link-by-hash.diff

   1 Jason M. Felice wrote:
   2
   3 This patch adds the --link-by-hash=DIR option, which hard links received
   4 files in a link farm arranged by MD4 file hash.  The result is that the system
   5 will only store one copy of the unique contents of each file, regardless of
   6 the file's name.
   7
   8 To use this patch, run these commands for a successful build:
   9
  10     patch -p1 <patches/link-by-hash.diff
  11     ./prepare-source
  12     ./configure
  13     make
  14
  15 diff --git a/Makefile.in b/Makefile.in
  16 --- a/Makefile.in
  17 +++ b/Makefile.in
  18 @@ -36,7 +36,7 @@ OBJS1=flist.o rsync.o generator.o receiver.o cleanup.o sender.o exclude.o \
  19         util.o main.o checksum.o match.o syscall.o log.o backup.o
  20  OBJS2=options.o io.o compat.o hlink.o token.o uidlist.o socket.o hashtable.o \
  21         fileio.o batch.o clientname.o chmod.o acls.o xattrs.o
  22 -OBJS3=progress.o pipe.o
  23 +OBJS3=progress.o pipe.o hashlink.o
  24  DAEMON_OBJ = params.o loadparm.o clientserver.o access.o connection.o authenticate.o
  25  popt_OBJS=popt/findme.o  popt/popt.o  popt/poptconfig.o \
  26         popt/popthelp.o popt/poptparse.o
  27 diff --git a/flist.c b/flist.c
  28 --- a/flist.c
  29 +++ b/flist.c
  30 @@ -68,6 +68,7 @@ extern int need_unsorted_flist;
  31  extern int unsort_ndx;
  32  extern struct stats stats;
  33  extern char *filesfrom_host;
  34 +extern char *link_by_hash_dir;
  35
  36  extern char curr_dir[MAXPATHLEN];
  37
  38 @@ -822,7 +823,7 @@ static struct file_struct *recv_file_entry(struct file_list *flist,
  39                 extra_len += (S_ISDIR(mode) ? 2 : 1) * EXTRA_LEN;
  40  #endif
  41
  42 -       if (always_checksum && S_ISREG(mode))
  43 +       if ((always_checksum || link_by_hash_dir) && S_ISREG(mode))
  44                 extra_len += SUM_EXTRA_CNT * EXTRA_LEN;
  45
  46         if (file_length > 0xFFFFFFFFu && S_ISREG(mode))
  47 diff --git a/hashlink.c b/hashlink.c
  48 new file mode 100644
  49 --- /dev/null
  50 +++ b/hashlink.c
  51 @@ -0,0 +1,336 @@
  52 +/*
  53 +   Copyright (C) Cronosys, LLC 2004
  54 +
  55 +   This program is free software; you can redistribute it and/or modify
  56 +   it under the terms of the GNU General Public License as published by
  57 +   the Free Software Foundation; either version 2 of the License, or
  58 +   (at your option) any later version.
  59 +
  60 +   This program is distributed in the hope that it will be useful,
  61 +   but WITHOUT ANY WARRANTY; without even the implied warranty of
  62 +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  63 +   GNU General Public License for more details.
  64 +
  65 +   You should have received a copy of the GNU General Public License
  66 +   along with this program; if not, write to the Free Software
  67 +   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  68 +*/
  69 +
  70 +/* This file contains code used by the --link-by-hash option. */
  71 +
  72 +#include "rsync.h"
  73 +
  74 +extern char *link_by_hash_dir;
  75 +
  76 +#ifdef HAVE_LINK
  77 +
  78 +char *make_hash_name(struct file_struct *file)
  79 +{
  80 +       char hash[33], *dst;
  81 +       uchar c, *src = (uchar*)F_SUM(file);
  82 +       int i;
  83 +
  84 +       for (dst = hash, i = 0; i < 4; i++, src++) {
  85 +               c = *src >> 4;
  86 +               *(dst++) = (c >= 10) ? (c - 10 + 'a') : (c + '0');
  87 +               c = *src & 0x0f;
  88 +               *(dst++) = (c >= 10) ? (c - 10 + 'a') : (c + '0');
  89 +       }
  90 +       *dst++ = '/';
  91 +       for (i = 0; i < 12; i++, src++) {
  92 +               c = *src >> 4;
  93 +               *(dst++) = (c >= 10) ? (c - 10 + 'a') : (c + '0');
  94 +               c = *src & 0x0f;
  95 +               *(dst++) = (c >= 10) ? (c - 10 + 'a') : (c + '0');
  96 +       }
  97 +       *dst = 0;
  98 +
  99 +       asprintf(&dst,"%s/%s",link_by_hash_dir,hash);
 100 +       return dst;
 101 +}
 102 +
 103 +
 104 +void kill_hashfile(struct hashfile_struct *hashfile)
 105 +{
 106 +       if (!hashfile)
 107 +               return;
 108 +       free(hashfile->name);
 109 +       close(hashfile->fd);
 110 +       free(hashfile);
 111 +}
 112 +
 113 +
 114 +void kill_hashfiles(struct hashfile_struct *hashfiles)
 115 +{
 116 +       struct hashfile_struct *iter, *next;
 117 +       if ((iter = hashfiles) != NULL) {
 118 +               do {
 119 +                       next = iter->next;
 120 +                       kill_hashfile(iter);
 121 +                       iter = next;
 122 +               } while (iter != hashfiles);
 123 +       }
 124 +}
 125 +
 126 +
 127 +struct hashfile_struct *find_hashfiles(char *hashname, int64 size, long *fnbr)
 128 +{
 129 +       DIR *d;
 130 +       struct dirent *di;
 131 +       struct hashfile_struct *hashfiles = NULL, *hashfile;
 132 +       STRUCT_STAT st;
 133 +       long this_fnbr;
 134 +
 135 +       *fnbr = 0;
 136 +
 137 +       /* Build a list of potential candidates and open
 138 +        * them. */
 139 +       if ((d = opendir(hashname)) == NULL) {
 140 +               rsyserr(FERROR, errno, "opendir failed: \"%s\"", hashname);
 141 +               free(hashname);
 142 +               return NULL;
 143 +       }
 144 +       while ((di = readdir(d)) != NULL) {
 145 +               if (!strcmp(di->d_name,".") || !strcmp(di->d_name,"..")) {
 146 +                       continue;
 147 +               }
 148 +
 149 +               /* We need to have the largest fnbr in case we need to store
 150 +                * a new file. */
 151 +               this_fnbr = atol(di->d_name);
 152 +               if (this_fnbr > *fnbr)
 153 +                       *fnbr = this_fnbr;
 154 +
 155 +               hashfile = new_array(struct hashfile_struct, 1);
 156 +               asprintf(&hashfile->name,"%s/%s",hashname,
 157 +                        di->d_name);
 158 +               if (do_stat(hashfile->name,&st) == -1) {
 159 +                       rsyserr(FERROR, errno, "stat failed: %s", hashfile->name);
 160 +                       kill_hashfile(hashfile);
 161 +                       continue;
 162 +               }
 163 +               if (st.st_size != size) {
 164 +                       kill_hashfile(hashfile);
 165 +                       continue;
 166 +               }
 167 +               hashfile->nlink = st.st_nlink;
 168 +               hashfile->fd = open(hashfile->name,O_RDONLY|O_BINARY);
 169 +               if (hashfile->fd == -1) {
 170 +                       rsyserr(FERROR, errno, "open failed: %s", hashfile->name);
 171 +                       kill_hashfile(hashfile);
 172 +                       continue;
 173 +               }
 174 +               if (hashfiles == NULL)
 175 +                       hashfiles = hashfile->next = hashfile->prev = hashfile;
 176 +               else {
 177 +                       hashfile->next = hashfiles;
 178 +                       hashfile->prev = hashfiles->prev;
 179 +                       hashfile->next->prev = hashfile;
 180 +                       hashfile->prev->next = hashfile;
 181 +               }
 182 +       }
 183 +       closedir(d);
 184 +
 185 +       return hashfiles;
 186 +}
 187 +
 188 +
 189 +struct hashfile_struct *compare_hashfiles(int fd,struct hashfile_struct *files)
 190 +{
 191 +       int amt, hamt;
 192 +       char buffer[BUFSIZ], cmpbuffer[BUFSIZ];
 193 +       struct hashfile_struct *iter, *next, *best;
 194 +       uint32 nlink;
 195 +
 196 +       if (!files)
 197 +               return NULL;
 198 +
 199 +       iter = files; /* in case files are 0 bytes */
 200 +       while ((amt = read(fd, buffer, BUFSIZ)) > 0) {
 201 +               iter = files;
 202 +               do {
 203 +                       /* Icky bit to resync when we steal the first node. */
 204 +                       if (!files)
 205 +                               files = iter;
 206 +
 207 +                       next = iter->next;
 208 +
 209 +                       hamt = read(iter->fd, cmpbuffer, BUFSIZ);
 210 +                       if (amt != hamt || memcmp(buffer, cmpbuffer, amt)) {
 211 +                               if (iter == files) {
 212 +                                       files = files->prev;
 213 +                               }
 214 +                               if (iter->next == iter) {
 215 +                                       files = next = NULL;
 216 +                               } else {
 217 +                                       next = iter->next;
 218 +                                       if (iter == files) {
 219 +                                               /* So we know to resync */
 220 +                                               files = NULL;
 221 +                                       }
 222 +                               }
 223 +                               iter->next->prev = iter->prev;
 224 +                               iter->prev->next = iter->next;
 225 +                               kill_hashfile(iter);
 226 +                       }
 227 +
 228 +                       iter = next;
 229 +               } while (iter != files);
 230 +
 231 +               if (iter == NULL && files == NULL) {
 232 +                       /* There are no matches. */
 233 +                       return NULL;
 234 +               }
 235 +       }
 236 +
 237 +       if (amt == -1) {
 238 +               rsyserr(FERROR, errno, "read failed in compare_hashfiles()");
 239 +               kill_hashfiles(files);
 240 +               return NULL;
 241 +       }
 242 +
 243 +       /* If we only have one file left, use it. */
 244 +       if (files == files->next) {
 245 +               return files;
 246 +       }
 247 +
 248 +       /* All files which remain in the list are identical and should have
 249 +        * the same size.  We pick the one with the lowest link count (we
 250 +        * may have rolled over because we hit the maximum link count for
 251 +        * the filesystem). */
 252 +       best = iter = files;
 253 +       nlink = iter->nlink;
 254 +       do {
 255 +               if (iter->nlink < nlink) {
 256 +                       nlink = iter->nlink;
 257 +                       best = iter;
 258 +               }
 259 +               iter = iter->next;
 260 +       } while (iter != files);
 261 +
 262 +       best->next->prev = best->prev;
 263 +       best->prev->next = best->next;
 264 +       if (files == best)
 265 +               files = files->next;
 266 +       kill_hashfiles(files);
 267 +       return best;
 268 +}
 269 +
 270 +
 271 +int link_by_hash(const char *fnametmp, const char *fname, struct file_struct *file)
 272 +{
 273 +       STRUCT_STAT st;
 274 +       char *hashname = make_hash_name(file);
 275 +       int first = 0, rc;
 276 +       char *linkname;
 277 +       long last_fnbr;
 278 +
 279 +       if (F_LENGTH(file) == 0)
 280 +               return robust_rename(fnametmp, fname, NULL, 0644);
 281 +
 282 +       if (do_stat(hashname, &st) == -1) {
 283 +               char *dirname;
 284 +
 285 +               /* Directory does not exist. */
 286 +               dirname = strdup(hashname);
 287 +               *strrchr(dirname,'/') = 0;
 288 +               if (do_mkdir(dirname, 0755) == -1 && errno != EEXIST) {
 289 +                       rsyserr(FERROR, errno, "mkdir failed: %s", dirname);
 290 +                       free(hashname);
 291 +                       free(dirname);
 292 +                       return robust_rename(fnametmp, fname, NULL, 0644);
 293 +               }
 294 +               free(dirname);
 295 +
 296 +               if (do_mkdir(hashname, 0755) == -1 && errno != EEXIST) {
 297 +                       rsyserr(FERROR, errno, "mkdir failed: %s", hashname);
 298 +                       free(hashname);
 299 +                       return robust_rename(fnametmp, fname, NULL, 0644);
 300 +               }
 301 +
 302 +               first = 1;
 303 +               asprintf(&linkname,"%s/0",hashname);
 304 +               rprintf(FINFO, "(1) linkname = %s\n", linkname);
 305 +       } else {
 306 +               struct hashfile_struct *hashfiles, *hashfile;
 307 +
 308 +               if (do_stat(fnametmp,&st) == -1) {
 309 +                       rsyserr(FERROR, errno, "stat failed: %s", fname);
 310 +                       return -1;
 311 +               }
 312 +               hashfiles = find_hashfiles(hashname, st.st_size, &last_fnbr);
 313 +
 314 +               if (hashfiles == NULL) {
 315 +                       first = 1;
 316 +                       asprintf(&linkname,"%s/0",hashname);
 317 +                       rprintf(FINFO, "(2) linkname = %s\n", linkname);
 318 +               } else {
 319 +                       int fd;
 320 +                       /* Search for one identical to us. */
 321 +                       if ((fd = open(fnametmp,O_RDONLY|O_BINARY)) == -1) {
 322 +                               rsyserr(FERROR, errno, "open failed: %s", fnametmp);
 323 +                               kill_hashfiles(hashfiles);
 324 +                               return -1;
 325 +                       }
 326 +                       hashfile = compare_hashfiles(fd, hashfiles);
 327 +                       hashfiles = NULL;
 328 +                       close(fd);
 329 +
 330 +                       if (hashfile) {
 331 +                               first = 0;
 332 +                               linkname = strdup(hashfile->name);
 333 +                               rprintf(FINFO, "(3) linkname = %s\n", linkname);
 334 +                               kill_hashfile(hashfile);
 335 +                       } else {
 336 +                               first = 1;
 337 +                               asprintf(&linkname, "%s/%ld", hashname,
 338 +                                        last_fnbr + 1);
 339 +                               rprintf(FINFO, "(4) linkname = %s\n", linkname);
 340 +                       }
 341 +               }
 342 +       }
 343 +
 344 +       if (!first) {
 345 +               rprintf(FINFO, "link-by-hash (existing): \"%s\" -> %s\n",
 346 +                               linkname, full_fname(fname));
 347 +               robust_unlink(fname);
 348 +               rc = do_link(linkname, fname);
 349 +               if (rc == -1) {
 350 +                       if (errno == EMLINK) {
 351 +                               first = 1;
 352 +                               free(linkname);
 353 +                               asprintf(&linkname,"%s/%ld",hashname,
 354 +                                        last_fnbr + 1);
 355 +                               rprintf(FINFO, "(5) linkname = %s\n", linkname);
 356 +                               rprintf(FINFO,"link-by-hash: max link count exceeded, starting new file \"%s\".\n", linkname);
 357 +                       } else {
 358 +                               rsyserr(FERROR, errno, "link \"%s\" -> \"%s\"",
 359 +                                       linkname, full_fname(fname));
 360 +                               rc = robust_rename(fnametmp, fname, NULL, 0644);
 361 +                       }
 362 +               } else {
 363 +                       do_unlink(fnametmp);
 364 +               }
 365 +       }
 366 +
 367 +       if (first) {
 368 +               rprintf(FINFO, "link-by-hash (new): %s -> \"%s\"\n",
 369 +                               full_fname(fname),linkname);
 370 +
 371 +               rc = robust_rename(fnametmp, fname, NULL, 0644);
 372 +               if (rc != 0) {
 373 +                       rsyserr(FERROR, errno, "rename \"%s\" -> \"%s\"",
 374 +                               full_fname(fnametmp), full_fname(fname));
 375 +               }
 376 +               rc = do_link(fname,linkname);
 377 +               if (rc != 0) {
 378 +                       rsyserr(FERROR, errno, "link \"%s\" -> \"%s\"",
 379 +                               full_fname(fname), linkname);
 380 +               }
 381 +       }
 382 +
 383 +       free(linkname);
 384 +       free(hashname);
 385 +       return rc;
 386 +}
 387 +#endif
 388 diff --git a/options.c b/options.c
 389 --- a/options.c
 390 +++ b/options.c
 391 @@ -156,6 +156,7 @@ char *backup_suffix = NULL;
 392  char *tmpdir = NULL;
 393  char *partial_dir = NULL;
 394  char *basis_dir[MAX_BASIS_DIRS+1];
 395 +char *link_by_hash_dir = NULL;
 396  char *config_file = NULL;
 397  char *shell_cmd = NULL;
 398  char *logfile_name = NULL;
 399 @@ -389,6 +390,7 @@ void usage(enum logcode F)
 400    rprintf(F,"     --compare-dest=DIR      also compare destination files relative to DIR\n");
 401    rprintf(F,"     --copy-dest=DIR         ... and include copies of unchanged files\n");
 402    rprintf(F,"     --link-dest=DIR         hardlink to files in DIR when unchanged\n");
 403 +  rprintf(F,"     --link-by-hash=DIR      create hardlinks by hash into DIR\n");
 404    rprintf(F," -z, --compress              compress file data during the transfer\n");
 405    rprintf(F,"     --compress-level=NUM    explicitly set compression level\n");
 406    rprintf(F,"     --skip-compress=LIST    skip compressing files with a suffix in LIST\n");
 407 @@ -441,7 +443,7 @@ enum {OPT_VERSION = 1000, OPT_DAEMON, OPT_SENDER, OPT_EXCLUDE, OPT_EXCLUDE_FROM,
 408        OPT_FILTER, OPT_COMPARE_DEST, OPT_COPY_DEST, OPT_LINK_DEST, OPT_HELP,
 409        OPT_INCLUDE, OPT_INCLUDE_FROM, OPT_MODIFY_WINDOW, OPT_MIN_SIZE, OPT_CHMOD,
 410        OPT_READ_BATCH, OPT_WRITE_BATCH, OPT_ONLY_WRITE_BATCH, OPT_MAX_SIZE,
 411 -      OPT_NO_D, OPT_APPEND,
 412 +      OPT_NO_D, OPT_APPEND, OPT_LINK_BY_HASH,
 413        OPT_SERVER, OPT_REFUSED_BASE = 9000};
 414
 415  static struct poptOption long_options[] = {
 416 @@ -564,6 +566,7 @@ static struct poptOption long_options[] = {
 417    {"compare-dest",     0,  POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 },
 418    {"copy-dest",        0,  POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 },
 419    {"link-dest",        0,  POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 },
 420 +  {"link-by-hash",     0,  POPT_ARG_STRING, 0, OPT_LINK_BY_HASH, 0, 0},
 421    {"fuzzy",           'y', POPT_ARG_NONE,   &fuzzy_basis, 0, 0, 0 },
 422    {"compress",        'z', POPT_ARG_NONE,   0, 'z', 0, 0 },
 423    {"no-compress",      0,  POPT_ARG_VAL,    &do_compression, 0, 0, 0 },
 424 @@ -1226,6 +1229,21 @@ int parse_arguments(int *argc_p, const char ***argv_p, int frommain)
 425                         return 0;
 426  #endif
 427
 428 +                case OPT_LINK_BY_HASH:
 429 +#ifdef HAVE_LINK
 430 +                       arg = poptGetOptArg(pc);
 431 +                       if (sanitize_paths)
 432 +                               arg = sanitize_path(NULL, arg, NULL, 0, NULL);
 433 +                       link_by_hash_dir = (char *)arg;
 434 +                       break;
 435 +#else
 436 +                       snprintf(err_buf, sizeof err_buf,
 437 +                                "hard links are not supported on this %s\n",
 438 +                                am_server ? "server" : "client");
 439 +                       rprintf(FERROR, "ERROR: %s", err_buf);
 440 +                       return 0;
 441 +#endif
 442 +
 443                 default:
 444                         /* A large opt value means that set_refuse_options()
 445                          * turned this option off. */
 446 @@ -1974,6 +1992,11 @@ void server_options(char **args, int *argc_p)
 447         } else if (inplace)
 448                 args[ac++] = "--inplace";
 449
 450 +       if (link_by_hash_dir && am_sender) {
 451 +               args[ac++] = "--link-by-hash";
 452 +               args[ac++] = link_by_hash_dir;
 453 +       }
 454 +
 455         if (files_from && (!am_sender || filesfrom_host)) {
 456                 if (filesfrom_host) {
 457                         args[ac++] = "--files-from";
 458 diff --git a/receiver.c b/receiver.c
 459 --- a/receiver.c
 460 +++ b/receiver.c
 461 @@ -162,12 +162,14 @@ int open_tmpfile(char *fnametmp, const char *fname, struct file_struct *file)
 462  }
 463
 464  static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r,
 465 -                       const char *fname, int fd, OFF_T total_size)
 466 +                       const char *fname, int fd, OFF_T total_size,
 467 +                       const char *md4)
 468  {
 469         static char file_sum1[MAX_DIGEST_LEN];
 470         static char file_sum2[MAX_DIGEST_LEN];
 471         struct map_struct *mapbuf;
 472         struct sum_struct sum;
 473 +       md_context mdfour_data;
 474         int32 len, sum_len;
 475         OFF_T offset = 0;
 476         OFF_T offset2;
 477 @@ -187,6 +189,9 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r,
 478         } else
 479                 mapbuf = NULL;
 480
 481 +       if (md4)
 482 +               mdfour_begin(&mdfour_data);
 483 +
 484         sum_init(checksum_seed);
 485
 486         if (append_mode > 0) {
 487 @@ -231,6 +236,8 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r,
 488                         cleanup_got_literal = 1;
 489
 490                         sum_update(data, i);
 491 +                       if (md4)
 492 +                               mdfour_update(&mdfour_data, (uchar*)data, i);
 493
 494                         if (fd != -1 && write_file(fd,data,i) != i)
 495                                 goto report_write_error;
 496 @@ -257,6 +264,8 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r,
 497
 498                         see_token(map, len);
 499                         sum_update(map, len);
 500 +                       if (md4)
 501 +                               mdfour_update(&mdfour_data, (uchar*)map, len);
 502                 }
 503
 504                 if (updating_basis_or_equiv) {
 505 @@ -299,6 +308,8 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r,
 506         }
 507
 508         sum_len = sum_end(file_sum1);
 509 +       if (md4)
 510 +               mdfour_result(&mdfour_data, (uchar*)md4);
 511
 512         if (mapbuf)
 513                 unmap_file(mapbuf);
 514 @@ -314,7 +325,7 @@ static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r,
 515
 516  static void discard_receive_data(int f_in, OFF_T length)
 517  {
 518 -       receive_data(f_in, NULL, -1, 0, NULL, -1, length);
 519 +       receive_data(f_in, NULL, -1, 0, NULL, -1, length, NULL);
 520  }
 521
 522  static void handle_delayed_updates(char *local_name)
 523 @@ -675,7 +686,7 @@ int recv_files(int f_in, char *local_name)
 524
 525                 /* recv file data */
 526                 recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size,
 527 -                                      fname, fd2, F_LENGTH(file));
 528 +                                      fname, fd2, F_LENGTH(file), F_SUM(file));
 529
 530                 log_item(log_code, file, &initial_stats, iflags, NULL);
 531
 532 diff --git a/rsync.c b/rsync.c
 533 --- a/rsync.c
 534 +++ b/rsync.c
 535 @@ -48,6 +48,7 @@ extern int inplace;
 536  extern int flist_eof;
 537  extern int keep_dirlinks;
 538  extern int make_backups;
 539 +extern char *link_by_hash_dir;
 540  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 541  extern struct chmod_mode_struct *daemon_chmod_modes;
 542  #ifdef ICONV_OPTION
 543 @@ -531,8 +532,15 @@ void finish_transfer(const char *fname, const char *fnametmp,
 544         /* move tmp file over real file */
 545         if (verbose > 2)
 546                 rprintf(FINFO, "renaming %s to %s\n", fnametmp, fname);
 547 -       ret = robust_rename(fnametmp, fname, partialptr,
 548 -                           file->mode & INITACCESSPERMS);
 549 +#ifdef HAVE_LINK
 550 +       if (link_by_hash_dir)
 551 +               ret = link_by_hash(fnametmp, fname, file);
 552 +       else
 553 +#endif
 554 +       {
 555 +               ret = robust_rename(fnametmp, fname, partialptr,
 556 +                                   file->mode & INITACCESSPERMS);
 557 +       }
 558         if (ret < 0) {
 559                 rsyserr(FERROR_XFER, errno, "%s %s -> \"%s\"",
 560                         ret == -2 ? "copy" : "rename",
 561 diff --git a/rsync.h b/rsync.h
 562 --- a/rsync.h
 563 +++ b/rsync.h
 564 @@ -816,6 +816,14 @@ struct stats {
 565         int num_transferred_files;
 566  };
 567
 568 +struct hashfile_struct {
 569 +       struct hashfile_struct *next;
 570 +       struct hashfile_struct *prev;
 571 +       char *name;
 572 +       int fd;
 573 +       uint32 nlink;
 574 +};
 575 +
 576  struct chmod_mode_struct;
 577
 578  #define EMPTY_ITEM_LIST {NULL, 0, 0}
 579 diff --git a/rsync.yo b/rsync.yo
 580 --- a/rsync.yo
 581 +++ b/rsync.yo
 582 @@ -388,6 +388,7 @@ to the detailed description below for a complete description.  verb(
 583       --compare-dest=DIR      also compare received files relative to DIR
 584       --copy-dest=DIR         ... and include copies of unchanged files
 585       --link-dest=DIR         hardlink to files in DIR when unchanged
 586 +     --link-by-hash=DIR      create hardlinks by hash into DIR
 587   -z, --compress              compress file data during the transfer
 588       --compress-level=NUM    explicitly set compression level
 589       --skip-compress=LIST    skip compressing files with suffix in LIST