style(9) only; no functional changes
[got-portable.git] / lib / pack_create.c
blobf69fe8b0f0d9292efbd835da4ba3669bdceff718
1 /*
2 * Copyright (c) 2020 Ori Bernstein
3 * Copyright (c) 2021 Stefan Sperling <stsp@openbsd.org>
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.
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.
18 #include "got_compat.h"
20 #include <sys/types.h>
21 #include <sys/queue.h>
22 #include <sys/uio.h>
23 #include <sys/stat.h>
24 #include <sys/time.h>
25 #include <sys/mman.h>
27 #include <errno.h>
28 #include <stdint.h>
29 #include <inttypes.h>
30 #include <poll.h>
31 #include <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <time.h>
35 #include <unistd.h>
36 #include <limits.h>
37 #include <zlib.h>
39 #include "got_error.h"
40 #include "got_cancel.h"
41 #include "got_object.h"
42 #include "got_path.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_repository_admin.h"
47 #include "got_lib_deltify.h"
48 #include "got_lib_delta.h"
49 #include "got_lib_hash.h"
50 #include "got_lib_object.h"
51 #include "got_lib_object_idset.h"
52 #include "got_lib_object_cache.h"
53 #include "got_lib_deflate.h"
54 #include "got_lib_ratelimit.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_pack_create.h"
57 #include "got_lib_repository.h"
58 #include "got_lib_inflate.h"
59 #include "got_lib_poll.h"
61 #include "murmurhash2.h"
63 #ifndef MIN
64 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
65 #endif
67 #ifndef MAX
68 #define MAX(_a,_b) ((_a) > (_b) ? (_a) : (_b))
69 #endif
71 #ifndef nitems
72 #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
73 #endif
75 static const struct got_error *
76 alloc_meta(struct got_pack_meta **new, struct got_object_id *id,
77 const char *path, int obj_type, time_t mtime, uint32_t seed)
79 struct got_pack_meta *m;
81 *new = NULL;
83 m = calloc(1, sizeof(*m));
84 if (m == NULL)
85 return got_error_from_errno("calloc");
87 memcpy(&m->id, id, sizeof(m->id));
89 m->path_hash = murmurhash2(path, strlen(path), seed);
90 m->obj_type = obj_type;
91 m->mtime = mtime;
92 *new = m;
93 return NULL;
96 static void
97 clear_meta(struct got_pack_meta *meta)
99 if (meta == NULL)
100 return;
101 meta->path_hash = 0;
102 free(meta->delta_buf);
103 meta->delta_buf = NULL;
104 free(meta->base_obj_id);
105 meta->base_obj_id = NULL;
106 meta->reused_delta_offset = 0;
107 got_deltify_free(meta->dtab);
108 meta->dtab = NULL;
111 static void
112 free_nmeta(struct got_pack_meta **meta, int nmeta)
114 int i;
116 for (i = 0; i < nmeta; i++)
117 clear_meta(meta[i]);
119 free(meta);
122 static int
123 delta_order_cmp(const void *pa, const void *pb)
125 struct got_pack_meta *a, *b;
127 a = *(struct got_pack_meta **)pa;
128 b = *(struct got_pack_meta **)pb;
130 if (a->obj_type != b->obj_type)
131 return a->obj_type - b->obj_type;
132 if (a->path_hash < b->path_hash)
133 return -1;
134 if (a->path_hash > b->path_hash)
135 return 1;
136 if (a->mtime < b->mtime)
137 return -1;
138 if (a->mtime > b->mtime)
139 return 1;
140 return got_object_id_cmp(&a->id, &b->id);
143 static off_t
144 delta_size(struct got_delta_instruction *deltas, int ndeltas)
146 int i;
147 off_t size = 32;
148 for (i = 0; i < ndeltas; i++) {
149 if (deltas[i].copy)
150 size += GOT_DELTA_SIZE_SHIFT;
151 else
152 size += deltas[i].len + 1;
154 return size;
157 static const struct got_error *
158 append(unsigned char **p, size_t *len, off_t *sz, void *seg, int nseg)
160 char *n;
162 if (*len + nseg >= *sz) {
163 while (*len + nseg >= *sz)
164 *sz += *sz / 2;
165 n = realloc(*p, *sz);
166 if (n == NULL)
167 return got_error_from_errno("realloc");
168 *p = n;
170 memcpy(*p + *len, seg, nseg);
171 *len += nseg;
172 return NULL;
175 static const struct got_error *
176 encode_delta_in_mem(struct got_pack_meta *m, struct got_raw_object *o,
177 struct got_delta_instruction *deltas, int ndeltas,
178 off_t delta_size, off_t base_size)
180 const struct got_error *err;
181 unsigned char buf[16], *bp;
182 int i, j;
183 size_t len = 0, compressed_len;
184 off_t bufsize = delta_size;
185 off_t n;
186 struct got_delta_instruction *d;
187 uint8_t *delta_buf;
189 delta_buf = malloc(bufsize);
190 if (delta_buf == NULL)
191 return got_error_from_errno("malloc");
193 /* base object size */
194 buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
195 n = base_size >> GOT_DELTA_SIZE_SHIFT;
196 for (i = 1; n > 0; i++) {
197 buf[i - 1] |= GOT_DELTA_SIZE_MORE;
198 buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
199 n >>= GOT_DELTA_SIZE_SHIFT;
201 err = append(&delta_buf, &len, &bufsize, buf, i);
202 if (err)
203 goto done;
205 /* target object size */
206 buf[0] = o->size & GOT_DELTA_SIZE_VAL_MASK;
207 n = o->size >> GOT_DELTA_SIZE_SHIFT;
208 for (i = 1; n > 0; i++) {
209 buf[i - 1] |= GOT_DELTA_SIZE_MORE;
210 buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
211 n >>= GOT_DELTA_SIZE_SHIFT;
213 err = append(&delta_buf, &len, &bufsize, buf, i);
214 if (err)
215 goto done;
217 for (j = 0; j < ndeltas; j++) {
218 d = &deltas[j];
219 if (d->copy) {
220 n = d->offset;
221 bp = &buf[1];
222 buf[0] = GOT_DELTA_BASE_COPY;
223 for (i = 0; i < 4; i++) {
224 /* DELTA_COPY_OFF1 ... DELTA_COPY_OFF4 */
225 buf[0] |= 1 << i;
226 *bp++ = n & 0xff;
227 n >>= 8;
228 if (n == 0)
229 break;
232 n = d->len;
233 if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
234 /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
235 for (i = 0; i < 3 && n > 0; i++) {
236 buf[0] |= 1 << (i + 4);
237 *bp++ = n & 0xff;
238 n >>= 8;
241 err = append(&delta_buf, &len, &bufsize,
242 buf, bp - buf);
243 if (err)
244 goto done;
245 } else if (o->f == NULL) {
246 n = 0;
247 while (n != d->len) {
248 buf[0] = (d->len - n < 127) ? d->len - n : 127;
249 err = append(&delta_buf, &len, &bufsize,
250 buf, 1);
251 if (err)
252 goto done;
253 err = append(&delta_buf, &len, &bufsize,
254 o->data + o->hdrlen + d->offset + n,
255 buf[0]);
256 if (err)
257 goto done;
258 n += buf[0];
260 } else {
261 char content[128];
262 size_t r;
263 if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
264 err = got_error_from_errno("fseeko");
265 goto done;
267 n = 0;
268 while (n != d->len) {
269 buf[0] = (d->len - n < 127) ? d->len - n : 127;
270 err = append(&delta_buf, &len, &bufsize,
271 buf, 1);
272 if (err)
273 goto done;
274 r = fread(content, 1, buf[0], o->f);
275 if (r != buf[0]) {
276 err = got_ferror(o->f, GOT_ERR_IO);
277 goto done;
279 err = append(&delta_buf, &len, &bufsize,
280 content, buf[0]);
281 if (err)
282 goto done;
283 n += buf[0];
288 err = got_deflate_to_mem_mmap(&m->delta_buf, &compressed_len,
289 NULL, NULL, delta_buf, 0, len);
290 if (err)
291 goto done;
293 m->delta_len = len;
294 m->delta_compressed_len = compressed_len;
295 done:
296 free(delta_buf);
297 return err;
300 static const struct got_error *
301 encode_delta(struct got_pack_meta *m, struct got_raw_object *o,
302 struct got_delta_instruction *deltas, int ndeltas,
303 off_t base_size, FILE *f)
305 const struct got_error *err;
306 unsigned char buf[16], *bp;
307 int i, j;
308 off_t n;
309 struct got_deflate_buf zb;
310 struct got_delta_instruction *d;
311 off_t delta_len = 0, compressed_len = 0;
313 err = got_deflate_init(&zb, NULL, GOT_DEFLATE_BUFSIZE);
314 if (err)
315 return err;
317 /* base object size */
318 buf[0] = base_size & GOT_DELTA_SIZE_VAL_MASK;
319 n = base_size >> GOT_DELTA_SIZE_SHIFT;
320 for (i = 1; n > 0; i++) {
321 buf[i - 1] |= GOT_DELTA_SIZE_MORE;
322 buf[i] = n & GOT_DELTA_SIZE_VAL_MASK;
323 n >>= GOT_DELTA_SIZE_SHIFT;
326 err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
327 buf, 0, i, f, NULL);
328 if (err)
329 goto done;
330 delta_len += i;
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;
341 err = got_deflate_append_to_file_mmap(&zb, &compressed_len,
342 buf, 0, i, f, NULL);
343 if (err)
344 goto done;
345 delta_len += i;
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;
361 n = d->len;
362 if (n != GOT_DELTA_COPY_DEFAULT_LEN) {
363 /* DELTA_COPY_LEN1 ... DELTA_COPY_LEN3 */
364 for (i = 0; i < 3 && n > 0; i++) {
365 buf[0] |= 1 << (i + 4);
366 *bp++ = n & 0xff;
367 n >>= 8;
370 err = got_deflate_append_to_file_mmap(&zb,
371 &compressed_len, buf, 0, bp - buf, f, NULL);
372 if (err)
373 goto done;
374 delta_len += (bp - buf);
375 } else if (o->f == NULL) {
376 n = 0;
377 while (n != d->len) {
378 buf[0] = (d->len - n < 127) ? d->len - n : 127;
379 err = got_deflate_append_to_file_mmap(&zb,
380 &compressed_len, buf, 0, 1, f, NULL);
381 if (err)
382 goto done;
383 delta_len++;
384 err = got_deflate_append_to_file_mmap(&zb,
385 &compressed_len,
386 o->data + o->hdrlen + d->offset + n, 0,
387 buf[0], f, NULL);
388 if (err)
389 goto done;
390 delta_len += buf[0];
391 n += buf[0];
393 } else {
394 char content[128];
395 size_t r;
396 if (fseeko(o->f, o->hdrlen + d->offset, SEEK_SET) == -1) {
397 err = got_error_from_errno("fseeko");
398 goto done;
400 n = 0;
401 while (n != d->len) {
402 buf[0] = (d->len - n < 127) ? d->len - n : 127;
403 err = got_deflate_append_to_file_mmap(&zb,
404 &compressed_len, buf, 0, 1, f, NULL);
405 if (err)
406 goto done;
407 delta_len++;
408 r = fread(content, 1, buf[0], o->f);
409 if (r != buf[0]) {
410 err = got_ferror(o->f, GOT_ERR_IO);
411 goto done;
413 err = got_deflate_append_to_file_mmap(&zb,
414 &compressed_len, content, 0, buf[0], f,
415 NULL);
416 if (err)
417 goto done;
418 delta_len += buf[0];
419 n += buf[0];
424 err = got_deflate_flush(&zb, f, NULL, &compressed_len);
425 if (err)
426 goto done;
428 /* sanity check */
429 if (compressed_len != ftello(f) - m->delta_offset) {
430 err = got_error(GOT_ERR_COMPRESSION);
431 goto done;
434 m->delta_len = delta_len;
435 m->delta_compressed_len = compressed_len;
436 done:
437 got_deflate_end(&zb);
438 return err;
441 const struct got_error *
442 got_pack_report_progress(got_pack_progress_cb progress_cb, void *progress_arg,
443 struct got_ratelimit *rl, int ncolored, int nfound, int ntrees,
444 off_t packfile_size, int ncommits, int nobj_total, int obj_deltify,
445 int nobj_written, int pack_done)
447 const struct got_error *err;
448 int elapsed;
450 if (progress_cb == NULL)
451 return NULL;
453 err = got_ratelimit_check(&elapsed, rl);
454 if (err || !elapsed)
455 return err;
457 return progress_cb(progress_arg, ncolored, nfound, ntrees,
458 packfile_size, ncommits, nobj_total, obj_deltify, nobj_written,
459 pack_done);
462 const struct got_error *
463 got_pack_add_meta(struct got_pack_meta *m, struct got_pack_metavec *v)
465 if (v->nmeta == v->metasz){
466 size_t newsize = 2 * v->metasz;
467 struct got_pack_meta **new;
468 new = reallocarray(v->meta, newsize, sizeof(*new));
469 if (new == NULL)
470 return got_error_from_errno("reallocarray");
471 v->meta = new;
472 v->metasz = newsize;
475 v->meta[v->nmeta++] = m;
476 return NULL;
479 const struct got_error *
480 got_pack_find_pack_for_reuse(struct got_packidx **best_packidx,
481 struct got_repository *repo)
483 const struct got_error *err = NULL;
484 struct got_pathlist_entry *pe;
485 const char *best_packidx_path = NULL;
486 int nobj_max = 0;
488 *best_packidx = NULL;
490 RB_FOREACH(pe, got_pathlist_head, &repo->packidx_paths) {
491 const char *path_packidx = pe->path;
492 struct got_packidx *packidx;
493 int nobj;
495 err = got_repo_get_packidx(&packidx, path_packidx, repo);
496 if (err)
497 break;
499 nobj = be32toh(packidx->hdr.fanout_table[0xff]);
500 if (nobj > nobj_max) {
501 best_packidx_path = path_packidx;
502 nobj_max = nobj;
506 if (best_packidx_path) {
507 err = got_repo_get_packidx(best_packidx, best_packidx_path,
508 repo);
511 return err;
514 const struct got_error *
515 got_pack_cache_pack_for_packidx(struct got_pack **pack,
516 struct got_packidx *packidx, struct got_repository *repo)
518 const struct got_error *err;
519 char *path_packfile = NULL;
521 err = got_packidx_get_packfile_path(&path_packfile,
522 packidx->path_packidx);
523 if (err)
524 return err;
526 *pack = got_repo_get_cached_pack(repo, path_packfile);
527 if (*pack == NULL) {
528 err = got_repo_cache_pack(pack, repo, path_packfile, packidx);
529 if (err)
530 goto done;
532 done:
533 free(path_packfile);
534 return err;
537 static const struct got_error *
538 pick_deltas(struct got_pack_meta **meta, int nmeta, int ncolored,
539 int nfound, int ntrees, int ncommits, int nreused, FILE *delta_cache,
540 struct got_repository *repo,
541 got_pack_progress_cb progress_cb, void *progress_arg,
542 struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
544 const struct got_error *err = NULL;
545 struct got_pack_meta *m = NULL, *base = NULL;
546 struct got_raw_object *raw = NULL, *base_raw = NULL;
547 struct got_delta_instruction *deltas = NULL, *best_deltas = NULL;
548 int i, j, ndeltas, best_ndeltas;
549 off_t size, best_size;
550 const int max_base_candidates = 3;
551 size_t delta_memsize = 0;
552 const size_t max_delta_memsize = 4 * GOT_DELTA_RESULT_SIZE_CACHED_MAX;
553 int outfd = -1;
554 uint32_t delta_seed;
556 delta_seed = arc4random();
558 qsort(meta, nmeta, sizeof(struct got_pack_meta *), delta_order_cmp);
559 for (i = 0; i < nmeta; i++) {
560 if (cancel_cb) {
561 err = (*cancel_cb)(cancel_arg);
562 if (err)
563 break;
565 err = got_pack_report_progress(progress_cb, progress_arg, rl,
566 ncolored, nfound, ntrees, 0L, ncommits, nreused + nmeta,
567 nreused + i, 0, 0);
568 if (err)
569 goto done;
570 m = meta[i];
572 if (m->obj_type == GOT_OBJ_TYPE_COMMIT ||
573 m->obj_type == GOT_OBJ_TYPE_TAG)
574 continue;
576 err = got_object_raw_open(&raw, &outfd, repo, &m->id);
577 if (err)
578 goto done;
579 m->size = raw->size;
581 if (raw->f == NULL) {
582 err = got_deltify_init_mem(&m->dtab, raw->data,
583 raw->hdrlen, raw->size + raw->hdrlen, delta_seed);
584 } else {
585 err = got_deltify_init(&m->dtab, raw->f, raw->hdrlen,
586 raw->size + raw->hdrlen, delta_seed);
588 if (err)
589 goto done;
591 if (i > max_base_candidates) {
592 struct got_pack_meta *n = NULL;
593 n = meta[i - (max_base_candidates + 1)];
594 got_deltify_free(n->dtab);
595 n->dtab = NULL;
598 best_size = raw->size;
599 best_ndeltas = 0;
600 for (j = MAX(0, i - max_base_candidates); j < i; j++) {
601 if (cancel_cb) {
602 err = (*cancel_cb)(cancel_arg);
603 if (err)
604 goto done;
606 base = meta[j];
607 /* long chains make unpacking slow, avoid such bases */
608 if (base->nchain >= 128 ||
609 base->obj_type != m->obj_type)
610 continue;
612 err = got_object_raw_open(&base_raw, &outfd, repo,
613 &base->id);
614 if (err)
615 goto done;
617 if (raw->f == NULL && base_raw->f == NULL) {
618 err = got_deltify_mem_mem(&deltas, &ndeltas,
619 raw->data, raw->hdrlen,
620 raw->size + raw->hdrlen, delta_seed,
621 base->dtab, base_raw->data,
622 base_raw->hdrlen,
623 base_raw->size + base_raw->hdrlen);
624 } else if (raw->f == NULL) {
625 err = got_deltify_mem_file(&deltas, &ndeltas,
626 raw->data, raw->hdrlen,
627 raw->size + raw->hdrlen, delta_seed,
628 base->dtab, base_raw->f,
629 base_raw->hdrlen,
630 base_raw->size + base_raw->hdrlen);
631 } else if (base_raw->f == NULL) {
632 err = got_deltify_file_mem(&deltas, &ndeltas,
633 raw->f, raw->hdrlen,
634 raw->size + raw->hdrlen, delta_seed,
635 base->dtab, base_raw->data,
636 base_raw->hdrlen,
637 base_raw->size + base_raw->hdrlen);
638 } else {
639 err = got_deltify(&deltas, &ndeltas,
640 raw->f, raw->hdrlen,
641 raw->size + raw->hdrlen, delta_seed,
642 base->dtab, base_raw->f, base_raw->hdrlen,
643 base_raw->size + base_raw->hdrlen);
645 got_object_raw_close(base_raw);
646 base_raw = NULL;
647 if (err)
648 goto done;
650 size = delta_size(deltas, ndeltas);
651 if (size + 32 < best_size){
653 * if we already picked a best delta,
654 * replace it.
656 best_size = size;
657 free(best_deltas);
658 best_deltas = deltas;
659 best_ndeltas = ndeltas;
660 deltas = NULL;
661 m->nchain = base->nchain + 1;
662 m->prev = base;
663 m->head = base->head;
664 if (m->head == NULL)
665 m->head = base;
666 } else {
667 free(deltas);
668 deltas = NULL;
669 ndeltas = 0;
673 if (best_ndeltas > 0) {
674 if (best_size <= GOT_DELTA_RESULT_SIZE_CACHED_MAX &&
675 delta_memsize + best_size <= max_delta_memsize) {
676 delta_memsize += best_size;
677 err = encode_delta_in_mem(m, raw, best_deltas,
678 best_ndeltas, best_size, m->prev->size);
679 } else {
680 m->delta_offset = ftello(delta_cache);
681 err = encode_delta(m, raw, best_deltas,
682 best_ndeltas, m->prev->size, delta_cache);
684 free(best_deltas);
685 best_deltas = NULL;
686 best_ndeltas = 0;
687 if (err)
688 goto done;
691 got_object_raw_close(raw);
692 raw = NULL;
694 done:
695 if (raw)
696 got_object_raw_close(raw);
697 if (base_raw)
698 got_object_raw_close(base_raw);
699 if (outfd != -1 && close(outfd) == -1 && err == NULL)
700 err = got_error_from_errno("close");
701 free(deltas);
702 free(best_deltas);
703 return err;
706 static const struct got_error *
707 search_packidx(int *found, struct got_object_id *id,
708 struct got_repository *repo)
710 const struct got_error *err = NULL;
711 struct got_packidx *packidx = NULL;
712 int idx;
714 *found = 0;
716 err = got_repo_search_packidx(&packidx, &idx, repo, id);
717 if (err == NULL)
718 *found = 1; /* object is already packed */
719 else if (err->code == GOT_ERR_NO_OBJ)
720 err = NULL;
721 return err;
724 const struct got_error *
725 got_pack_add_object(int want_meta, struct got_object_idset *idset,
726 struct got_object_id *id, const char *path, int obj_type,
727 time_t mtime, uint32_t seed, int loose_obj_only,
728 struct got_repository *repo, int *ncolored, int *nfound, int *ntrees,
729 got_pack_progress_cb progress_cb, void *progress_arg,
730 struct got_ratelimit *rl)
732 const struct got_error *err;
733 struct got_pack_meta *m = NULL;
735 if (loose_obj_only) {
736 int is_packed;
737 err = search_packidx(&is_packed, id, repo);
738 if (err)
739 return err;
740 if (is_packed && want_meta)
741 return NULL;
744 if (want_meta) {
745 err = alloc_meta(&m, id, path, obj_type, mtime, seed);
746 if (err)
747 return err;
749 (*nfound)++;
750 err = got_pack_report_progress(progress_cb, progress_arg, rl,
751 *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0, 0);
752 if (err) {
753 clear_meta(m);
754 free(m);
755 return err;
759 err = got_object_idset_add(idset, id, m);
760 if (err) {
761 clear_meta(m);
762 free(m);
764 return err;
767 const struct got_error *
768 got_pack_load_tree_entries(struct got_object_id_queue *ids, int want_meta,
769 struct got_object_idset *idset, struct got_object_idset *idset_exclude,
770 struct got_tree_object *tree,
771 const char *dpath, time_t mtime, uint32_t seed, struct got_repository *repo,
772 int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
773 got_pack_progress_cb progress_cb, void *progress_arg,
774 struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
776 const struct got_error *err;
777 char *p = NULL;
778 int i;
780 (*ntrees)++;
781 err = got_pack_report_progress(progress_cb, progress_arg, rl,
782 *ncolored, *nfound, *ntrees, 0L, 0, 0, 0, 0, 0);
783 if (err)
784 return err;
786 for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
787 struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
788 struct got_object_id *id = got_tree_entry_get_id(e);
789 mode_t mode = got_tree_entry_get_mode(e);
791 if (cancel_cb) {
792 err = (*cancel_cb)(cancel_arg);
793 if (err)
794 break;
797 if (got_object_tree_entry_is_submodule(e) ||
798 got_object_idset_contains(idset, id) ||
799 got_object_idset_contains(idset_exclude, id))
800 continue;
803 * If got-read-pack is crawling trees for us then
804 * we are only here to collect blob IDs.
806 if (ids == NULL && S_ISDIR(mode))
807 continue;
809 if (asprintf(&p, "%s%s%s", dpath,
810 got_path_is_root_dir(dpath) ? "" : "/",
811 got_tree_entry_get_name(e)) == -1) {
812 err = got_error_from_errno("asprintf");
813 break;
816 if (S_ISDIR(mode)) {
817 struct got_object_qid *qid;
818 err = got_object_qid_alloc(&qid, id);
819 if (err)
820 break;
821 qid->data = p;
822 p = NULL;
823 STAILQ_INSERT_TAIL(ids, qid, entry);
824 } else if (S_ISREG(mode) || S_ISLNK(mode)) {
825 err = got_pack_add_object(want_meta,
826 want_meta ? idset : idset_exclude, id, p,
827 GOT_OBJ_TYPE_BLOB, mtime, seed, loose_obj_only,
828 repo, ncolored, nfound, ntrees,
829 progress_cb, progress_arg, rl);
830 if (err)
831 break;
832 free(p);
833 p = NULL;
834 } else {
835 free(p);
836 p = NULL;
840 free(p);
841 return err;
844 const struct got_error *
845 got_pack_load_tree(int want_meta, struct got_object_idset *idset,
846 struct got_object_idset *idset_exclude,
847 struct got_object_id *tree_id, const char *dpath, time_t mtime,
848 uint32_t seed, struct got_repository *repo, int loose_obj_only,
849 int *ncolored, int *nfound, int *ntrees,
850 got_pack_progress_cb progress_cb, void *progress_arg,
851 struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
853 const struct got_error *err = NULL;
854 struct got_object_id_queue tree_ids;
855 struct got_object_qid *qid;
856 struct got_tree_object *tree = NULL;
858 if (got_object_idset_contains(idset, tree_id) ||
859 got_object_idset_contains(idset_exclude, tree_id))
860 return NULL;
862 err = got_object_qid_alloc(&qid, tree_id);
863 if (err)
864 return err;
865 qid->data = strdup(dpath);
866 if (qid->data == NULL) {
867 err = got_error_from_errno("strdup");
868 got_object_qid_free(qid);
869 return err;
872 STAILQ_INIT(&tree_ids);
873 STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
875 while (!STAILQ_EMPTY(&tree_ids)) {
876 const char *path;
877 if (cancel_cb) {
878 err = (*cancel_cb)(cancel_arg);
879 if (err)
880 break;
883 qid = STAILQ_FIRST(&tree_ids);
884 STAILQ_REMOVE_HEAD(&tree_ids, entry);
885 path = qid->data;
887 if (got_object_idset_contains(idset, &qid->id) ||
888 got_object_idset_contains(idset_exclude, &qid->id)) {
889 free(qid->data);
890 got_object_qid_free(qid);
891 continue;
894 err = got_pack_add_object(want_meta,
895 want_meta ? idset : idset_exclude,
896 &qid->id, path, GOT_OBJ_TYPE_TREE,
897 mtime, seed, loose_obj_only, repo,
898 ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
899 if (err) {
900 free(qid->data);
901 got_object_qid_free(qid);
902 break;
905 err = got_object_open_as_tree(&tree, repo, &qid->id);
906 if (err) {
907 free(qid->data);
908 got_object_qid_free(qid);
909 break;
912 err = got_pack_load_tree_entries(&tree_ids, want_meta, idset,
913 idset_exclude, tree, path, mtime, seed, repo,
914 loose_obj_only, ncolored, nfound, ntrees,
915 progress_cb, progress_arg, rl,
916 cancel_cb, cancel_arg);
917 free(qid->data);
918 got_object_qid_free(qid);
919 if (err)
920 break;
922 got_object_tree_close(tree);
923 tree = NULL;
926 STAILQ_FOREACH(qid, &tree_ids, entry)
927 free(qid->data);
928 got_object_id_queue_free(&tree_ids);
929 if (tree)
930 got_object_tree_close(tree);
931 return err;
934 static const struct got_error *
935 load_commit(int want_meta, struct got_object_idset *idset,
936 struct got_object_idset *idset_exclude,
937 struct got_object_id *id, struct got_repository *repo, uint32_t seed,
938 int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
939 got_pack_progress_cb progress_cb, void *progress_arg,
940 struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
942 const struct got_error *err;
943 struct got_commit_object *commit;
945 if (got_object_idset_contains(idset, id) ||
946 got_object_idset_contains(idset_exclude, id))
947 return NULL;
949 if (loose_obj_only) {
950 int is_packed;
951 err = search_packidx(&is_packed, id, repo);
952 if (err)
953 return err;
954 if (is_packed && want_meta)
955 return NULL;
958 err = got_object_open_as_commit(&commit, repo, id);
959 if (err)
960 return err;
962 err = got_pack_add_object(want_meta,
963 want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_COMMIT,
964 got_object_commit_get_committer_time(commit), seed,
965 loose_obj_only, repo,
966 ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
967 if (err)
968 goto done;
970 err = got_pack_load_tree(want_meta, idset, idset_exclude,
971 got_object_commit_get_tree_id(commit),
972 "", got_object_commit_get_committer_time(commit), seed,
973 repo, loose_obj_only, ncolored, nfound, ntrees,
974 progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
975 done:
976 got_object_commit_close(commit);
977 return err;
980 static const struct got_error *
981 load_tag(int want_meta, struct got_object_idset *idset,
982 struct got_object_idset *idset_exclude,
983 struct got_object_id *id, struct got_repository *repo, uint32_t seed,
984 int loose_obj_only, int *ncolored, int *nfound, int *ntrees,
985 got_pack_progress_cb progress_cb, void *progress_arg,
986 struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
988 const struct got_error *err;
989 struct got_tag_object *tag = NULL;
991 if (got_object_idset_contains(idset, id) ||
992 got_object_idset_contains(idset_exclude, id))
993 return NULL;
995 if (loose_obj_only) {
996 int is_packed;
997 err = search_packidx(&is_packed, id, repo);
998 if (err)
999 return err;
1000 if (is_packed && want_meta)
1001 return NULL;
1004 err = got_object_open_as_tag(&tag, repo, id);
1005 if (err)
1006 return err;
1008 err = got_pack_add_object(want_meta,
1009 want_meta ? idset : idset_exclude, id, "", GOT_OBJ_TYPE_TAG,
1010 got_object_tag_get_tagger_time(tag), seed, loose_obj_only, repo,
1011 ncolored, nfound, ntrees, progress_cb, progress_arg, rl);
1012 if (err)
1013 goto done;
1015 switch (got_object_tag_get_object_type(tag)) {
1016 case GOT_OBJ_TYPE_COMMIT:
1017 err = load_commit(want_meta, idset, idset_exclude,
1018 got_object_tag_get_object_id(tag), repo, seed,
1019 loose_obj_only, ncolored, nfound, ntrees,
1020 progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1021 break;
1022 case GOT_OBJ_TYPE_TREE:
1023 err = got_pack_load_tree(want_meta, idset, idset_exclude,
1024 got_object_tag_get_object_id(tag), "",
1025 got_object_tag_get_tagger_time(tag), seed, repo,
1026 loose_obj_only, ncolored, nfound, ntrees,
1027 progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1028 break;
1029 default:
1030 break;
1033 done:
1034 got_object_tag_close(tag);
1035 return err;
1038 const struct got_error *
1039 got_pack_paint_commit(struct got_object_qid *qid, intptr_t color)
1041 if (color < 0 || color >= COLOR_MAX)
1042 return got_error(GOT_ERR_RANGE);
1044 qid->data = (void *)color;
1045 return NULL;
1048 const struct got_error *
1049 got_pack_queue_commit_id(struct got_object_id_queue *ids,
1050 struct got_object_id *id, intptr_t color, struct got_repository *repo)
1052 const struct got_error *err;
1053 struct got_object_qid *qid;
1055 err = got_object_qid_alloc(&qid, id);
1056 if (err)
1057 return err;
1059 STAILQ_INSERT_TAIL(ids, qid, entry);
1060 return got_pack_paint_commit(qid, color);
1063 struct append_id_arg {
1064 struct got_object_id **array;
1065 int idx;
1066 struct got_object_idset *drop;
1067 struct got_object_idset *skip;
1070 static const struct got_error *
1071 append_id(struct got_object_id *id, void *data, void *arg)
1073 struct append_id_arg *a = arg;
1075 if (got_object_idset_contains(a->skip, id) ||
1076 got_object_idset_contains(a->drop, id))
1077 return NULL;
1079 a->array[++a->idx] = got_object_id_dup(id);
1080 if (a->array[a->idx] == NULL)
1081 return got_error_from_errno("got_object_id_dup");
1083 return NULL;
1086 static const struct got_error *
1087 free_meta(struct got_object_id *id, void *data, void *arg)
1089 struct got_pack_meta *meta = data;
1091 clear_meta(meta);
1092 free(meta);
1093 return NULL;
1096 static const struct got_error *
1097 queue_commit_or_tag_id(struct got_object_id *id, intptr_t color,
1098 struct got_object_id_queue *ids, struct got_repository *repo)
1100 const struct got_error *err;
1101 struct got_tag_object *tag = NULL;
1102 int obj_type;
1104 err = got_object_get_type(&obj_type, repo, id);
1105 if (err)
1106 return err;
1108 if (obj_type == GOT_OBJ_TYPE_TAG) {
1109 err = got_object_open_as_tag(&tag, repo, id);
1110 if (err)
1111 return err;
1112 obj_type = got_object_tag_get_object_type(tag);
1113 id = got_object_tag_get_object_id(tag);
1116 if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1117 err = got_pack_queue_commit_id(ids, id, color, repo);
1118 if (err)
1119 goto done;
1121 done:
1122 if (tag)
1123 got_object_tag_close(tag);
1124 return err;
1127 const struct got_error *
1128 got_pack_find_pack_for_commit_painting(struct got_packidx **best_packidx,
1129 struct got_object_id_queue *ids, int nids, struct got_repository *repo)
1131 const struct got_error *err = NULL;
1132 struct got_pathlist_entry *pe;
1133 const char *best_packidx_path = NULL;
1134 int nobj_max = 0;
1135 int ncommits_max = 0;
1137 *best_packidx = NULL;
1140 * Find the largest pack which contains at least some of the
1141 * commits we are interested in.
1143 RB_FOREACH(pe, got_pathlist_head, &repo->packidx_paths) {
1144 const char *path_packidx = pe->path;
1145 struct got_packidx *packidx;
1146 int nobj, idx, ncommits = 0;
1147 struct got_object_qid *qid;
1149 err = got_repo_get_packidx(&packidx, path_packidx, repo);
1150 if (err)
1151 break;
1153 nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1154 if (nobj <= nobj_max)
1155 continue;
1157 STAILQ_FOREACH(qid, ids, entry) {
1158 idx = got_packidx_get_object_idx(packidx, &qid->id);
1159 if (idx != -1)
1160 ncommits++;
1162 if (ncommits > ncommits_max) {
1163 best_packidx_path = path_packidx;
1164 nobj_max = nobj;
1165 ncommits_max = ncommits;
1169 if (best_packidx_path && err == NULL) {
1170 err = got_repo_get_packidx(best_packidx, best_packidx_path,
1171 repo);
1174 return err;
1177 static const struct got_error *
1178 findtwixt(struct got_object_id ***res, int *nres, int *ncolored,
1179 struct got_object_id **head, int nhead,
1180 struct got_object_id **tail, int ntail,
1181 struct got_repository *repo,
1182 got_pack_progress_cb progress_cb, void *progress_arg,
1183 struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1185 const struct got_error *err = NULL;
1186 struct got_object_id_queue ids;
1187 struct got_object_idset *keep, *drop, *skip = NULL;
1188 int i, nkeep;
1190 STAILQ_INIT(&ids);
1191 *res = NULL;
1192 *nres = 0;
1193 *ncolored = 0;
1195 keep = got_object_idset_alloc();
1196 if (keep == NULL)
1197 return got_error_from_errno("got_object_idset_alloc");
1199 drop = got_object_idset_alloc();
1200 if (drop == NULL) {
1201 err = got_error_from_errno("got_object_idset_alloc");
1202 goto done;
1205 skip = got_object_idset_alloc();
1206 if (skip == NULL) {
1207 err = got_error_from_errno("got_object_idset_alloc");
1208 goto done;
1211 for (i = 0; i < nhead; i++) {
1212 struct got_object_id *id = head[i];
1213 if (id == NULL)
1214 continue;
1215 err = queue_commit_or_tag_id(id, COLOR_KEEP, &ids, repo);
1216 if (err)
1217 goto done;
1220 for (i = 0; i < ntail; i++) {
1221 struct got_object_id *id = tail[i];
1222 if (id == NULL)
1223 continue;
1224 err = queue_commit_or_tag_id(id, COLOR_DROP, &ids, repo);
1225 if (err)
1226 goto done;
1229 err = got_pack_paint_commits(ncolored, &ids, nhead + ntail,
1230 keep, drop, skip, repo, progress_cb, progress_arg, rl,
1231 cancel_cb, cancel_arg);
1232 if (err)
1233 goto done;
1235 nkeep = got_object_idset_num_elements(keep);
1236 if (nkeep > 0) {
1237 struct append_id_arg arg;
1238 arg.array = calloc(nkeep, sizeof(struct got_object_id *));
1239 if (arg.array == NULL) {
1240 err = got_error_from_errno("calloc");
1241 goto done;
1243 arg.idx = -1;
1244 arg.skip = skip;
1245 arg.drop = drop;
1246 err = got_object_idset_for_each(keep, append_id, &arg);
1247 if (err) {
1248 free(arg.array);
1249 goto done;
1251 *res = arg.array;
1252 *nres = arg.idx + 1;
1254 done:
1255 got_object_idset_free(keep);
1256 got_object_idset_free(drop);
1257 if (skip)
1258 got_object_idset_free(skip);
1259 got_object_id_queue_free(&ids);
1260 return err;
1263 static const struct got_error *
1264 find_pack_for_enumeration(struct got_packidx **best_packidx,
1265 struct got_object_id **ids, int nids, struct got_repository *repo)
1267 const struct got_error *err = NULL;
1268 struct got_pathlist_entry *pe;
1269 const char *best_packidx_path = NULL;
1270 int nobj_max = 0;
1271 int ncommits_max = 0;
1273 *best_packidx = NULL;
1276 * Find the largest pack which contains at least some of the
1277 * commits and tags we are interested in.
1279 RB_FOREACH(pe, got_pathlist_head, &repo->packidx_paths) {
1280 const char *path_packidx = pe->path;
1281 struct got_packidx *packidx;
1282 int nobj, i, idx, ncommits = 0;
1284 err = got_repo_get_packidx(&packidx, path_packidx, repo);
1285 if (err)
1286 break;
1288 nobj = be32toh(packidx->hdr.fanout_table[0xff]);
1289 if (nobj <= nobj_max)
1290 continue;
1292 for (i = 0; i < nids; i++) {
1293 idx = got_packidx_get_object_idx(packidx, ids[i]);
1294 if (idx != -1)
1295 ncommits++;
1297 if (ncommits > ncommits_max) {
1298 best_packidx_path = path_packidx;
1299 nobj_max = nobj;
1300 ncommits_max = ncommits;
1304 if (best_packidx_path && err == NULL) {
1305 err = got_repo_get_packidx(best_packidx, best_packidx_path,
1306 repo);
1309 return err;
1312 static const struct got_error *
1313 load_object_ids(int *ncolored, int *nfound, int *ntrees,
1314 struct got_object_idset *idset, struct got_object_id **theirs, int ntheirs,
1315 struct got_object_id **ours, int nours, struct got_repository *repo,
1316 uint32_t seed, int loose_obj_only, got_pack_progress_cb progress_cb,
1317 void *progress_arg, struct got_ratelimit *rl, got_cancel_cb cancel_cb,
1318 void *cancel_arg)
1320 const struct got_error *err = NULL;
1321 struct got_object_id **ids = NULL;
1322 struct got_packidx *packidx = NULL;
1323 int i, nobj = 0, obj_type, found_all_objects = 0;
1324 struct got_object_idset *idset_exclude;
1326 idset_exclude = got_object_idset_alloc();
1327 if (idset_exclude == NULL)
1328 return got_error_from_errno("got_object_idset_alloc");
1330 *ncolored = 0;
1331 *nfound = 0;
1332 *ntrees = 0;
1334 err = findtwixt(&ids, &nobj, ncolored, ours, nours, theirs, ntheirs,
1335 repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1336 if (err)
1337 goto done;
1339 err = find_pack_for_enumeration(&packidx, theirs, ntheirs, repo);
1340 if (err)
1341 goto done;
1342 if (packidx) {
1343 err = got_pack_load_packed_object_ids(&found_all_objects,
1344 theirs, ntheirs, NULL, 0, 0, seed, idset, idset_exclude,
1345 loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1346 progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1347 if (err)
1348 goto done;
1351 for (i = 0; i < ntheirs; i++) {
1352 struct got_object_id *id = theirs[i];
1353 if (id == NULL)
1354 continue;
1355 err = got_object_get_type(&obj_type, repo, id);
1356 if (err)
1357 return err;
1358 if (obj_type == GOT_OBJ_TYPE_COMMIT) {
1359 if (!found_all_objects) {
1360 err = load_commit(0, idset, idset_exclude,
1361 id, repo, seed, loose_obj_only,
1362 ncolored, nfound, ntrees,
1363 progress_cb, progress_arg, rl,
1364 cancel_cb, cancel_arg);
1365 if (err)
1366 goto done;
1368 } else if (obj_type == GOT_OBJ_TYPE_TAG) {
1369 err = load_tag(0, idset, idset_exclude, id, repo,
1370 seed, loose_obj_only, ncolored, nfound, ntrees,
1371 progress_cb, progress_arg, rl,
1372 cancel_cb, cancel_arg);
1373 if (err)
1374 goto done;
1378 found_all_objects = 0;
1379 err = find_pack_for_enumeration(&packidx, ids, nobj, repo);
1380 if (err)
1381 goto done;
1382 if (packidx) {
1383 err = got_pack_load_packed_object_ids(&found_all_objects, ids,
1384 nobj, theirs, ntheirs, 1, seed, idset, idset_exclude,
1385 loose_obj_only, repo, packidx, ncolored, nfound, ntrees,
1386 progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1387 if (err)
1388 goto done;
1391 if (!found_all_objects) {
1392 for (i = 0; i < nobj; i++) {
1393 err = load_commit(1, idset, idset_exclude, ids[i],
1394 repo, seed, loose_obj_only, ncolored, nfound,
1395 ntrees, progress_cb, progress_arg, rl,
1396 cancel_cb, cancel_arg);
1397 if (err)
1398 goto done;
1402 for (i = 0; i < nours; i++) {
1403 struct got_object_id *id = ours[i];
1404 struct got_pack_meta *m;
1405 if (id == NULL)
1406 continue;
1407 m = got_object_idset_get(idset, id);
1408 if (m == NULL) {
1409 err = got_object_get_type(&obj_type, repo, id);
1410 if (err)
1411 goto done;
1412 } else
1413 obj_type = m->obj_type;
1414 if (obj_type != GOT_OBJ_TYPE_TAG)
1415 continue;
1416 err = load_tag(1, idset, idset_exclude, id, repo,
1417 seed, loose_obj_only, ncolored, nfound, ntrees,
1418 progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1419 if (err)
1420 goto done;
1422 done:
1423 for (i = 0; i < nobj; i++) {
1424 free(ids[i]);
1426 free(ids);
1427 got_object_idset_free(idset_exclude);
1428 return err;
1431 static const struct got_error *
1432 hwrite(int fd, const void *buf, off_t len, struct got_hash *ctx)
1434 got_hash_update(ctx, buf, len);
1435 return got_poll_write_full(fd, buf, len);
1438 static const struct got_error *
1439 hcopy(FILE *fsrc, int fd_dst, off_t len, struct got_hash *ctx)
1441 const struct got_error *err;
1442 unsigned char buf[65536];
1443 off_t remain = len;
1444 size_t n;
1446 while (remain > 0) {
1447 size_t copylen = MIN(sizeof(buf), remain);
1448 n = fread(buf, 1, copylen, fsrc);
1449 if (n != copylen)
1450 return got_ferror(fsrc, GOT_ERR_IO);
1451 got_hash_update(ctx, buf, copylen);
1452 err = got_poll_write_full(fd_dst, buf, copylen);
1453 if (err)
1454 return err;
1455 remain -= copylen;
1458 return NULL;
1461 static const struct got_error *
1462 hcopy_mmap(uint8_t *src, off_t src_offset, size_t src_size,
1463 int fd, off_t len, struct got_hash *ctx)
1465 if (src_offset + len > src_size)
1466 return got_error(GOT_ERR_RANGE);
1468 got_hash_update(ctx, src + src_offset, len);
1469 return got_poll_write_full(fd, src + src_offset, len);
1472 static void
1473 putbe32(char *b, uint32_t n)
1475 b[0] = n >> 24;
1476 b[1] = n >> 16;
1477 b[2] = n >> 8;
1478 b[3] = n >> 0;
1481 static int
1482 write_order_cmp(const void *pa, const void *pb)
1484 struct got_pack_meta *a, *b, *ahd, *bhd;
1486 a = *(struct got_pack_meta **)pa;
1487 b = *(struct got_pack_meta **)pb;
1488 ahd = (a->head == NULL) ? a : a->head;
1489 bhd = (b->head == NULL) ? b : b->head;
1490 if (bhd->mtime < ahd->mtime)
1491 return -1;
1492 if (bhd->mtime > ahd->mtime)
1493 return 1;
1494 if (bhd < ahd)
1495 return -1;
1496 if (bhd > ahd)
1497 return 1;
1498 if (a->nchain != b->nchain)
1499 return a->nchain - b->nchain;
1500 if (a->mtime < b->mtime)
1501 return -1;
1502 if (a->mtime > b->mtime)
1503 return 1;
1504 return got_object_id_cmp(&a->id, &b->id);
1507 static int
1508 reuse_write_order_cmp(const void *pa, const void *pb)
1510 struct got_pack_meta *a, *b;
1512 a = *(struct got_pack_meta **)pa;
1513 b = *(struct got_pack_meta **)pb;
1515 if (a->reused_delta_offset < b->reused_delta_offset)
1516 return -1;
1517 if (a->reused_delta_offset > b->reused_delta_offset)
1518 return 1;
1519 return 0;
1522 static const struct got_error *
1523 packhdr(int *hdrlen, char *hdr, size_t bufsize, int obj_type, size_t len)
1525 size_t i;
1527 *hdrlen = 0;
1529 hdr[0] = obj_type << 4;
1530 hdr[0] |= len & 0xf;
1531 len >>= 4;
1532 for (i = 1; len != 0; i++){
1533 if (i >= bufsize)
1534 return got_error(GOT_ERR_NO_SPACE);
1535 hdr[i - 1] |= GOT_DELTA_SIZE_MORE;
1536 hdr[i] = len & GOT_DELTA_SIZE_VAL_MASK;
1537 len >>= GOT_DELTA_SIZE_SHIFT;
1540 *hdrlen = i;
1541 return NULL;
1544 static int
1545 packoff(char *hdr, off_t off)
1547 int i, j;
1548 char rbuf[8];
1550 rbuf[0] = off & GOT_DELTA_SIZE_VAL_MASK;
1551 for (i = 1; (off >>= GOT_DELTA_SIZE_SHIFT) != 0; i++) {
1552 rbuf[i] = (--off & GOT_DELTA_SIZE_VAL_MASK) |
1553 GOT_DELTA_SIZE_MORE;
1556 j = 0;
1557 while (i > 0)
1558 hdr[j++] = rbuf[--i];
1559 return j;
1562 static const struct got_error *
1563 deltahdr(off_t *packfile_size, struct got_hash *ctx, int packfd,
1564 int force_refdelta, struct got_pack_meta *m)
1566 const struct got_error *err;
1567 char buf[32];
1568 int nh;
1569 size_t digest_len = got_hash_digest_length(m->prev->id.algo);
1571 if (m->prev->off != 0 && !force_refdelta) {
1572 err = packhdr(&nh, buf, sizeof(buf),
1573 GOT_OBJ_TYPE_OFFSET_DELTA, m->delta_len);
1574 if (err)
1575 return err;
1576 nh += packoff(buf + nh, m->off - m->prev->off);
1577 err = hwrite(packfd, buf, nh, ctx);
1578 if (err)
1579 return err;
1580 *packfile_size += nh;
1581 } else {
1582 err = packhdr(&nh, buf, sizeof(buf),
1583 GOT_OBJ_TYPE_REF_DELTA, m->delta_len);
1584 if (err)
1585 return err;
1586 err = hwrite(packfd, buf, nh, ctx);
1587 if (err)
1588 return err;
1589 *packfile_size += nh;
1590 err = hwrite(packfd, m->prev->id.hash, digest_len, ctx);
1591 if (err)
1592 return err;
1593 *packfile_size += digest_len;
1596 return NULL;
1599 static const struct got_error *
1600 write_packed_object(off_t *packfile_size, int packfd,
1601 FILE *delta_cache, uint8_t *delta_cache_map, size_t delta_cache_size,
1602 struct got_pack_meta *m, int *outfd, struct got_hash *ctx,
1603 struct got_repository *repo, int force_refdelta)
1605 const struct got_error *err = NULL;
1606 struct got_deflate_checksum csum;
1607 char buf[32];
1608 int nh;
1609 struct got_raw_object *raw = NULL;
1610 off_t outlen, delta_offset;
1612 memset(&csum, 0, sizeof(csum));
1613 csum.output_ctx = ctx;
1615 if (m->reused_delta_offset)
1616 delta_offset = m->reused_delta_offset;
1617 else
1618 delta_offset = m->delta_offset;
1620 m->off = *packfile_size;
1621 if (m->delta_len == 0) {
1622 err = got_object_raw_open(&raw, outfd, repo, &m->id);
1623 if (err)
1624 goto done;
1625 err = packhdr(&nh, buf, sizeof(buf),
1626 m->obj_type, raw->size);
1627 if (err)
1628 goto done;
1629 err = hwrite(packfd, buf, nh, ctx);
1630 if (err)
1631 goto done;
1632 *packfile_size += nh;
1633 if (raw->f == NULL) {
1634 err = got_deflate_to_fd_mmap(&outlen,
1635 raw->data + raw->hdrlen, 0, raw->size,
1636 packfd, &csum);
1637 if (err)
1638 goto done;
1639 } else {
1640 if (fseeko(raw->f, raw->hdrlen, SEEK_SET)
1641 == -1) {
1642 err = got_error_from_errno("fseeko");
1643 goto done;
1645 err = got_deflate_to_fd(&outlen, raw->f,
1646 raw->size, packfd, &csum);
1647 if (err)
1648 goto done;
1650 *packfile_size += outlen;
1651 got_object_raw_close(raw);
1652 raw = NULL;
1653 } else if (m->delta_buf) {
1654 err = deltahdr(packfile_size, ctx, packfd, force_refdelta, m);
1655 if (err)
1656 goto done;
1657 err = hwrite(packfd, m->delta_buf,
1658 m->delta_compressed_len, ctx);
1659 if (err)
1660 goto done;
1661 *packfile_size += m->delta_compressed_len;
1662 free(m->delta_buf);
1663 m->delta_buf = NULL;
1664 } else if (delta_cache_map) {
1665 err = deltahdr(packfile_size, ctx, packfd, force_refdelta, m);
1666 if (err)
1667 goto done;
1668 err = hcopy_mmap(delta_cache_map, delta_offset,
1669 delta_cache_size, packfd, m->delta_compressed_len,
1670 ctx);
1671 if (err)
1672 goto done;
1673 *packfile_size += m->delta_compressed_len;
1674 } else {
1675 if (fseeko(delta_cache, delta_offset, SEEK_SET) == -1) {
1676 err = got_error_from_errno("fseeko");
1677 goto done;
1679 err = deltahdr(packfile_size, ctx, packfd, force_refdelta, m);
1680 if (err)
1681 goto done;
1682 err = hcopy(delta_cache, packfd,
1683 m->delta_compressed_len, ctx);
1684 if (err)
1685 goto done;
1686 *packfile_size += m->delta_compressed_len;
1688 done:
1689 if (raw)
1690 got_object_raw_close(raw);
1691 return err;
1694 static const struct got_error *
1695 genpack(struct got_object_id *pack_hash, int packfd,
1696 struct got_pack *reuse_pack, FILE *delta_cache,
1697 struct got_pack_meta **deltify, int ndeltify,
1698 struct got_pack_meta **reuse, int nreuse,
1699 int ncolored, int nfound, int ntrees, int nours,
1700 struct got_repository *repo, int force_refdelta,
1701 got_pack_progress_cb progress_cb, void *progress_arg,
1702 struct got_ratelimit *rl,
1703 got_cancel_cb cancel_cb, void *cancel_arg)
1705 const struct got_error *err = NULL;
1706 int i;
1707 struct got_hash ctx;
1708 struct got_pack_meta *m;
1709 char buf[32];
1710 off_t packfile_size = 0;
1711 int outfd = -1;
1712 int delta_cache_fd = -1;
1713 uint8_t *delta_cache_map = NULL;
1714 size_t delta_cache_size = 0;
1715 FILE *packfile = NULL;
1716 enum got_hash_algorithm algo;
1717 size_t digest_len;
1719 algo = got_repo_get_object_format(repo);
1720 digest_len = got_hash_digest_length(algo);
1721 got_hash_init(&ctx, algo);
1723 memset(pack_hash, 0, sizeof(*pack_hash));
1724 pack_hash->algo = algo;
1726 #ifndef GOT_PACK_NO_MMAP
1727 delta_cache_fd = dup(fileno(delta_cache));
1728 if (delta_cache_fd != -1) {
1729 struct stat sb;
1730 if (fstat(delta_cache_fd, &sb) == -1) {
1731 err = got_error_from_errno("fstat");
1732 goto done;
1734 if (sb.st_size > 0 && sb.st_size <= SIZE_MAX) {
1735 delta_cache_map = mmap(NULL, sb.st_size,
1736 PROT_READ, MAP_PRIVATE, delta_cache_fd, 0);
1737 if (delta_cache_map == MAP_FAILED) {
1738 if (errno != ENOMEM) {
1739 err = got_error_from_errno("mmap");
1740 goto done;
1742 delta_cache_map = NULL; /* fallback on stdio */
1743 } else
1744 delta_cache_size = (size_t)sb.st_size;
1747 #endif
1748 err = hwrite(packfd, "PACK", 4, &ctx);
1749 if (err)
1750 goto done;
1751 putbe32(buf, GOT_PACKFILE_VERSION);
1752 err = hwrite(packfd, buf, 4, &ctx);
1753 if (err)
1754 goto done;
1755 putbe32(buf, ndeltify + nreuse);
1756 err = hwrite(packfd, buf, 4, &ctx);
1757 if (err)
1758 goto done;
1760 qsort(deltify, ndeltify, sizeof(struct got_pack_meta *),
1761 write_order_cmp);
1762 for (i = 0; i < ndeltify; i++) {
1763 err = got_pack_report_progress(progress_cb, progress_arg, rl,
1764 ncolored, nfound, ntrees, packfile_size, nours,
1765 ndeltify + nreuse, ndeltify + nreuse, i, 0);
1766 if (err)
1767 goto done;
1768 m = deltify[i];
1769 err = write_packed_object(&packfile_size, packfd,
1770 delta_cache, delta_cache_map, delta_cache_size,
1771 m, &outfd, &ctx, repo, force_refdelta);
1772 if (err)
1773 goto done;
1776 qsort(reuse, nreuse, sizeof(struct got_pack_meta *),
1777 reuse_write_order_cmp);
1778 if (nreuse > 0 && reuse_pack->map == NULL) {
1779 int fd = dup(reuse_pack->fd);
1780 if (fd == -1) {
1781 err = got_error_from_errno("dup");
1782 goto done;
1784 packfile = fdopen(fd, "r");
1785 if (packfile == NULL) {
1786 err = got_error_from_errno("fdopen");
1787 close(fd);
1788 goto done;
1791 for (i = 0; i < nreuse; i++) {
1792 err = got_pack_report_progress(progress_cb, progress_arg, rl,
1793 ncolored, nfound, ntrees, packfile_size, nours,
1794 ndeltify + nreuse, ndeltify + nreuse, ndeltify + i, 0);
1795 if (err)
1796 goto done;
1797 m = reuse[i];
1798 err = write_packed_object(&packfile_size, packfd,
1799 packfile, reuse_pack->map, reuse_pack->filesize,
1800 m, &outfd, &ctx, repo, force_refdelta);
1801 if (err)
1802 goto done;
1805 got_hash_final_object_id(&ctx, pack_hash);
1806 err = got_poll_write_full(packfd, pack_hash->hash, digest_len);
1807 if (err)
1808 goto done;
1809 packfile_size += digest_len;
1810 packfile_size += sizeof(struct got_packfile_hdr);
1811 if (progress_cb) {
1812 err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1813 packfile_size, nours, ndeltify + nreuse,
1814 ndeltify + nreuse, ndeltify + nreuse, 1);
1815 if (err)
1816 goto done;
1818 done:
1819 if (outfd != -1 && close(outfd) == -1 && err == NULL)
1820 err = got_error_from_errno("close");
1821 if (delta_cache_map && munmap(delta_cache_map, delta_cache_size) == -1)
1822 err = got_error_from_errno("munmap");
1823 if (delta_cache_fd != -1 && close(delta_cache_fd) == -1 && err == NULL)
1824 err = got_error_from_errno("close");
1825 if (packfile && fclose(packfile) == EOF && err == NULL)
1826 err = got_error_from_errno("fclose");
1827 return err;
1830 static const struct got_error *
1831 add_meta_idset_cb(struct got_object_id *id, void *data, void *arg)
1833 struct got_pack_meta *m = data;
1834 struct got_pack_metavec *v = arg;
1836 if (m->reused_delta_offset != 0)
1837 return NULL;
1839 return got_pack_add_meta(m, v);
1842 const struct got_error *
1843 got_pack_create(struct got_object_id *packhash, int packfd, FILE *delta_cache,
1844 struct got_object_id **theirs, int ntheirs,
1845 struct got_object_id **ours, int nours,
1846 struct got_repository *repo, int loose_obj_only, int allow_empty,
1847 int force_refdelta, got_pack_progress_cb progress_cb, void *progress_arg,
1848 struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
1850 const struct got_error *err;
1851 struct got_object_idset *idset;
1852 struct got_packidx *reuse_packidx = NULL;
1853 struct got_pack *reuse_pack = NULL;
1854 struct got_pack_metavec deltify, reuse;
1855 int ncolored = 0, nfound = 0, ntrees = 0;
1856 size_t ndeltify;
1857 uint32_t seed;
1859 seed = arc4random();
1861 memset(&deltify, 0, sizeof(deltify));
1862 memset(&reuse, 0, sizeof(reuse));
1864 idset = got_object_idset_alloc();
1865 if (idset == NULL)
1866 return got_error_from_errno("got_object_idset_alloc");
1868 err = load_object_ids(&ncolored, &nfound, &ntrees, idset, theirs,
1869 ntheirs, ours, nours, repo, seed, loose_obj_only,
1870 progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1871 if (err)
1872 goto done;
1874 if (progress_cb) {
1875 err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1876 0L, nours, got_object_idset_num_elements(idset), 0, 0, 0);
1877 if (err)
1878 goto done;
1881 if (got_object_idset_num_elements(idset) == 0 && !allow_empty) {
1882 err = got_error(GOT_ERR_CANNOT_PACK);
1883 goto done;
1886 reuse.metasz = 64;
1887 reuse.meta = calloc(reuse.metasz,
1888 sizeof(struct got_pack_meta *));
1889 if (reuse.meta == NULL) {
1890 err = got_error_from_errno("calloc");
1891 goto done;
1894 err = got_pack_search_deltas(&reuse_packidx, &reuse_pack,
1895 &reuse, idset, ncolored, nfound, ntrees, nours,
1896 repo, progress_cb, progress_arg, rl, cancel_cb, cancel_arg);
1897 if (err)
1898 goto done;
1900 if (reuse_packidx && reuse_pack) {
1901 err = got_repo_pin_pack(repo, reuse_packidx, reuse_pack);
1902 if (err)
1903 goto done;
1906 if (fseeko(delta_cache, 0L, SEEK_END) == -1) {
1907 err = got_error_from_errno("fseeko");
1908 goto done;
1911 ndeltify = got_object_idset_num_elements(idset) - reuse.nmeta;
1912 if (ndeltify > 0) {
1913 deltify.meta = calloc(ndeltify, sizeof(struct got_pack_meta *));
1914 if (deltify.meta == NULL) {
1915 err = got_error_from_errno("calloc");
1916 goto done;
1918 deltify.metasz = ndeltify;
1920 err = got_object_idset_for_each(idset, add_meta_idset_cb,
1921 &deltify);
1922 if (err)
1923 goto done;
1924 if (deltify.nmeta > 0) {
1925 err = pick_deltas(deltify.meta, deltify.nmeta,
1926 ncolored, nfound, ntrees, nours, reuse.nmeta,
1927 delta_cache, repo, progress_cb, progress_arg, rl,
1928 cancel_cb, cancel_arg);
1929 if (err)
1930 goto done;
1934 if (fflush(delta_cache) == EOF) {
1935 err = got_error_from_errno("fflush");
1936 goto done;
1939 if (progress_cb) {
1941 * Report a 1-byte packfile write to indicate we are about
1942 * to start sending packfile data. gotd(8) needs this.
1944 err = progress_cb(progress_arg, ncolored, nfound, ntrees,
1945 1 /* packfile_size */, nours,
1946 got_object_idset_num_elements(idset),
1947 deltify.nmeta + reuse.nmeta, 0, 0);
1948 if (err)
1949 goto done;
1952 /* Pinned pack may have moved to different cache slot. */
1953 reuse_pack = got_repo_get_pinned_pack(repo);
1955 err = genpack(packhash, packfd, reuse_pack, delta_cache, deltify.meta,
1956 deltify.nmeta, reuse.meta, reuse.nmeta, ncolored, nfound, ntrees,
1957 nours, repo, force_refdelta, progress_cb, progress_arg, rl,
1958 cancel_cb, cancel_arg);
1959 if (err)
1960 goto done;
1961 done:
1962 free_nmeta(deltify.meta, deltify.nmeta);
1963 free_nmeta(reuse.meta, reuse.nmeta);
1964 got_object_idset_for_each(idset, free_meta, NULL);
1965 got_object_idset_free(idset);
1966 got_repo_unpin_pack(repo);
1967 return err;