lib/object.c

   1 /*
   2  * Copyright (c) 2018, 2019 Stefan Sperling <stsp@openbsd.org>
   3  *
   4  * Permission to use, copy, modify, and distribute this software for any
   5  * purpose with or without fee is hereby granted, provided that the above
   6  * copyright notice and this permission notice appear in all copies.
   7  *
   8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
   9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  15  */
  16
  17 #include <sys/types.h>
  18 #include <sys/stat.h>
  19 #include <sys/queue.h>
  20 #include <sys/uio.h>
  21 #include <sys/mman.h>
  22
  23 #include <errno.h>
  24 #include <fcntl.h>
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27 #include <string.h>
  28 #include <unistd.h>
  29 #include <zlib.h>
  30 #include <libgen.h>
  31 #include <limits.h>
  32 #include <imsg.h>
  33
  34 #include "got_compat.h"
  35
  36 #include "got_error.h"
  37 #include "got_object.h"
  38 #include "got_repository.h"
  39 #include "got_opentemp.h"
  40 #include "got_path.h"
  41
  42 #include "got_lib_sha1.h"
  43 #include "got_lib_delta.h"
  44 #include "got_lib_inflate.h"
  45 #include "got_lib_object.h"
  46 #include "got_lib_object_idcache.h"
  47 #include "got_lib_object_cache.h"
  48 #include "got_lib_object_parse.h"
  49 #include "got_lib_pack.h"
  50 #include "got_lib_repository.h"
  51
  52 #ifndef MIN
  53 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
  54 #endif
  55
  56 #ifndef nitems
  57 #define nitems(_a)      (sizeof((_a)) / sizeof((_a)[0]))
  58 #endif
  59
  60 struct got_object_id *
  61 got_object_get_id(struct got_object *obj)
  62 {
  63         return &obj->id;
  64 }
  65
  66 const struct got_error *
  67 got_object_get_id_str(char **outbuf, struct got_object *obj)
  68 {
  69         return got_object_id_str(outbuf, &obj->id);
  70 }
  71
  72 const struct got_error *
  73 got_object_get_type(int *type, struct got_repository *repo,
  74     struct got_object_id *id)
  75 {
  76         const struct got_error *err = NULL;
  77         struct got_object *obj;
  78
  79         err = got_object_open(&obj, repo, id);
  80         if (err)
  81                 return err;
  82
  83         switch (obj->type) {
  84         case GOT_OBJ_TYPE_COMMIT:
  85         case GOT_OBJ_TYPE_TREE:
  86         case GOT_OBJ_TYPE_BLOB:
  87         case GOT_OBJ_TYPE_TAG:
  88                 *type = obj->type;
  89                 break;
  90         default:
  91                 err = got_error(GOT_ERR_OBJ_TYPE);
  92                 break;
  93         }
  94
  95         got_object_close(obj);
  96         return err;
  97 }
  98
  99 const struct got_error *
 100 got_object_get_path(char **path, struct got_object_id *id,
 101     struct got_repository *repo)
 102 {
 103         const struct got_error *err = NULL;
 104         char *hex = NULL;
 105         char *path_objects;
 106
 107         *path = NULL;
 108
 109         path_objects = got_repo_get_path_objects(repo);
 110         if (path_objects == NULL)
 111                 return got_error_from_errno("got_repo_get_path_objects");
 112
 113         err = got_object_id_str(&hex, id);
 114         if (err)
 115                 goto done;
 116
 117         if (asprintf(path, "%s/%.2x/%s", path_objects,
 118             id->sha1[0], hex + 2) == -1)
 119                 err = got_error_from_errno("asprintf");
 120
 121 done:
 122         free(hex);
 123         free(path_objects);
 124         return err;
 125 }
 126
 127 const struct got_error *
 128 got_object_open_loose_fd(int *fd, struct got_object_id *id,
 129     struct got_repository *repo)
 130 {
 131         const struct got_error *err = NULL;
 132         char *path;
 133
 134         err = got_object_get_path(&path, id, repo);
 135         if (err)
 136                 return err;
 137         *fd = open(path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
 138         if (*fd == -1) {
 139                 err = got_error_from_errno2("open", path);
 140                 goto done;
 141         }
 142 done:
 143         free(path);
 144         return err;
 145 }
 146
 147 const struct got_error *
 148 got_object_open_by_id_str(struct got_object **obj, struct got_repository *repo,
 149     const char *id_str)
 150 {
 151         struct got_object_id id;
 152
 153         if (!got_parse_sha1_digest(id.sha1, id_str))
 154                 return got_error_path(id_str, GOT_ERR_BAD_OBJ_ID_STR);
 155
 156         return got_object_open(obj, repo, &id);
 157 }
 158
 159 const struct got_error *
 160 got_object_resolve_id_str(struct got_object_id **id,
 161     struct got_repository *repo, const char *id_str)
 162 {
 163         const struct got_error *err = NULL;
 164         struct got_object *obj;
 165
 166         err = got_object_open_by_id_str(&obj, repo, id_str);
 167         if (err)
 168                 return err;
 169
 170         *id = got_object_id_dup(got_object_get_id(obj));
 171         got_object_close(obj);
 172         if (*id == NULL)
 173                 return got_error_from_errno("got_object_id_dup");
 174
 175         return NULL;
 176 }
 177
 178 const struct got_error *
 179 got_object_qid_alloc(struct got_object_qid **qid, struct got_object_id *id)
 180 {
 181         *qid = calloc(1, sizeof(**qid));
 182         if (*qid == NULL)
 183                 return got_error_from_errno("calloc");
 184
 185         memcpy(&(*qid)->id, id, sizeof((*qid)->id));
 186         return NULL;
 187 }
 188
 189 const struct got_error *
 190 got_object_id_queue_copy(const struct got_object_id_queue *src,
 191     struct got_object_id_queue *dest)
 192 {
 193         const struct got_error *err;
 194         struct got_object_qid *qid;
 195
 196         STAILQ_FOREACH(qid, src, entry) {
 197                 struct got_object_qid *new;
 198                 /*
 199                  * Deep-copy the object ID only. Let the caller deal
 200                  * with setting up the new->data pointer if needed.
 201                  */
 202                 err = got_object_qid_alloc(&new, &qid->id);
 203                 if (err) {
 204                         got_object_id_queue_free(dest);
 205                         return err;
 206                 }
 207                 STAILQ_INSERT_TAIL(dest, new, entry);
 208         }
 209
 210         return NULL;
 211 }
 212
 213 int
 214 got_object_tree_get_nentries(struct got_tree_object *tree)
 215 {
 216         return tree->nentries;
 217 }
 218
 219 struct got_tree_entry *
 220 got_object_tree_get_first_entry(struct got_tree_object *tree)
 221 {
 222         return got_object_tree_get_entry(tree, 0);
 223 }
 224
 225 struct got_tree_entry *
 226 got_object_tree_get_last_entry(struct got_tree_object *tree)
 227 {
 228         return got_object_tree_get_entry(tree, tree->nentries - 1);
 229 }
 230
 231 struct got_tree_entry *
 232 got_object_tree_get_entry(struct got_tree_object *tree, int i)
 233 {
 234         if (i < 0 || i >= tree->nentries)
 235                 return NULL;
 236         return &tree->entries[i];
 237 }
 238
 239 mode_t
 240 got_tree_entry_get_mode(struct got_tree_entry *te)
 241 {
 242         return te->mode;
 243 }
 244
 245 const char *
 246 got_tree_entry_get_name(struct got_tree_entry *te)
 247 {
 248         return &te->name[0];
 249 }
 250
 251 struct got_object_id *
 252 got_tree_entry_get_id(struct got_tree_entry *te)
 253 {
 254         return &te->id;
 255 }
 256
 257 const struct got_error *
 258 got_object_blob_read_to_str(char **s, struct got_blob_object *blob)
 259 {
 260         const struct got_error *err = NULL;
 261         size_t len, totlen, hdrlen, offset;
 262
 263         *s = NULL;
 264
 265         hdrlen = got_object_blob_get_hdrlen(blob);
 266         totlen = 0;
 267         offset = 0;
 268         do {
 269                 char *p;
 270
 271                 err = got_object_blob_read_block(&len, blob);
 272                 if (err)
 273                         return err;
 274
 275                 if (len == 0)
 276                         break;
 277
 278                 totlen += len - hdrlen;
 279                 p = realloc(*s, totlen + 1);
 280                 if (p == NULL) {
 281                         err = got_error_from_errno("realloc");
 282                         free(*s);
 283                         *s = NULL;
 284                         return err;
 285                 }
 286                 *s = p;
 287                 /* Skip blob object header first time around. */
 288                 memcpy(*s + offset,
 289                     got_object_blob_get_read_buf(blob) + hdrlen, len - hdrlen);
 290                 hdrlen = 0;
 291                 offset = totlen;
 292         } while (len > 0);
 293
 294         (*s)[totlen] = '\0';
 295         return NULL;
 296 }
 297
 298 const struct got_error *
 299 got_tree_entry_get_symlink_target(char **link_target, struct got_tree_entry *te,
 300     struct got_repository *repo)
 301 {
 302         const struct got_error *err = NULL;
 303         struct got_blob_object *blob = NULL;
 304         int fd = -1;
 305
 306         *link_target = NULL;
 307
 308         if (!got_object_tree_entry_is_symlink(te))
 309                 return got_error(GOT_ERR_TREE_ENTRY_TYPE);
 310
 311         fd = got_opentempfd();
 312         if (fd == -1) {
 313                 err = got_error_from_errno("got_opentempfd");
 314                 goto done;
 315         }
 316
 317         err = got_object_open_as_blob(&blob, repo,
 318             got_tree_entry_get_id(te), PATH_MAX, fd);
 319         if (err)
 320                 goto done;
 321
 322         err = got_object_blob_read_to_str(link_target, blob);
 323 done:
 324         if (fd != -1 && close(fd) == -1 && err == NULL)
 325                 err = got_error_from_errno("close");
 326         if (blob)
 327                 got_object_blob_close(blob);
 328         if (err) {
 329                 free(*link_target);
 330                 *link_target = NULL;
 331         }
 332         return err;
 333 }
 334
 335 int
 336 got_tree_entry_get_index(struct got_tree_entry *te)
 337 {
 338         return te->idx;
 339 }
 340
 341 struct got_tree_entry *
 342 got_tree_entry_get_next(struct got_tree_object *tree,
 343     struct got_tree_entry *te)
 344 {
 345         return got_object_tree_get_entry(tree, te->idx + 1);
 346 }
 347
 348 struct got_tree_entry *
 349 got_tree_entry_get_prev(struct got_tree_object *tree,
 350     struct got_tree_entry *te)
 351 {
 352         return got_object_tree_get_entry(tree, te->idx - 1);
 353 }
 354
 355 const struct got_error *
 356 got_object_blob_close(struct got_blob_object *blob)
 357 {
 358         const struct got_error *err = NULL;
 359         free(blob->read_buf);
 360         if (blob->f && fclose(blob->f) == EOF)
 361                 err = got_error_from_errno("fclose");
 362         free(blob->data);
 363         free(blob);
 364         return err;
 365 }
 366
 367 void
 368 got_object_blob_rewind(struct got_blob_object *blob)
 369 {
 370         if (blob->f)
 371                 rewind(blob->f);
 372 }
 373
 374 char *
 375 got_object_blob_id_str(struct got_blob_object *blob, char *buf, size_t size)
 376 {
 377         return got_object_id_hex(&blob->id, buf, size);
 378 }
 379
 380 size_t
 381 got_object_blob_get_hdrlen(struct got_blob_object *blob)
 382 {
 383         return blob->hdrlen;
 384 }
 385
 386 const uint8_t *
 387 got_object_blob_get_read_buf(struct got_blob_object *blob)
 388 {
 389         return blob->read_buf;
 390 }
 391
 392 const struct got_error *
 393 got_object_blob_read_block(size_t *outlenp, struct got_blob_object *blob)
 394 {
 395         size_t n;
 396
 397         n = fread(blob->read_buf, 1, blob->blocksize, blob->f);
 398         if (n == 0 && ferror(blob->f))
 399                 return got_ferror(blob->f, GOT_ERR_IO);
 400         *outlenp = n;
 401         return NULL;
 402 }
 403
 404 const struct got_error *
 405 got_object_blob_is_binary(int *binary, struct got_blob_object *blob)
 406 {
 407         const struct got_error *err;
 408         size_t hdrlen, len;
 409
 410         *binary = 0;
 411         hdrlen = got_object_blob_get_hdrlen(blob);
 412
 413         if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
 414                 return got_error_from_errno("fseeko");
 415
 416         err = got_object_blob_read_block(&len, blob);
 417         if (err)
 418                 return err;
 419
 420         *binary = memchr(blob->read_buf, '\0', len) != NULL;
 421
 422         if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
 423                 return got_error_from_errno("fseeko");
 424         return NULL;
 425 }
 426
 427 const struct got_error *
 428 got_object_blob_getline(char **line, ssize_t *linelen, size_t *linesize,
 429     struct got_blob_object *blob)
 430 {
 431         *linelen = getline(line, linesize, blob->f);
 432         if (*linelen == -1 && !feof(blob->f))
 433                 return got_error_from_errno("getline");
 434         return NULL;
 435 }
 436
 437 const struct got_error *
 438 got_object_blob_dump_to_file(off_t *filesize, int *nlines,
 439     off_t **line_offsets, FILE *outfile, struct got_blob_object *blob)
 440 {
 441         const struct got_error *err = NULL;
 442         size_t n, len, hdrlen;
 443         const uint8_t *buf;
 444         int i;
 445         const int alloc_chunksz = 512;
 446         size_t nalloc = 0;
 447         off_t off = 0, total_len = 0;
 448
 449         if (line_offsets)
 450                 *line_offsets = NULL;
 451         if (filesize)
 452                 *filesize = 0;
 453         if (nlines)
 454                 *nlines = 0;
 455
 456         hdrlen = got_object_blob_get_hdrlen(blob);
 457         do {
 458                 err = got_object_blob_read_block(&len, blob);
 459                 if (err)
 460                         return err;
 461                 if (len == 0)
 462                         break;
 463                 buf = got_object_blob_get_read_buf(blob);
 464                 i = hdrlen;
 465                 if (nlines) {
 466                         if (line_offsets && *line_offsets == NULL) {
 467                                 /* Have some data but perhaps no '\n'. */
 468                                 *nlines = 1;
 469                                 nalloc = alloc_chunksz;
 470                                 *line_offsets = calloc(nalloc,
 471                                     sizeof(**line_offsets));
 472                                 if (*line_offsets == NULL)
 473                                         return got_error_from_errno("calloc");
 474
 475                                 /* Skip forward over end of first line. */
 476                                 while (i < len) {
 477                                         if (buf[i] == '\n')
 478                                                 break;
 479                                         i++;
 480                                 }
 481                         }
 482                         /* Scan '\n' offsets in remaining chunk of data. */
 483                         while (i < len) {
 484                                 if (buf[i] != '\n') {
 485                                         i++;
 486                                         continue;
 487                                 }
 488                                 (*nlines)++;
 489                                 if (line_offsets && nalloc < *nlines) {
 490                                         size_t n = *nlines + alloc_chunksz;
 491                                         off_t *o = recallocarray(*line_offsets,
 492                                             nalloc, n, sizeof(**line_offsets));
 493                                         if (o == NULL) {
 494                                                 free(*line_offsets);
 495                                                 *line_offsets = NULL;
 496                                                 return got_error_from_errno(
 497                                                     "recallocarray");
 498                                         }
 499                                         *line_offsets = o;
 500                                         nalloc = n;
 501                                 }
 502                                 if (line_offsets) {
 503                                         off = total_len + i - hdrlen + 1;
 504                                         (*line_offsets)[*nlines - 1] = off;
 505                                 }
 506                                 i++;
 507                         }
 508                 }
 509                 /* Skip blob object header first time around. */
 510                 n = fwrite(buf + hdrlen, 1, len - hdrlen, outfile);
 511                 if (n != len - hdrlen)
 512                         return got_ferror(outfile, GOT_ERR_IO);
 513                 total_len += len - hdrlen;
 514                 hdrlen = 0;
 515         } while (len != 0);
 516
 517         if (fflush(outfile) != 0)
 518                 return got_error_from_errno("fflush");
 519         rewind(outfile);
 520
 521         if (filesize)
 522                 *filesize = total_len;
 523
 524         return NULL;
 525 }
 526
 527 const char *
 528 got_object_tag_get_name(struct got_tag_object *tag)
 529 {
 530         return tag->tag;
 531 }
 532
 533 int
 534 got_object_tag_get_object_type(struct got_tag_object *tag)
 535 {
 536         return tag->obj_type;
 537 }
 538
 539 struct got_object_id *
 540 got_object_tag_get_object_id(struct got_tag_object *tag)
 541 {
 542         return &tag->id;
 543 }
 544
 545 time_t
 546 got_object_tag_get_tagger_time(struct got_tag_object *tag)
 547 {
 548         return tag->tagger_time;
 549 }
 550
 551 time_t
 552 got_object_tag_get_tagger_gmtoff(struct got_tag_object *tag)
 553 {
 554         return tag->tagger_gmtoff;
 555 }
 556
 557 const char *
 558 got_object_tag_get_tagger(struct got_tag_object *tag)
 559 {
 560         return tag->tagger;
 561 }
 562
 563 const char *
 564 got_object_tag_get_message(struct got_tag_object *tag)
 565 {
 566         return tag->tagmsg;
 567 }
 568
 569 static struct got_tree_entry *
 570 find_entry_by_name(struct got_tree_object *tree, const char *name, size_t len)
 571 {
 572         int i;
 573
 574         /* Note that tree entries are sorted in strncmp() order. */
 575         for (i = 0; i < tree->nentries; i++) {
 576                 struct got_tree_entry *te = &tree->entries[i];
 577                 int cmp = strncmp(te->name, name, len);
 578                 if (cmp < 0)
 579                         continue;
 580                 if (cmp > 0)
 581                         break;
 582                 if (te->name[len] == '\0')
 583                         return te;
 584         }
 585         return NULL;
 586 }
 587
 588 struct got_tree_entry *
 589 got_object_tree_find_entry(struct got_tree_object *tree, const char *name)
 590 {
 591         return find_entry_by_name(tree, name, strlen(name));
 592 }
 593
 594 const struct got_error *
 595 got_object_tree_find_path(struct got_object_id **id, mode_t *mode,
 596     struct got_repository *repo, struct got_tree_object *tree,
 597     const char *path)
 598 {
 599         const struct got_error *err = NULL;
 600         struct got_tree_object *subtree = NULL;
 601         struct got_tree_entry *te = NULL;
 602         const char *seg, *s;
 603         size_t seglen;
 604
 605         *id = NULL;
 606
 607         s = path;
 608         while (s[0] == '/')
 609                 s++;
 610         seg = s;
 611         seglen = 0;
 612         subtree = tree;
 613         while (*s) {
 614                 struct got_tree_object *next_tree;
 615
 616                 if (*s != '/') {
 617                         s++;
 618                         seglen++;
 619                         if (*s)
 620                                 continue;
 621                 }
 622
 623                 te = find_entry_by_name(subtree, seg, seglen);
 624                 if (te == NULL) {
 625                         err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
 626                         goto done;
 627                 }
 628
 629                 if (*s == '\0')
 630                         break;
 631
 632                 seg = s + 1;
 633                 seglen = 0;
 634                 s++;
 635                 if (*s) {
 636                         err = got_object_open_as_tree(&next_tree, repo,
 637                             &te->id);
 638                         te = NULL;
 639                         if (err)
 640                                 goto done;
 641                         if (subtree != tree)
 642                                 got_object_tree_close(subtree);
 643                         subtree = next_tree;
 644                 }
 645         }
 646
 647         if (te) {
 648                 *id = got_object_id_dup(&te->id);
 649                 if (*id == NULL)
 650                         return got_error_from_errno("got_object_id_dup");
 651                 if (mode)
 652                         *mode = te->mode;
 653         } else
 654                 err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
 655 done:
 656         if (subtree && subtree != tree)
 657                 got_object_tree_close(subtree);
 658         return err;
 659 }
 660
 661 const struct got_error *
 662 got_object_id_by_path(struct got_object_id **id, struct got_repository *repo,
 663     struct got_commit_object *commit, const char *path)
 664 {
 665         const struct got_error *err = NULL;
 666         struct got_tree_object *tree = NULL;
 667
 668         *id = NULL;
 669
 670         /* Handle opening of root of commit's tree. */
 671         if (got_path_is_root_dir(path)) {
 672                 *id = got_object_id_dup(commit->tree_id);
 673                 if (*id == NULL)
 674                         err = got_error_from_errno("got_object_id_dup");
 675         } else {
 676                 err = got_object_open_as_tree(&tree, repo, commit->tree_id);
 677                 if (err)
 678                         goto done;
 679                 err = got_object_tree_find_path(id, NULL, repo, tree, path);
 680         }
 681 done:
 682         if (tree)
 683                 got_object_tree_close(tree);
 684         return err;
 685 }
 686
 687 /*
 688  * Normalize file mode bits to avoid false positive tree entry differences
 689  * in case tree entries have unexpected mode bits set.
 690  */
 691 static mode_t
 692 normalize_mode_for_comparison(mode_t mode)
 693 {
 694         /*
 695          * For directories, the only relevant bit is the IFDIR bit.
 696          * This allows us to detect paths changing from a directory
 697          * to a file and vice versa.
 698          */
 699         if (S_ISDIR(mode))
 700                 return mode & S_IFDIR;
 701
 702         /*
 703          * For symlinks, the only relevant bit is the IFLNK bit.
 704          * This allows us to detect paths changing from a symlinks
 705          * to a file or directory and vice versa.
 706          */
 707         if (S_ISLNK(mode))
 708                 return mode & S_IFLNK;
 709
 710         /* For files, the only change we care about is the executable bit. */
 711         return mode & S_IXUSR;
 712 }
 713
 714 const struct got_error *
 715 got_object_tree_path_changed(int *changed,
 716     struct got_tree_object *tree01, struct got_tree_object *tree02,
 717     const char *path, struct got_repository *repo)
 718 {
 719         const struct got_error *err = NULL;
 720         struct got_tree_object *tree1 = NULL, *tree2 = NULL;
 721         struct got_tree_entry *te1 = NULL, *te2 = NULL;
 722         const char *seg, *s;
 723         size_t seglen;
 724
 725         *changed = 0;
 726
 727         /* We not do support comparing the root path. */
 728         if (got_path_is_root_dir(path))
 729                 return got_error_path(path, GOT_ERR_BAD_PATH);
 730
 731         tree1 = tree01;
 732         tree2 = tree02;
 733         s = path;
 734         while (*s == '/')
 735                 s++;
 736         seg = s;
 737         seglen = 0;
 738         while (*s) {
 739                 struct got_tree_object *next_tree1, *next_tree2;
 740                 mode_t mode1, mode2;
 741
 742                 if (*s != '/') {
 743                         s++;
 744                         seglen++;
 745                         if (*s)
 746                                 continue;
 747                 }
 748
 749                 te1 = find_entry_by_name(tree1, seg, seglen);
 750                 if (te1 == NULL) {
 751                         err = got_error(GOT_ERR_NO_OBJ);
 752                         goto done;
 753                 }
 754
 755                 if (tree2)
 756                         te2 = find_entry_by_name(tree2, seg, seglen);
 757
 758                 if (te2) {
 759                         mode1 = normalize_mode_for_comparison(te1->mode);
 760                         mode2 = normalize_mode_for_comparison(te2->mode);
 761                         if (mode1 != mode2) {
 762                                 *changed = 1;
 763                                 goto done;
 764                         }
 765
 766                         if (got_object_id_cmp(&te1->id, &te2->id) == 0) {
 767                                 *changed = 0;
 768                                 goto done;
 769                         }
 770                 }
 771
 772                 if (*s == '\0') { /* final path element */
 773                         *changed = 1;
 774                         goto done;
 775                 }
 776
 777                 seg = s + 1;
 778                 s++;
 779                 seglen = 0;
 780                 if (*s) {
 781                         err = got_object_open_as_tree(&next_tree1, repo,
 782                             &te1->id);
 783                         te1 = NULL;
 784                         if (err)
 785                                 goto done;
 786                         if (tree1 != tree01)
 787                                 got_object_tree_close(tree1);
 788                         tree1 = next_tree1;
 789
 790                         if (te2) {
 791                                 err = got_object_open_as_tree(&next_tree2, repo,
 792                                     &te2->id);
 793                                 te2 = NULL;
 794                                 if (err)
 795                                         goto done;
 796                                 if (tree2 != tree02)
 797                                         got_object_tree_close(tree2);
 798                                 tree2 = next_tree2;
 799                         } else if (tree2) {
 800                                 if (tree2 != tree02)
 801                                         got_object_tree_close(tree2);
 802                                 tree2 = NULL;
 803                         }
 804                 }
 805         }
 806 done:
 807         if (tree1 && tree1 != tree01)
 808                 got_object_tree_close(tree1);
 809         if (tree2 && tree2 != tree02)
 810                 got_object_tree_close(tree2);
 811         return err;
 812 }
 813
 814 const struct got_error *
 815 got_object_tree_entry_dup(struct got_tree_entry **new_te,
 816     struct got_tree_entry *te)
 817 {
 818         const struct got_error *err = NULL;
 819
 820         *new_te = calloc(1, sizeof(**new_te));
 821         if (*new_te == NULL)
 822                 return got_error_from_errno("calloc");
 823
 824         (*new_te)->mode = te->mode;
 825         memcpy((*new_te)->name, te->name, sizeof((*new_te)->name));
 826         memcpy(&(*new_te)->id, &te->id, sizeof((*new_te)->id));
 827         return err;
 828 }
 829
 830 int
 831 got_object_tree_entry_is_submodule(struct got_tree_entry *te)
 832 {
 833         return (te->mode & S_IFMT) == (S_IFDIR | S_IFLNK);
 834 }
 835
 836 int
 837 got_object_tree_entry_is_symlink(struct got_tree_entry *te)
 838 {
 839         /* S_IFDIR check avoids confusing symlinks with submodules. */
 840         return ((te->mode & (S_IFDIR | S_IFLNK)) == S_IFLNK);
 841 }
 842
 843 static const struct got_error *
 844 resolve_symlink(char **link_target, const char *path,
 845     struct got_commit_object *commit, struct got_repository *repo)
 846 {
 847         const struct got_error *err = NULL;
 848         char buf[PATH_MAX];
 849         char *name, *parent_path = NULL;
 850         struct got_object_id *tree_obj_id = NULL;
 851         struct got_tree_object *tree = NULL;
 852         struct got_tree_entry *te = NULL;
 853
 854         *link_target = NULL;
 855
 856         if (strlcpy(buf, path, sizeof(buf)) >= sizeof(buf))
 857                 return got_error(GOT_ERR_NO_SPACE);
 858
 859         name = basename(buf);
 860         if (name == NULL)
 861                 return got_error_from_errno2("basename", path);
 862
 863         err = got_path_dirname(&parent_path, path);
 864         if (err)
 865                 return err;
 866
 867         err = got_object_id_by_path(&tree_obj_id, repo, commit,
 868             parent_path);
 869         if (err) {
 870                 if (err->code == GOT_ERR_NO_TREE_ENTRY) {
 871                         /* Display the complete path in error message. */
 872                         err = got_error_path(path, err->code);
 873                 }
 874                 goto done;
 875         }
 876
 877         err = got_object_open_as_tree(&tree, repo, tree_obj_id);
 878         if (err)
 879                 goto done;
 880
 881         te = got_object_tree_find_entry(tree, name);
 882         if (te == NULL) {
 883                 err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
 884                 goto done;
 885         }
 886
 887         if (got_object_tree_entry_is_symlink(te)) {
 888                 err = got_tree_entry_get_symlink_target(link_target, te, repo);
 889                 if (err)
 890                         goto done;
 891                 if (!got_path_is_absolute(*link_target)) {
 892                         char *abspath;
 893                         if (asprintf(&abspath, "%s/%s", parent_path,
 894                             *link_target) == -1) {
 895                                 err = got_error_from_errno("asprintf");
 896                                 goto done;
 897                         }
 898                         free(*link_target);
 899                         *link_target = malloc(PATH_MAX);
 900                         if (*link_target == NULL) {
 901                                 err = got_error_from_errno("malloc");
 902                                 goto done;
 903                         }
 904                         err = got_canonpath(abspath, *link_target, PATH_MAX);
 905                         free(abspath);
 906                         if (err)
 907                                 goto done;
 908                 }
 909         }
 910 done:
 911         free(parent_path);
 912         free(tree_obj_id);
 913         if (tree)
 914                 got_object_tree_close(tree);
 915         if (err) {
 916                 free(*link_target);
 917                 *link_target = NULL;
 918         }
 919         return err;
 920 }
 921
 922 const struct got_error *
 923 got_object_resolve_symlinks(char **link_target, const char *path,
 924     struct got_commit_object *commit, struct got_repository *repo)
 925 {
 926         const struct got_error *err = NULL;
 927         char *next_target = NULL;
 928         int max_recursion = 40; /* matches Git */
 929
 930         *link_target = NULL;
 931
 932         do {
 933                 err = resolve_symlink(&next_target,
 934                     *link_target ? *link_target : path, commit, repo);
 935                 if (err)
 936                         break;
 937                 if (next_target) {
 938                         free(*link_target);
 939                         if (--max_recursion == 0) {
 940                                 err = got_error_path(path, GOT_ERR_RECURSION);
 941                                 *link_target = NULL;
 942                                 break;
 943                         }
 944                         *link_target = next_target;
 945                 }
 946         } while (next_target);
 947
 948         return err;
 949 }
 950
 951 void
 952 got_object_commit_retain(struct got_commit_object *commit)
 953 {
 954         commit->refcnt++;
 955 }
 956
 957 const struct got_error *
 958 got_object_raw_alloc(struct got_raw_object **obj, uint8_t *outbuf, int *outfd,
 959     size_t max_in_mem_size, size_t hdrlen, off_t size)
 960 {
 961         const struct got_error *err = NULL;
 962         off_t tot;
 963
 964         tot = hdrlen + size;
 965
 966         *obj = calloc(1, sizeof(**obj));
 967         if (*obj == NULL) {
 968                 err = got_error_from_errno("calloc");
 969                 goto done;
 970         }
 971         (*obj)->fd = -1;
 972         (*obj)->tempfile_idx = -1;
 973
 974         if (outbuf) {
 975                 (*obj)->data = outbuf;
 976         } else {
 977                 struct stat sb;
 978                 if (fstat(*outfd, &sb) == -1) {
 979                         err = got_error_from_errno("fstat");
 980                         goto done;
 981                 }
 982
 983                 if (sb.st_size != tot) {
 984                         err = got_error_msg(GOT_ERR_BAD_OBJ_HDR,
 985                             "raw object has unexpected size");
 986                         goto done;
 987                 }
 988 #ifndef GOT_PACK_NO_MMAP
 989                 if (tot > 0 && tot <= max_in_mem_size) {
 990                         (*obj)->data = mmap(NULL, tot, PROT_READ,
 991                             MAP_PRIVATE, *outfd, 0);
 992                         if ((*obj)->data == MAP_FAILED) {
 993                                 if (errno != ENOMEM) {
 994                                         err = got_error_from_errno("mmap");
 995                                         goto done;
 996                                 }
 997                                 (*obj)->data = NULL;
 998                         } else {
 999                                 (*obj)->fd = *outfd;
1000                                 *outfd = -1;
1001                         }
1002                 }
1003 #endif
1004                 if (*outfd != -1) {
1005                         (*obj)->f = fdopen(*outfd, "r");
1006                         if ((*obj)->f == NULL) {
1007                                 err = got_error_from_errno("fdopen");
1008                                 goto done;
1009                         }
1010                         *outfd = -1;
1011                 }
1012         }
1013         (*obj)->hdrlen = hdrlen;
1014         (*obj)->size = size;
1015 done:
1016         if (err) {
1017                 if (*obj) {
1018                         got_object_raw_close(*obj);
1019                         *obj = NULL;
1020                 }
1021         } else
1022                 (*obj)->refcnt++;
1023         return err;
1024 }