entry.c

   1 #define USE_THE_REPOSITORY_VARIABLE
   2
   3 #include "git-compat-util.h"
   4 #include "object-store-ll.h"
   5 #include "dir.h"
   6 #include "environment.h"
   7 #include "gettext.h"
   8 #include "hex.h"
   9 #include "name-hash.h"
  10 #include "sparse-index.h"
  11 #include "streaming.h"
  12 #include "submodule.h"
  13 #include "symlinks.h"
  14 #include "progress.h"
  15 #include "fsmonitor.h"
  16 #include "entry.h"
  17 #include "parallel-checkout.h"
  18
  19 static void create_directories(const char *path, int path_len,
  20                                const struct checkout *state)
  21 {
  22         char *buf = xmallocz(path_len);
  23         int len = 0;
  24
  25         while (len < path_len) {
  26                 do {
  27                         buf[len] = path[len];
  28                         len++;
  29                 } while (len < path_len && path[len] != '/');
  30                 if (len >= path_len)
  31                         break;
  32                 buf[len] = 0;
  33
  34                 /*
  35                  * For 'checkout-index --prefix=<dir>', <dir> is
  36                  * allowed to be a symlink to an existing directory,
  37                  * and we set 'state->base_dir_len' below, such that
  38                  * we test the path components of the prefix with the
  39                  * stat() function instead of the lstat() function.
  40                  */
  41                 if (has_dirs_only_path(buf, len, state->base_dir_len))
  42                         continue; /* ok, it is already a directory. */
  43
  44                 /*
  45                  * If this mkdir() would fail, it could be that there
  46                  * is already a symlink or something else exists
  47                  * there, therefore we then try to unlink it and try
  48                  * one more time to create the directory.
  49                  */
  50                 if (mkdir(buf, 0777)) {
  51                         if (errno == EEXIST && state->force &&
  52                             !unlink_or_warn(buf) && !mkdir(buf, 0777))
  53                                 continue;
  54                         die_errno("cannot create directory at '%s'", buf);
  55                 }
  56         }
  57         free(buf);
  58 }
  59
  60 static void remove_subtree(struct strbuf *path)
  61 {
  62         DIR *dir = opendir(path->buf);
  63         struct dirent *de;
  64         int origlen = path->len;
  65
  66         if (!dir)
  67                 die_errno("cannot opendir '%s'", path->buf);
  68         while ((de = readdir_skip_dot_and_dotdot(dir)) != NULL) {
  69                 struct stat st;
  70
  71                 strbuf_addch(path, '/');
  72                 strbuf_addstr(path, de->d_name);
  73                 if (lstat(path->buf, &st))
  74                         die_errno("cannot lstat '%s'", path->buf);
  75                 if (S_ISDIR(st.st_mode))
  76                         remove_subtree(path);
  77                 else if (unlink(path->buf))
  78                         die_errno("cannot unlink '%s'", path->buf);
  79                 strbuf_setlen(path, origlen);
  80         }
  81         closedir(dir);
  82         if (rmdir(path->buf))
  83                 die_errno("cannot rmdir '%s'", path->buf);
  84 }
  85
  86 static int create_file(const char *path, unsigned int mode)
  87 {
  88         mode = (mode & 0100) ? 0777 : 0666;
  89         return open(path, O_WRONLY | O_CREAT | O_EXCL, mode);
  90 }
  91
  92 void *read_blob_entry(const struct cache_entry *ce, size_t *size)
  93 {
  94         enum object_type type;
  95         unsigned long ul;
  96         void *blob_data = repo_read_object_file(the_repository, &ce->oid,
  97                                                 &type, &ul);
  98
  99         *size = ul;
 100         if (blob_data) {
 101                 if (type == OBJ_BLOB)
 102                         return blob_data;
 103                 free(blob_data);
 104         }
 105         return NULL;
 106 }
 107
 108 static int open_output_fd(char *path, const struct cache_entry *ce, int to_tempfile)
 109 {
 110         int symlink = (ce->ce_mode & S_IFMT) != S_IFREG;
 111         if (to_tempfile) {
 112                 xsnprintf(path, TEMPORARY_FILENAME_LENGTH, "%s",
 113                           symlink ? ".merge_link_XXXXXX" : ".merge_file_XXXXXX");
 114                 return mkstemp(path);
 115         } else {
 116                 return create_file(path, !symlink ? ce->ce_mode : 0666);
 117         }
 118 }
 119
 120 int fstat_checkout_output(int fd, const struct checkout *state, struct stat *st)
 121 {
 122         /* use fstat() only when path == ce->name */
 123         if (fstat_is_reliable() &&
 124             state->refresh_cache && !state->base_dir_len) {
 125                 return !fstat(fd, st);
 126         }
 127         return 0;
 128 }
 129
 130 static int streaming_write_entry(const struct cache_entry *ce, char *path,
 131                                  struct stream_filter *filter,
 132                                  const struct checkout *state, int to_tempfile,
 133                                  int *fstat_done, struct stat *statbuf)
 134 {
 135         int result = 0;
 136         int fd;
 137
 138         fd = open_output_fd(path, ce, to_tempfile);
 139         if (fd < 0)
 140                 return -1;
 141
 142         result |= stream_blob_to_fd(fd, &ce->oid, filter, 1);
 143         *fstat_done = fstat_checkout_output(fd, state, statbuf);
 144         result |= close(fd);
 145
 146         if (result)
 147                 unlink(path);
 148         return result;
 149 }
 150
 151 void enable_delayed_checkout(struct checkout *state)
 152 {
 153         if (!state->delayed_checkout) {
 154                 state->delayed_checkout = xmalloc(sizeof(*state->delayed_checkout));
 155                 state->delayed_checkout->state = CE_CAN_DELAY;
 156                 string_list_init_nodup(&state->delayed_checkout->filters);
 157                 string_list_init_nodup(&state->delayed_checkout->paths);
 158         }
 159 }
 160
 161 static int remove_available_paths(struct string_list_item *item, void *cb_data)
 162 {
 163         struct string_list *available_paths = cb_data;
 164         struct string_list_item *available;
 165
 166         available = string_list_lookup(available_paths, item->string);
 167         if (available)
 168                 available->util = item->util;
 169         return !available;
 170 }
 171
 172 static int string_is_not_null(struct string_list_item *item, void *data UNUSED)
 173 {
 174         return !!item->string;
 175 }
 176
 177 int finish_delayed_checkout(struct checkout *state, int show_progress)
 178 {
 179         int errs = 0;
 180         unsigned processed_paths = 0;
 181         off_t filtered_bytes = 0;
 182         struct string_list_item *filter, *path;
 183         struct progress *progress = NULL;
 184         struct delayed_checkout *dco = state->delayed_checkout;
 185
 186         if (!state->delayed_checkout)
 187                 return errs;
 188
 189         dco->state = CE_RETRY;
 190         if (show_progress)
 191                 progress = start_delayed_progress(_("Filtering content"), dco->paths.nr);
 192         while (dco->filters.nr > 0) {
 193                 for_each_string_list_item(filter, &dco->filters) {
 194                         struct string_list available_paths = STRING_LIST_INIT_NODUP;
 195
 196                         if (!async_query_available_blobs(filter->string, &available_paths)) {
 197                                 /* Filter reported an error */
 198                                 errs = 1;
 199                                 filter->string = NULL;
 200                                 continue;
 201                         }
 202                         if (available_paths.nr <= 0) {
 203                                 /*
 204                                  * Filter responded with no entries. That means
 205                                  * the filter is done and we can remove the
 206                                  * filter from the list (see
 207                                  * "string_list_remove_empty_items" call below).
 208                                  */
 209                                 filter->string = NULL;
 210                                 continue;
 211                         }
 212
 213                         /*
 214                          * In dco->paths we store a list of all delayed paths.
 215                          * The filter just send us a list of available paths.
 216                          * Remove them from the list.
 217                          */
 218                         filter_string_list(&dco->paths, 0,
 219                                 &remove_available_paths, &available_paths);
 220
 221                         for_each_string_list_item(path, &available_paths) {
 222                                 struct cache_entry* ce;
 223
 224                                 if (!path->util) {
 225                                         error("external filter '%s' signaled that '%s' "
 226                                               "is now available although it has not been "
 227                                               "delayed earlier",
 228                                               filter->string, path->string);
 229                                         errs |= 1;
 230
 231                                         /*
 232                                          * Do not ask the filter for available blobs,
 233                                          * again, as the filter is likely buggy.
 234                                          */
 235                                         filter->string = NULL;
 236                                         continue;
 237                                 }
 238                                 ce = index_file_exists(state->istate, path->string,
 239                                                        strlen(path->string), 0);
 240                                 if (ce) {
 241                                         display_progress(progress, ++processed_paths);
 242                                         errs |= checkout_entry(ce, state, NULL, path->util);
 243                                         filtered_bytes += ce->ce_stat_data.sd_size;
 244                                         display_throughput(progress, filtered_bytes);
 245                                 } else
 246                                         errs = 1;
 247                         }
 248                 }
 249
 250                 filter_string_list(&dco->filters, 0, string_is_not_null, NULL);
 251         }
 252         stop_progress(&progress);
 253         string_list_clear(&dco->filters, 0);
 254
 255         /* At this point we should not have any delayed paths anymore. */
 256         errs |= dco->paths.nr;
 257         for_each_string_list_item(path, &dco->paths) {
 258                 error("'%s' was not filtered properly", path->string);
 259         }
 260         string_list_clear(&dco->paths, 0);
 261
 262         free(dco);
 263         state->delayed_checkout = NULL;
 264
 265         return errs;
 266 }
 267
 268 void update_ce_after_write(const struct checkout *state, struct cache_entry *ce,
 269                            struct stat *st)
 270 {
 271         if (state->refresh_cache) {
 272                 assert(state->istate);
 273                 fill_stat_cache_info(state->istate, ce, st);
 274                 ce->ce_flags |= CE_UPDATE_IN_BASE;
 275                 mark_fsmonitor_invalid(state->istate, ce);
 276                 state->istate->cache_changed |= CE_ENTRY_CHANGED;
 277         }
 278 }
 279
 280 /* Note: ca is used (and required) iff the entry refers to a regular file. */
 281 static int write_entry(struct cache_entry *ce, char *path, struct conv_attrs *ca,
 282                        const struct checkout *state, int to_tempfile,
 283                        int *nr_checkouts)
 284 {
 285         unsigned int ce_mode_s_ifmt = ce->ce_mode & S_IFMT;
 286         struct delayed_checkout *dco = state->delayed_checkout;
 287         int fd, ret, fstat_done = 0;
 288         char *new_blob;
 289         struct strbuf buf = STRBUF_INIT;
 290         size_t size;
 291         ssize_t wrote;
 292         size_t newsize = 0;
 293         struct stat st;
 294         const struct submodule *sub;
 295         struct checkout_metadata meta;
 296         static int scratch_nr_checkouts;
 297
 298         clone_checkout_metadata(&meta, &state->meta, &ce->oid);
 299
 300         if (ce_mode_s_ifmt == S_IFREG) {
 301                 struct stream_filter *filter = get_stream_filter_ca(ca, &ce->oid);
 302                 if (filter &&
 303                     !streaming_write_entry(ce, path, filter,
 304                                            state, to_tempfile,
 305                                            &fstat_done, &st))
 306                         goto finish;
 307         }
 308
 309         switch (ce_mode_s_ifmt) {
 310         case S_IFLNK:
 311                 new_blob = read_blob_entry(ce, &size);
 312                 if (!new_blob)
 313                         return error("unable to read sha1 file of %s (%s)",
 314                                      ce->name, oid_to_hex(&ce->oid));
 315
 316                 /*
 317                  * We can't make a real symlink; write out a regular file entry
 318                  * with the symlink destination as its contents.
 319                  */
 320                 if (!has_symlinks || to_tempfile)
 321                         goto write_file_entry;
 322
 323                 ret = symlink(new_blob, path);
 324                 free(new_blob);
 325                 if (ret)
 326                         return error_errno("unable to create symlink %s", path);
 327                 break;
 328
 329         case S_IFREG:
 330                 /*
 331                  * We do not send the blob in case of a retry, so do not
 332                  * bother reading it at all.
 333                  */
 334                 if (dco && dco->state == CE_RETRY) {
 335                         new_blob = NULL;
 336                         size = 0;
 337                 } else {
 338                         new_blob = read_blob_entry(ce, &size);
 339                         if (!new_blob)
 340                                 return error("unable to read sha1 file of %s (%s)",
 341                                              ce->name, oid_to_hex(&ce->oid));
 342                 }
 343
 344                 /*
 345                  * Convert from git internal format to working tree format
 346                  */
 347                 if (dco && dco->state != CE_NO_DELAY) {
 348                         ret = async_convert_to_working_tree_ca(ca, ce->name,
 349                                                                new_blob, size,
 350                                                                &buf, &meta, dco);
 351                         if (ret) {
 352                                 struct string_list_item *item =
 353                                         string_list_lookup(&dco->paths, ce->name);
 354                                 if (item) {
 355                                         item->util = nr_checkouts ? nr_checkouts
 356                                                         : &scratch_nr_checkouts;
 357                                         free(new_blob);
 358                                         goto delayed;
 359                                 }
 360                         }
 361                 } else {
 362                         ret = convert_to_working_tree_ca(ca, ce->name, new_blob,
 363                                                          size, &buf, &meta);
 364                 }
 365
 366                 if (ret) {
 367                         free(new_blob);
 368                         new_blob = strbuf_detach(&buf, &newsize);
 369                         size = newsize;
 370                 }
 371                 /*
 372                  * No "else" here as errors from convert are OK at this
 373                  * point. If the error would have been fatal (e.g.
 374                  * filter is required), then we would have died already.
 375                  */
 376
 377         write_file_entry:
 378                 fd = open_output_fd(path, ce, to_tempfile);
 379                 if (fd < 0) {
 380                         free(new_blob);
 381                         return error_errno("unable to create file %s", path);
 382                 }
 383
 384                 wrote = write_in_full(fd, new_blob, size);
 385                 if (!to_tempfile)
 386                         fstat_done = fstat_checkout_output(fd, state, &st);
 387                 close(fd);
 388                 free(new_blob);
 389                 if (wrote < 0)
 390                         return error("unable to write file %s", path);
 391                 break;
 392
 393         case S_IFGITLINK:
 394                 if (to_tempfile)
 395                         return error("cannot create temporary submodule %s", ce->name);
 396                 if (mkdir(path, 0777) < 0)
 397                         return error("cannot create submodule directory %s", path);
 398                 sub = submodule_from_ce(ce);
 399                 if (sub)
 400                         return submodule_move_head(ce->name, state->super_prefix,
 401                                 NULL, oid_to_hex(&ce->oid),
 402                                 state->force ? SUBMODULE_MOVE_HEAD_FORCE : 0);
 403                 break;
 404
 405         default:
 406                 return error("unknown file mode for %s in index", ce->name);
 407         }
 408
 409 finish:
 410         if (state->refresh_cache) {
 411                 if (!fstat_done && lstat(ce->name, &st) < 0)
 412                         return error_errno("unable to stat just-written file %s",
 413                                            ce->name);
 414                 update_ce_after_write(state, ce , &st);
 415         }
 416         if (nr_checkouts)
 417                 (*nr_checkouts)++;
 418 delayed:
 419         return 0;
 420 }
 421
 422 /*
 423  * This is like 'lstat()', except it refuses to follow symlinks
 424  * in the path, after skipping "skiplen".
 425  */
 426 static int check_path(const char *path, int len, struct stat *st, int skiplen)
 427 {
 428         const char *slash = path + len;
 429
 430         while (path < slash && *slash != '/')
 431                 slash--;
 432         if (!has_dirs_only_path(path, slash - path, skiplen)) {
 433                 errno = ENOENT;
 434                 return -1;
 435         }
 436         return lstat(path, st);
 437 }
 438
 439 static void mark_colliding_entries(const struct checkout *state,
 440                                    struct cache_entry *ce, struct stat *st)
 441 {
 442         int i, trust_ino = check_stat;
 443
 444 #if defined(GIT_WINDOWS_NATIVE) || defined(__CYGWIN__)
 445         trust_ino = 0;
 446 #endif
 447
 448         ce->ce_flags |= CE_MATCHED;
 449
 450         /* TODO: audit for interaction with sparse-index. */
 451         ensure_full_index(state->istate);
 452         for (i = 0; i < state->istate->cache_nr; i++) {
 453                 struct cache_entry *dup = state->istate->cache[i];
 454
 455                 if (dup == ce) {
 456                         /*
 457                          * Parallel checkout doesn't create the files in index
 458                          * order. So the other side of the collision may appear
 459                          * after the given cache_entry in the array.
 460                          */
 461                         if (parallel_checkout_status() == PC_RUNNING)
 462                                 continue;
 463                         else
 464                                 break;
 465                 }
 466
 467                 if (dup->ce_flags & (CE_MATCHED | CE_VALID | CE_SKIP_WORKTREE))
 468                         continue;
 469
 470                 if ((trust_ino && !match_stat_data(&dup->ce_stat_data, st)) ||
 471                     paths_collide(ce->name, dup->name)) {
 472                         dup->ce_flags |= CE_MATCHED;
 473                         break;
 474                 }
 475         }
 476 }
 477
 478 int checkout_entry_ca(struct cache_entry *ce, struct conv_attrs *ca,
 479                       const struct checkout *state, char *topath,
 480                       int *nr_checkouts)
 481 {
 482         static struct strbuf path = STRBUF_INIT;
 483         struct stat st;
 484         struct conv_attrs ca_buf;
 485
 486         if (ce->ce_flags & CE_WT_REMOVE) {
 487                 if (topath)
 488                         /*
 489                          * No content and thus no path to create, so we have
 490                          * no pathname to return.
 491                          */
 492                         BUG("Can't remove entry to a path");
 493                 unlink_entry(ce, state->super_prefix);
 494                 return 0;
 495         }
 496
 497         if (topath) {
 498                 if (S_ISREG(ce->ce_mode) && !ca) {
 499                         convert_attrs(state->istate, &ca_buf, ce->name);
 500                         ca = &ca_buf;
 501                 }
 502                 return write_entry(ce, topath, ca, state, 1, nr_checkouts);
 503         }
 504
 505         strbuf_reset(&path);
 506         strbuf_add(&path, state->base_dir, state->base_dir_len);
 507         strbuf_add(&path, ce->name, ce_namelen(ce));
 508
 509         if (!check_path(path.buf, path.len, &st, state->base_dir_len)) {
 510                 const struct submodule *sub;
 511                 unsigned changed = ie_match_stat(state->istate, ce, &st,
 512                                                  CE_MATCH_IGNORE_VALID | CE_MATCH_IGNORE_SKIP_WORKTREE);
 513                 /*
 514                  * Needs to be checked before !changed returns early,
 515                  * as the possibly empty directory was not changed
 516                  */
 517                 sub = submodule_from_ce(ce);
 518                 if (sub) {
 519                         int err;
 520                         if (!is_submodule_populated_gently(ce->name, &err)) {
 521                                 struct stat sb;
 522                                 if (lstat(ce->name, &sb))
 523                                         die(_("could not stat file '%s'"), ce->name);
 524                                 if (!(st.st_mode & S_IFDIR))
 525                                         unlink_or_warn(ce->name);
 526
 527                                 return submodule_move_head(ce->name, state->super_prefix,
 528                                         NULL, oid_to_hex(&ce->oid), 0);
 529                         } else
 530                                 return submodule_move_head(ce->name, state->super_prefix,
 531                                         "HEAD", oid_to_hex(&ce->oid),
 532                                         state->force ? SUBMODULE_MOVE_HEAD_FORCE : 0);
 533                 }
 534
 535                 if (!changed)
 536                         return 0;
 537                 if (!state->force) {
 538                         if (!state->quiet)
 539                                 fprintf(stderr,
 540                                         "%s already exists, no checkout\n",
 541                                         path.buf);
 542                         return -1;
 543                 }
 544
 545                 if (state->clone)
 546                         mark_colliding_entries(state, ce, &st);
 547
 548                 /*
 549                  * We unlink the old file, to get the new one with the
 550                  * right permissions (including umask, which is nasty
 551                  * to emulate by hand - much easier to let the system
 552                  * just do the right thing)
 553                  */
 554                 if (S_ISDIR(st.st_mode)) {
 555                         /* If it is a gitlink, leave it alone! */
 556                         if (S_ISGITLINK(ce->ce_mode))
 557                                 return 0;
 558                         /*
 559                          * We must avoid replacing submodules' leading
 560                          * directories with symbolic links, lest recursive
 561                          * clones can write into arbitrary locations.
 562                          *
 563                          * Technically, this logic is not limited
 564                          * to recursive clones, or for that matter to
 565                          * submodules' paths colliding with symbolic links'
 566                          * paths. Yet it strikes a balance in favor of
 567                          * simplicity, and if paths are colliding, we might
 568                          * just as well keep the directories during a clone.
 569                          */
 570                         if (state->clone && S_ISLNK(ce->ce_mode))
 571                                 return 0;
 572                         remove_subtree(&path);
 573                 } else if (unlink(path.buf))
 574                         return error_errno("unable to unlink old '%s'", path.buf);
 575         } else if (state->not_new)
 576                 return 0;
 577
 578         create_directories(path.buf, path.len, state);
 579
 580         if (S_ISREG(ce->ce_mode) && !ca) {
 581                 convert_attrs(state->istate, &ca_buf, ce->name);
 582                 ca = &ca_buf;
 583         }
 584
 585         if (!enqueue_checkout(ce, ca, nr_checkouts))
 586                 return 0;
 587
 588         return write_entry(ce, path.buf, ca, state, 0, nr_checkouts);
 589 }
 590
 591 void unlink_entry(const struct cache_entry *ce, const char *super_prefix)
 592 {
 593         const struct submodule *sub = submodule_from_ce(ce);
 594         if (sub) {
 595                 /* state.force is set at the caller. */
 596                 submodule_move_head(ce->name, super_prefix, "HEAD", NULL,
 597                                     SUBMODULE_MOVE_HEAD_FORCE);
 598         }
 599         if (check_leading_path(ce->name, ce_namelen(ce), 1) >= 0)
 600                 return;
 601         if (remove_or_warn(ce->ce_mode, ce->name))
 602                 return;
 603         schedule_dir_for_removal(ce->name, ce_namelen(ce));
 604 }
 605
 606 int remove_or_warn(unsigned int mode, const char *file)
 607 {
 608         return S_ISGITLINK(mode) ? rmdir_or_warn(file) : unlink_or_warn(file);
 609 }