
1859 	if (delta_cache_fd == -1) {
1860 		err = got_error_from_errno("got_opentemp");
1861 		goto done;
1862 	}
1863 
1864 	reuse.metasz = 64;
1865 	reuse.meta = calloc(reuse.metasz,
1866 	    sizeof(struct got_pack_meta *));
1867 	if (reuse.meta == NULL) {
1868 		err = got_error_from_errno("calloc");
1869 		goto done;
1870 	}
1871 
1872 	err = search_deltas(&reuse, idset, delta_cache_fd, ncolored, nfound,
1873 	    ntrees, nours, repo, progress_cb, progress_arg, &rl,
1874 	    cancel_cb, cancel_arg);
1875 	if (err)
1876 		goto done;
1877 	if (reuse.nmeta > 0) {
1878 		err = got_object_idset_for_each(idset,
1879 		    remove_reused_object, idset);
1880 		if (err)
1881 			goto done;
1882 	}
1883 
1884 	delta_cache = fdopen(delta_cache_fd, "a+");
1885 	if (delta_cache == NULL) {
1886 		err = got_error_from_errno("fdopen");
1887 		goto done;
1888 	}
1889 	delta_cache_fd = -1;
1890 
1891 	if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1892 		err = got_error_from_errno("fseeko");
1893 		goto done;
1894 	}
1895 
1896 	deltify.meta = calloc(got_object_idset_num_elements(idset),
1897 	    sizeof(struct got_pack_meta *));
1898 	if (deltify.meta == NULL) {
1899 		err = got_error_from_errno("calloc");
1900 		goto done;
1901 	}
1902 	deltify.metasz = got_object_idset_num_elements(idset);
1903 
1904 	err = got_object_idset_for_each(idset, add_meta_idset_cb, &deltify);
1905 	if (err)
1906 		goto done;
1907 	if (deltify.nmeta > 0) {
1908 		err = pick_deltas(deltify.meta, deltify.nmeta, ncolored,
1909 		    nfound, ntrees, nours, reuse.nmeta, delta_cache, repo,
1910 		    progress_cb, progress_arg, &rl, cancel_cb, cancel_arg);
1911 		if (err)
1912 			goto done;
1913 		if (fseeko(delta_cache, 0L, SEEK_SET) == -1) {
1914 			err = got_error_from_errno("fseeko");
1915 			goto done;
1916 		}
1917 	}
1918 
1919 	err = genpack(packsha1, packfile, delta_cache, deltify.meta,
1920 	    deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1921 	    nours, repo, progress_cb, progress_arg, &rl,
1922 	    cancel_cb, cancel_arg);
1923 	if (err)
1924 		goto done;
1925 done:
1926 	free_nmeta(deltify.meta, deltify.nmeta);
1927 	free_nmeta(reuse.meta, reuse.nmeta);
1928 	got_object_idset_free(idset);
1929 	if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1930 		err = got_error_from_errno("close");
1931 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
1932 		err = got_error_from_errno("fclose");
1933 	return err;
1934 }