bin/cpdup/cpdup.c

   1 /*-
   2  * CPDUP.C
   3  *
   4  * CPDUP <options> source destination
   5  *
   6  * (c) Copyright 1997-1999 by Matthew Dillon and Dima Ruban.  Permission to
   7  *     use and distribute based on the FreeBSD copyright.  Supplied as-is,
   8  *     USE WITH EXTREME CAUTION.
   9  *
  10  * This program attempts to duplicate the source onto the destination as
  11  * exactly as possible, retaining modify times, flags, perms, uid, and gid.
  12  * It can duplicate devices, files (including hardlinks), softlinks,
  13  * directories, and so forth.  It is recursive by default!  The duplication
  14  * is inclusive of removal of files/directories on the destination that do
  15  * not exist on the source.  This program supports a per-directory exception
  16  * file called .cpignore, or a user-specified exception file.
  17  *
  18  * Safety features:
  19  *
  20  *      - does not cross partition boundries on source
  21  *      - asks for confirmation on deletions unless -i0 is specified
  22  *      - refuses to replace a destination directory with a source file
  23  *        unless -s0 is specified.
  24  *      - terminates on error
  25  *
  26  * Copying features:
  27  *
  28  *      - does not copy file if mtime, flags, perms, and size match unless
  29  *        forced
  30  *
  31  *      - copies to temporary and renames-over the original, allowing
  32  *        you to update live systems
  33  *
  34  *      - copies uid, gid, mtime, perms, flags, softlinks, devices, hardlinks,
  35  *        and recurses through directories.
  36  *
  37  *      - accesses a per-directory exclusion file, .cpignore, containing
  38  *        standard wildcarded ( ? / * style, NOT regex) exclusions.
  39  *
  40  *      - tries to play permissions and flags smart in regards to overwriting
  41  *        schg files and doing related stuff.
  42  *
  43  *      - Can do MD5 consistancy checks
  44  *
  45  *      - Is able to do incremental mirroring/backups via hardlinks from
  46  *        the 'previous' version (supplied with -H path).
  47  *
  48  * $DragonFly: src/bin/cpdup/cpdup.c,v 1.32 2008/11/11 04:36:00 dillon Exp $
  49  */
  50
  51 /*-
  52  * Example: cc -O cpdup.c -o cpdup -lmd
  53  *
  54  * ".MD5.CHECKSUMS" contains md5 checksumms for the current directory.
  55  * This file is stored on the source.
  56  */
  57
  58 #include "cpdup.h"
  59 #include "hclink.h"
  60 #include "hcproto.h"
  61
  62 #define HSIZE   8192
  63 #define HMASK   (HSIZE-1)
  64 #define HLSIZE  8192
  65 #define HLMASK  (HLSIZE - 1)
  66
  67 #define GETBUFSIZE      8192
  68 #define GETPATHSIZE     2048
  69 #define GETLINKSIZE     1024
  70 #define GETIOSIZE       65536
  71
  72 #ifndef _ST_FLAGS_PRESENT_
  73 #define st_flags        st_mode
  74 #endif
  75
  76 typedef struct Node {
  77     struct Node *no_Next;
  78     struct Node *no_HNext;
  79     int  no_Value;
  80     char no_Name[4];
  81 } Node;
  82
  83 typedef struct List {
  84     Node        li_Node;
  85     Node        *li_Hash[HSIZE];
  86 } List;
  87
  88 struct hlink {
  89     ino_t ino;
  90     ino_t dino;
  91     int refs;
  92     struct hlink *next;
  93     struct hlink *prev;
  94     nlink_t nlinked;
  95     char name[0];
  96 };
  97
  98 typedef struct copy_info {
  99         char *spath;
 100         char *dpath;
 101         dev_t sdevNo;
 102         dev_t ddevNo;
 103 } *copy_info_t;
 104
 105 struct hlink *hltable[HLSIZE];
 106
 107 void RemoveRecur(const char *dpath, dev_t devNo);
 108 void InitList(List *list);
 109 void ResetList(List *list);
 110 char *IterateList(List *list, Node **nodeptr, int n);
 111 int AddList(List *list, const char *name, int n);
 112 static int getbool(const char *str);
 113 static char *SplitRemote(char *path);
 114 static int ChgrpAllowed(gid_t g);
 115 static int OwnerMatch(struct stat *st1, struct stat *st2);
 116 #ifdef _ST_FLAGS_PRESENT_
 117 static int FlagsMatch(struct stat *st1, struct stat *st2);
 118 #else
 119 #define FlagsMatch(st1, st2)    1
 120 #endif
 121 static struct hlink *hltlookup(struct stat *);
 122 static struct hlink *hltadd(struct stat *, const char *);
 123 static char *checkHLPath(struct stat *st, const char *spath, const char *dpath);
 124 static int validate_check(const char *spath, const char *dpath);
 125 static int shash(const char *s);
 126 static void hltdelete(struct hlink *);
 127 static void hltsetdino(struct hlink *, ino_t);
 128 int YesNo(const char *path);
 129 static int xrename(const char *src, const char *dst, u_long flags);
 130 static int xlink(const char *src, const char *dst, u_long flags);
 131 static int xremove(struct HostConf *host, const char *path);
 132 static int DoCopy(copy_info_t info, int depth);
 133 static int ScanDir(List *list, struct HostConf *host, const char *path,
 134         int64_t *CountReadBytes, int n);
 135
 136 int AskConfirmation = 1;
 137 int SafetyOpt = 1;
 138 int ForceOpt;
 139 int DeviceOpt = 1;
 140 int VerboseOpt;
 141 int DirShowOpt;
 142 int QuietOpt;
 143 int NoRemoveOpt;
 144 int UseMD5Opt;
 145 int UseFSMIDOpt;
 146 int SummaryOpt;
 147 int CompressOpt;
 148 int SlaveOpt;
 149 int EnableDirectoryRetries;
 150 int DstBaseLen;
 151 int ValidateOpt;
 152 int HardLinkCount;
 153 int ssh_argc;
 154 const char *ssh_argv[16];
 155 int DstRootPrivs;
 156 int GroupCount;
 157 gid_t *GroupList;
 158 const char *UseCpFile;
 159 const char *UseHLPath;
 160 const char *MD5CacheFile;
 161 const char *FSMIDCacheFile;
 162
 163 int64_t CountSourceBytes;
 164 int64_t CountSourceItems;
 165 int64_t CountCopiedItems;
 166 int64_t CountSourceReadBytes;
 167 int64_t CountTargetReadBytes;
 168 int64_t CountWriteBytes;
 169 int64_t CountRemovedItems;
 170 int64_t CountLinkedItems;
 171
 172 struct HostConf SrcHost;
 173 struct HostConf DstHost;
 174
 175 int
 176 main(int ac, char **av)
 177 {
 178     int i;
 179     int opt;
 180     char *src = NULL;
 181     char *dst = NULL;
 182     char *ptr;
 183     struct timeval start;
 184     struct copy_info info;
 185
 186     signal(SIGPIPE, SIG_IGN);
 187
 188     gettimeofday(&start, NULL);
 189     opterr = 0;
 190     while ((opt = getopt(ac, av, ":CdF:fH:Ii:j:K:klM:mopqSs:uVvX:x")) != -1) {
 191         switch (opt) {
 192         /* TODO: sort the branches */
 193         case 'C':
 194             CompressOpt = 1;
 195             break;
 196         case 'v':
 197             ++VerboseOpt;
 198             break;
 199         case 'd':
 200             DirShowOpt = 1;
 201             break;
 202         case 'l':
 203             setlinebuf(stdout);
 204             setlinebuf(stderr);
 205             break;
 206         case 'V':
 207             ++ValidateOpt;
 208             break;
 209         case 'I':
 210             SummaryOpt = 1;
 211             break;
 212         case 'o':
 213             NoRemoveOpt = 1;
 214             break;
 215         case 'x':
 216             UseCpFile = ".cpignore";
 217             break;
 218         case 'X':
 219             UseCpFile = optarg;
 220             break;
 221         case 'H':
 222             UseHLPath = optarg;
 223             break;
 224         case 'F':
 225             if (ssh_argc >= 16)
 226                 fatal("too many -F options");
 227             ssh_argv[ssh_argc++] = optarg;
 228             break;
 229         case 'S':
 230             SlaveOpt = 1;
 231             break;
 232         case 'f':
 233             ForceOpt = 1;
 234             break;
 235         case 'i':
 236             AskConfirmation = getbool(optarg);
 237             break;
 238         case 'j':
 239             DeviceOpt = getbool(optarg);
 240             break;
 241         case 's':
 242             SafetyOpt = getbool(optarg);
 243             break;
 244         case 'q':
 245             QuietOpt = 1;
 246             break;
 247         case 'k':
 248             UseFSMIDOpt = 1;
 249             FSMIDCacheFile = ".FSMID.CHECK";
 250             break;
 251         case 'K':
 252             UseFSMIDOpt = 1;
 253             FSMIDCacheFile = optarg;
 254             break;
 255         case 'M':
 256             UseMD5Opt = 1;
 257             MD5CacheFile = optarg;
 258             break;
 259         case 'm':
 260             UseMD5Opt = 1;
 261             MD5CacheFile = ".MD5.CHECKSUMS";
 262             break;
 263         case 'u':
 264             setvbuf(stdout, NULL, _IOLBF, 0);
 265             break;
 266         case ':':
 267             fatal("missing argument for option: -%c\n", optopt);
 268             /* not reached */
 269             break;
 270         case '?':
 271             fatal("illegal option: -%c\n", optopt);
 272             /* not reached */
 273             break;
 274         default:
 275             fatal(NULL);
 276             /* not reached */
 277             break;
 278         }
 279     }
 280     ac -= optind;
 281     av += optind;
 282     if (ac > 0)
 283         src = av[0];
 284     if (ac > 1)
 285         dst = av[1];
 286     if (ac > 2)
 287         fatal("too many arguments");
 288
 289     /*
 290      * If we are told to go into slave mode, run the HC protocol
 291      */
 292     if (SlaveOpt) {
 293         DstRootPrivs = (geteuid() == 0);
 294         hc_slave(0, 1);
 295         exit(0);
 296     }
 297
 298     /*
 299      * Extract the source and/or/neither target [user@]host and
 300      * make any required connections.
 301      */
 302     if (src && (ptr = SplitRemote(src)) != NULL) {
 303         SrcHost.host = src;
 304         src = ptr;
 305         if (UseMD5Opt) {
 306             fprintf(stderr, "The MD5 options are not currently supported for remote sources\n");
 307             exit(1);
 308         }
 309         if (hc_connect(&SrcHost) < 0)
 310             exit(1);
 311     }
 312     if (dst && (ptr = SplitRemote(dst)) != NULL) {
 313         DstHost.host = dst;
 314         dst = ptr;
 315         if (UseFSMIDOpt) {
 316             fprintf(stderr, "The FSMID options are not currently supported for remote targets\n");
 317             exit(1);
 318         }
 319         if (hc_connect(&DstHost) < 0)
 320             exit(1);
 321     }
 322
 323     /*
 324      * dst may be NULL only if -m option is specified,
 325      * which forces an update of the MD5 checksums
 326      */
 327     if (dst == NULL && UseMD5Opt == 0) {
 328         fatal(NULL);
 329         /* not reached */
 330     }
 331
 332     if (dst) {
 333         DstRootPrivs = (hc_geteuid(&DstHost) == 0);
 334         if (!DstRootPrivs)
 335             GroupCount = hc_getgroups(&DstHost, &GroupList);
 336     }
 337 #if 0
 338     /* XXXX DEBUG */
 339     fprintf(stderr, "DstRootPrivs == %s\n", DstRootPrivs ? "true" : "false");
 340     fprintf(stderr, "GroupCount == %d\n", GroupCount);
 341     for (i = 0; i < GroupCount; i++)
 342         fprintf(stderr, "Group[%d] == %d\n", i, GroupList[i]);
 343 #endif
 344
 345     bzero(&info, sizeof(info));
 346     if (dst) {
 347         DstBaseLen = strlen(dst);
 348         info.spath = src;
 349         info.dpath = dst;
 350         info.sdevNo = (dev_t)-1;
 351         info.ddevNo = (dev_t)-1;
 352         i = DoCopy(&info, -1);
 353     } else {
 354         info.spath = src;
 355         info.dpath = NULL;
 356         info.sdevNo = (dev_t)-1;
 357         info.ddevNo = (dev_t)-1;
 358         i = DoCopy(&info, -1);
 359     }
 360 #ifndef NOMD5
 361     md5_flush();
 362 #endif
 363     fsmid_flush();
 364
 365     if (SummaryOpt && i == 0) {
 366         double duration;
 367         struct timeval end;
 368
 369         gettimeofday(&end, NULL);
 370 #if 0
 371         /* don't count stat's in our byte statistics */
 372         CountSourceBytes += sizeof(struct stat) * CountSourceItems;
 373         CountSourceReadBytes += sizeof(struct stat) * CountSourceItems;
 374         CountWriteBytes +=  sizeof(struct stat) * CountCopiedItems;
 375         CountWriteBytes +=  sizeof(struct stat) * CountRemovedItems;
 376 #endif
 377
 378         duration = (end.tv_sec - start.tv_sec);
 379         duration += (double)(end.tv_usec - start.tv_usec) / 1000000.0;
 380         if (duration == 0.0)
 381                 duration = 1.0;
 382         logstd("cpdup completed successfully\n");
 383         logstd("%lld bytes source, %lld src bytes read, %lld tgt bytes read\n"
 384                "%lld bytes written (%.1fX speedup)\n",
 385             (long long)CountSourceBytes,
 386             (long long)CountSourceReadBytes,
 387             (long long)CountTargetReadBytes,
 388             (long long)CountWriteBytes,
 389             ((double)CountSourceBytes * 2.0) / ((double)(CountSourceReadBytes + CountTargetReadBytes + CountWriteBytes)));
 390         logstd("%lld source items, %lld items copied, %lld items linked, "
 391                "%lld things deleted\n",
 392             (long long)CountSourceItems,
 393             (long long)CountCopiedItems,
 394             (long long)CountLinkedItems,
 395             (long long)CountRemovedItems);
 396         logstd("%.1f seconds %5d Kbytes/sec synced %5d Kbytes/sec scanned\n",
 397             duration,
 398             (int)((CountSourceReadBytes + CountTargetReadBytes + CountWriteBytes) / duration  / 1024.0),
 399             (int)(CountSourceBytes / duration / 1024.0));
 400     }
 401     exit((i == 0) ? 0 : 1);
 402 }
 403
 404 static int
 405 getbool(const char *str)
 406 {
 407     if (strcmp(str, "0") == 0)
 408         return (0);
 409     if (strcmp(str, "1") == 0)
 410         return (1);
 411     fatal("option requires boolean argument (0 or 1): -%c\n", optopt);
 412     /* not reached */
 413     return (0);
 414 }
 415
 416 /*
 417  * Check if path specifies a remote path, using the same syntax as scp(1),
 418  * i.e. a path is considered remote if the first colon is not preceded by
 419  * a slash, so e.g. "./foo:bar" is considered local.
 420  * If a remote path is detected, the colon is replaced with a null byte,
 421  * and the return value is a pointer to the next character.
 422  * Otherwise NULL is returned.
 423  */
 424 static char *
 425 SplitRemote(char *path)
 426 {
 427     int cindex;
 428
 429     if (path[(cindex = strcspn(path, ":/"))] == ':') {
 430         path[cindex++] = 0;
 431         return (path + cindex);
 432     }
 433     return (NULL);
 434 }
 435
 436 /*
 437  * Check if group g is in our GroupList.
 438  *
 439  * Typically the number of groups a user belongs to isn't large
 440  * enough to warrant more effort than a simple linear search.
 441  * However, we perform an optimization by moving a group to the
 442  * top of the list when we have a hit.  This assumes that there
 443  * isn't much variance in the gids of files that a non-root user
 444  * copies.  So most of the time the search will terminate on the
 445  * first element of the list.
 446  */
 447 static int
 448 ChgrpAllowed(gid_t g)
 449 {
 450     int i;
 451
 452     for (i = 0; i < GroupCount; i++)
 453         if (GroupList[i] == g) {
 454             if (i > 0) {
 455                 /* Optimize: Move g to the front of the list. */
 456                 for (; i > 0; i--)
 457                     GroupList[i] = GroupList[i - 1];
 458                 GroupList[0] = g;
 459             }
 460             return (1);
 461         }
 462     return (0);
 463 }
 464
 465 /*
 466  * The following two functions return true if the ownership (UID + GID)
 467  * or the flags of two files match, respectively.
 468  *
 469  * Only perform weak checking if we don't have sufficient privileges on
 470  * the target machine, so we don't waste transfers with things that are
 471  * bound to fail anyway.
 472  */
 473 static int
 474 OwnerMatch(struct stat *st1, struct stat *st2)
 475 {
 476     if (DstRootPrivs)
 477         /* Both UID and GID must match. */
 478         return (st1->st_uid == st2->st_uid && st1->st_gid == st2->st_gid);
 479     else
 480         /* Ignore UID, and also ignore GID if we can't chgrp to that group. */
 481         return (st1->st_gid == st2->st_gid || !ChgrpAllowed(st1->st_gid));
 482 }
 483
 484 #ifdef _ST_FLAGS_PRESENT_
 485 static int
 486 FlagsMatch(struct stat *st1, struct stat *st2)
 487 {
 488     if (DstRootPrivs)
 489         return (st1->st_flags == st2->st_flags);
 490     else
 491         /* Only consider the user-settable flags. */
 492         return (((st1->st_flags ^ st2->st_flags) & UF_SETTABLE) == 0);
 493 }
 494 #endif
 495
 496
 497 static struct hlink *
 498 hltlookup(struct stat *stp)
 499 {
 500     struct hlink *hl;
 501     int n;
 502
 503     n = stp->st_ino & HLMASK;
 504
 505     for (hl = hltable[n]; hl; hl = hl->next) {
 506         if (hl->ino == stp->st_ino) {
 507             ++hl->refs;
 508             return hl;
 509         }
 510     }
 511
 512     return NULL;
 513 }
 514
 515 static struct hlink *
 516 hltadd(struct stat *stp, const char *path)
 517 {
 518     struct hlink *new;
 519     int plen = strlen(path);
 520     int n;
 521
 522     new = malloc(offsetof(struct hlink, name[plen + 1]));
 523     if (new == NULL) {
 524         fprintf(stderr, "out of memory\n");
 525         exit(EXIT_FAILURE);
 526     }
 527     ++HardLinkCount;
 528
 529     /* initialize and link the new element into the table */
 530     new->ino = stp->st_ino;
 531     new->dino = (ino_t)-1;
 532     new->refs = 1;
 533     bcopy(path, new->name, plen + 1);
 534     new->nlinked = 1;
 535     new->prev = NULL;
 536     n = stp->st_ino & HLMASK;
 537     new->next = hltable[n];
 538     if (hltable[n])
 539         hltable[n]->prev = new;
 540     hltable[n] = new;
 541
 542     return new;
 543 }
 544
 545 static void
 546 hltsetdino(struct hlink *hl, ino_t inum)
 547 {
 548     hl->dino = inum;
 549 }
 550
 551 static void
 552 hltdelete(struct hlink *hl)
 553 {
 554     assert(hl->refs == 1);
 555     --hl->refs;
 556     if (hl->prev) {
 557         if (hl->next)
 558             hl->next->prev = hl->prev;
 559         hl->prev->next = hl->next;
 560     } else {
 561         if (hl->next)
 562             hl->next->prev = NULL;
 563
 564         hltable[hl->ino & HLMASK] = hl->next;
 565     }
 566     --HardLinkCount;
 567     free(hl);
 568 }
 569
 570 static void
 571 hltrels(struct hlink *hl)
 572 {
 573     assert(hl->refs == 1);
 574     --hl->refs;
 575 }
 576
 577 /*
 578  * If UseHLPath is defined check to see if the file in question is
 579  * the same as the source file, and if it is return a pointer to the
 580  * -H path based file for hardlinking.  Else return NULL.
 581  */
 582 static char *
 583 checkHLPath(struct stat *st1, const char *spath, const char *dpath)
 584 {
 585     struct stat sthl;
 586     char *hpath;
 587     int error;
 588
 589     asprintf(&hpath, "%s%s", UseHLPath, dpath + DstBaseLen);
 590
 591     /*
 592      * stat info matches ?
 593      */
 594     if (hc_stat(&DstHost, hpath, &sthl) < 0 ||
 595         st1->st_size != sthl.st_size ||
 596         st1->st_mtime != sthl.st_mtime ||
 597         !OwnerMatch(st1, &sthl) ||
 598         !FlagsMatch(st1, &sthl)
 599     ) {
 600         free(hpath);
 601         return(NULL);
 602     }
 603
 604     /*
 605      * If ForceOpt or ValidateOpt is set we have to compare the files
 606      */
 607     if (ForceOpt || ValidateOpt) {
 608         error = validate_check(spath, hpath);
 609         if (error) {
 610             free(hpath);
 611             hpath = NULL;
 612         }
 613     }
 614     return(hpath);
 615 }
 616
 617 /*
 618  * Return 0 if the contents of the file <spath> matches the contents of
 619  * the file <dpath>.
 620  */
 621 static int
 622 validate_check(const char *spath, const char *dpath)
 623 {
 624     int error;
 625     int fd1;
 626     int fd2;
 627
 628     fd1 = hc_open(&SrcHost, spath, O_RDONLY, 0);
 629     fd2 = hc_open(&DstHost, dpath, O_RDONLY, 0);
 630     error = -1;
 631
 632     if (fd1 >= 0 && fd2 >= 0) {
 633         int n;
 634         int x;
 635         char *iobuf1 = malloc(GETIOSIZE);
 636         char *iobuf2 = malloc(GETIOSIZE);
 637
 638         while ((n = hc_read(&SrcHost, fd1, iobuf1, GETIOSIZE)) > 0) {
 639             CountSourceReadBytes += n;
 640             x = hc_read(&DstHost, fd2, iobuf2, GETIOSIZE);
 641             if (x > 0)
 642                     CountTargetReadBytes += x;
 643             if (x != n)
 644                 break;
 645             if (bcmp(iobuf1, iobuf2, n) != 0)
 646                 break;
 647         }
 648         free(iobuf1);
 649         free(iobuf2);
 650         if (n == 0)
 651             error = 0;
 652     }
 653     if (fd1 >= 0)
 654         hc_close(&SrcHost, fd1);
 655     if (fd2 >= 0)
 656         hc_close(&DstHost, fd2);
 657     return (error);
 658 }
 659
 660 int
 661 DoCopy(copy_info_t info, int depth)
 662 {
 663     const char *spath = info->spath;
 664     const char *dpath = info->dpath;
 665     dev_t sdevNo = info->sdevNo;
 666     dev_t ddevNo = info->ddevNo;
 667     struct stat st1;
 668     struct stat st2;
 669     unsigned long st2_flags;
 670     int r, mres, fres, st2Valid;
 671     struct hlink *hln;
 672     u_int64_t size;
 673
 674     r = mres = fres = st2Valid = 0;
 675     st2_flags = 0;
 676     size = 0;
 677     hln = NULL;
 678
 679     if (hc_lstat(&SrcHost, spath, &st1) != 0) {
 680         r = 1;
 681         goto done;
 682     }
 683 #ifdef SF_SNAPSHOT
 684     /* skip snapshot files because they're sparse and _huge_ */
 685     if (st1.st_flags & SF_SNAPSHOT)
 686        return(0);
 687 #endif
 688     st2.st_mode = 0;    /* in case lstat fails */
 689     st2.st_flags = 0;   /* in case lstat fails */
 690     if (dpath && hc_lstat(&DstHost, dpath, &st2) == 0) {
 691         st2Valid = 1;
 692 #ifdef _ST_FLAGS_PRESENT_
 693         st2_flags = st2.st_flags;
 694 #endif
 695     }
 696
 697     if (S_ISREG(st1.st_mode)) {
 698         size = st1.st_size;
 699     }
 700
 701     /*
 702      * Handle hardlinks
 703      */
 704
 705     if (S_ISREG(st1.st_mode) && st1.st_nlink > 1 && dpath) {
 706         if ((hln = hltlookup(&st1)) != NULL) {
 707             hln->nlinked++;
 708
 709             if (st2Valid) {
 710                 if (st2.st_ino == hln->dino) {
 711                     /*
 712                      * hard link is already correct, nothing to do
 713                      */
 714                     if (VerboseOpt >= 3)
 715                         logstd("%-32s nochange\n", (dpath) ? dpath : spath);
 716                     if (hln->nlinked == st1.st_nlink) {
 717                         hltdelete(hln);
 718                         hln = NULL;
 719                     }
 720                     CountSourceItems++;
 721                     r = 0;
 722                     goto done;
 723                 } else {
 724                     /*
 725                      * hard link is not correct, attempt to unlink it
 726                      */
 727                     if (xremove(&DstHost, dpath) < 0) {
 728                         logerr("%-32s hardlink: unable to unlink: %s\n",
 729                             ((dpath) ? dpath : spath), strerror(errno));
 730                         hltdelete(hln);
 731                         hln = NULL;
 732                         ++r;
 733                         goto done;
 734                     }
 735                 }
 736             }
 737
 738             if (xlink(hln->name, dpath, st1.st_flags) < 0) {
 739                 int tryrelink = (errno == EMLINK);
 740                 logerr("%-32s hardlink: unable to link to %s: %s\n",
 741                     (dpath ? dpath : spath), hln->name, strerror(errno)
 742                 );
 743                 hltdelete(hln);
 744                 hln = NULL;
 745                 if (tryrelink) {
 746                     logerr("%-20s hardlink: will attempt to copy normally\n");
 747                     goto relink;
 748                 }
 749                 ++r;
 750             } else {
 751                 if (hln->nlinked == st1.st_nlink) {
 752                     hltdelete(hln);
 753                     hln = NULL;
 754                 }
 755                 if (r == 0) {
 756                     if (VerboseOpt) {
 757                         logstd("%-32s hardlink: %s\n",
 758                             (dpath ? dpath : spath),
 759                             (st2Valid ? "relinked" : "linked")
 760                         );
 761                     }
 762                     CountSourceItems++;
 763                     CountCopiedItems++;
 764                     r = 0;
 765                     goto done;
 766                 }
 767             }
 768         } else {
 769             /*
 770              * first instance of hardlink must be copied normally
 771              */
 772 relink:
 773             hln = hltadd(&st1, dpath);
 774         }
 775     }
 776
 777     /*
 778      * Do we need to copy the file/dir/link/whatever?  Early termination
 779      * if we do not.  Always redo links.  Directories are always traversed
 780      * except when the FSMID options are used.
 781      *
 782      * NOTE: st2Valid is true only if dpath != NULL *and* dpath stats good.
 783      */
 784
 785     if (
 786         st2Valid
 787         && st1.st_mode == st2.st_mode
 788         && FlagsMatch(&st1, &st2)
 789     ) {
 790         if (S_ISLNK(st1.st_mode) || S_ISDIR(st1.st_mode)) {
 791             /*
 792              * If FSMID tracking is turned on we can avoid recursing through
 793              * an entire directory subtree if the FSMID matches.
 794              */
 795 #ifdef _ST_FSMID_PRESENT_
 796             if (ForceOpt == 0 &&
 797                 (UseFSMIDOpt && (fres = fsmid_check(st1.st_fsmid, dpath)) == 0)
 798             ) {
 799                 if (VerboseOpt >= 3) {
 800                     if (UseFSMIDOpt) /* always true!?! */
 801                         logstd("%-32s fsmid-nochange\n", (dpath ? dpath : spath));
 802                     else
 803                         logstd("%-32s nochange\n", (dpath ? dpath : spath));
 804                 }
 805                 r = 0;
 806                 goto done;
 807             }
 808 #endif
 809         } else {
 810             if (ForceOpt == 0 &&
 811                 st1.st_size == st2.st_size &&
 812                 (ValidateOpt == 2 || st1.st_mtime == st2.st_mtime) &&
 813                 OwnerMatch(&st1, &st2)
 814 #ifndef NOMD5
 815                 && (UseMD5Opt == 0 || !S_ISREG(st1.st_mode) ||
 816                     (mres = md5_check(spath, dpath)) == 0)
 817 #endif
 818 #ifdef _ST_FSMID_PRESENT_
 819                 && (UseFSMIDOpt == 0 ||
 820                     (fres = fsmid_check(st1.st_fsmid, dpath)) == 0)
 821 #endif
 822                 && (ValidateOpt == 0 || !S_ISREG(st1.st_mode) ||
 823                     validate_check(spath, dpath) == 0)
 824             ) {
 825                 /*
 826                  * The files are identical, but if we are running as
 827                  * root we might need to adjust ownership/group/flags.
 828                  */
 829                 int changedown = 0;
 830                 int changedflags = 0;
 831
 832                 if (hln)
 833                     hltsetdino(hln, st2.st_ino);
 834
 835                 if (!OwnerMatch(&st1, &st2)) {
 836                     hc_chown(&DstHost, dpath, st1.st_uid, st1.st_gid);
 837                     changedown = 1;
 838                 }
 839 #ifdef _ST_FLAGS_PRESENT_
 840                 if (!FlagsMatch(&st1, &st2)) {
 841                     hc_chflags(&DstHost, dpath, st1.st_flags);
 842                     changedflags = 1;
 843                 }
 844 #endif
 845                 if (VerboseOpt >= 3) {
 846 #ifndef NOMD5
 847                     if (UseMD5Opt) {
 848                         logstd("%-32s md5-nochange",
 849                                 (dpath ? dpath : spath));
 850                     } else
 851 #endif
 852                     if (UseFSMIDOpt) {
 853                         logstd("%-32s fsmid-nochange",
 854                                 (dpath ? dpath : spath));
 855                     } else if (ValidateOpt) {
 856                         logstd("%-32s nochange (contents validated)",
 857                                 (dpath ? dpath : spath));
 858                     } else {
 859                         logstd("%-32s nochange", (dpath ? dpath : spath));
 860                     }
 861                     if (changedown)
 862                         logstd(" (uid/gid differ)");
 863                     if (changedflags)
 864                         logstd(" (flags differ)");
 865                     logstd("\n");
 866                 }
 867                 CountSourceBytes += size;
 868                 CountSourceItems++;
 869                 r = 0;
 870                 goto done;
 871             }
 872         }
 873     }
 874     if (st2Valid && !S_ISDIR(st1.st_mode) && S_ISDIR(st2.st_mode)) {
 875         if (SafetyOpt) {
 876             logerr("%-32s SAFETY - refusing to copy file over directory\n",
 877                 (dpath ? dpath : spath)
 878             );
 879             ++r;                /* XXX */
 880             r = 0;
 881             goto done;          /* continue with the cpdup anyway */
 882         }
 883         if (QuietOpt == 0 || AskConfirmation) {
 884             logstd("%-32s WARNING: non-directory source will blow away\n"
 885                    "%-32s preexisting dest directory, continuing anyway!\n",
 886                    ((dpath) ? dpath : spath), "");
 887         }
 888         if (dpath)
 889             RemoveRecur(dpath, ddevNo);
 890         st2Valid = 0;
 891     }
 892
 893     /*
 894      * The various comparisons failed, copy it.
 895      */
 896     if (S_ISDIR(st1.st_mode)) {
 897         int skipdir = 0;
 898
 899         if (fres < 0)
 900             logerr("%-32s/ fsmid-CHECK-FAILED\n", (dpath) ? dpath : spath);
 901
 902         if (dpath) {
 903             if (!st2Valid || S_ISDIR(st2.st_mode) == 0) {
 904                 if (st2Valid)
 905                     xremove(&DstHost, dpath);
 906                 if (hc_mkdir(&DstHost, dpath, st1.st_mode | 0700) != 0) {
 907                     logerr("%s: mkdir failed: %s\n",
 908                         (dpath ? dpath : spath), strerror(errno));
 909                     r = 1;
 910                     skipdir = 1;
 911                 }
 912                 if (hc_lstat(&DstHost, dpath, &st2) != 0) {
 913                     logerr("%s: lstat of newly made dir failed: %s\n",
 914                         (dpath ? dpath : spath), strerror(errno));
 915                     st2Valid = 0;
 916                     r = 1;
 917                     skipdir = 1;
 918                 }
 919                 else {
 920                     st2Valid = 1;
 921                     if (!OwnerMatch(&st1, &st2) &&
 922                         hc_chown(&DstHost, dpath, st1.st_uid, st1.st_gid) != 0
 923                     ) {
 924                         logerr("%s: chown of newly made dir failed: %s\n",
 925                             (dpath ? dpath : spath), strerror(errno));
 926                         r = 1;
 927                         /* Note that we should not set skipdir = 1 here. */
 928                     }
 929                 }
 930                 CountCopiedItems++;
 931             } else {
 932                 /*
 933                  * Directory must be scanable by root for cpdup to
 934                  * work.  We'll fix it later if the directory isn't
 935                  * supposed to be readable ( which is why we fixup
 936                  * st2.st_mode to match what we did ).
 937                  */
 938                 if ((st2.st_mode & 0700) != 0700) {
 939                     hc_chmod(&DstHost, dpath, st2.st_mode | 0700);
 940                     st2.st_mode |= 0700;
 941                 }
 942                 if (VerboseOpt >= 2)
 943                     logstd("%s\n", dpath ? dpath : spath);
 944             }
 945         }
 946
 947         /*
 948          * When copying a directory, stop if the source crosses a mount
 949          * point.
 950          */
 951         if (sdevNo != (dev_t)-1 && st1.st_dev != sdevNo)
 952             skipdir = 1;
 953         else
 954             sdevNo = st1.st_dev;
 955
 956         /*
 957          * When copying a directory, stop if the destination crosses
 958          * a mount point.
 959          *
 960          * The target directory will have been created and stat'd
 961          * for st2 if it did not previously exist.   st2Valid is left
 962          * as a flag.  If the stat failed st2 will still only have its
 963          * default initialization.
 964          *
 965          * So we simply assume here that the directory is within the
 966          * current target mount if we had to create it (aka st2Valid is 0)
 967          * and we leave ddevNo alone.
 968          */
 969         if (st2Valid) {
 970             if (ddevNo != (dev_t)-1 && st2.st_dev != ddevNo)
 971                 skipdir = 1;
 972             else
 973                 ddevNo = st2.st_dev;
 974         }
 975
 976         if (!skipdir) {
 977             List *list = malloc(sizeof(List));
 978             Node *node = NULL;
 979             char *name;
 980
 981             if (DirShowOpt)
 982                 logstd("Scanning %s ...\n", spath);
 983             InitList(list);
 984             if (ScanDir(list, &SrcHost, spath, &CountSourceReadBytes, 0) == 0) {
 985                 while ((name = IterateList(list, &node, 0)) != NULL) {
 986                     char *nspath;
 987                     char *ndpath = NULL;
 988
 989                     nspath = mprintf("%s/%s", spath, name);
 990                     if (dpath)
 991                         ndpath = mprintf("%s/%s", dpath, name);
 992
 993                     info->spath = nspath;
 994                     info->dpath = ndpath;
 995                     info->sdevNo = sdevNo;
 996                     info->ddevNo = ddevNo;
 997                     if (depth < 0)
 998                         r += DoCopy(info, depth);
 999                     else
1000                         r += DoCopy(info, depth + 1);
1001                     free(nspath);
1002                     if (ndpath)
1003                         free(ndpath);
1004                     info->spath = NULL;
1005                     info->dpath = NULL;
1006                 }
1007
1008                 /*
1009                  * Remove files/directories from destination that do not appear
1010                  * in the source.
1011                  */
1012                 if (dpath && ScanDir(list, &DstHost, dpath,
1013                         &CountTargetReadBytes, 3) == 0) {
1014                     while ((name = IterateList(list, &node, 3)) != NULL) {
1015                         /*
1016                          * If object does not exist in source or .cpignore
1017                          * then recursively remove it.
1018                          */
1019                         char *ndpath;
1020
1021                         ndpath = mprintf("%s/%s", dpath, name);
1022                         RemoveRecur(ndpath, ddevNo);
1023                         free(ndpath);
1024                     }
1025                 }
1026             }
1027             ResetList(list);
1028             free(list);
1029         }
1030
1031         if (dpath && st2Valid) {
1032             struct timeval tv[2];
1033
1034             if (ForceOpt || !OwnerMatch(&st1, &st2))
1035                 hc_chown(&DstHost, dpath, st1.st_uid, st1.st_gid);
1036             if (st1.st_mode != st2.st_mode)
1037                 hc_chmod(&DstHost, dpath, st1.st_mode);
1038 #ifdef _ST_FLAGS_PRESENT_
1039             if (!FlagsMatch(&st1, &st2))
1040                 hc_chflags(&DstHost, dpath, st1.st_flags);
1041 #endif
1042             if (ForceOpt || st1.st_mtime != st2.st_mtime) {
1043                 bzero(tv, sizeof(tv));
1044                 tv[0].tv_sec = st1.st_mtime;
1045                 tv[1].tv_sec = st1.st_mtime;
1046                 hc_utimes(&DstHost, dpath, tv);
1047             }
1048         }
1049     } else if (dpath == NULL) {
1050         /*
1051          * If dpath is NULL, we are just updating the MD5
1052          */
1053 #ifndef NOMD5
1054         if (UseMD5Opt && S_ISREG(st1.st_mode)) {
1055             mres = md5_check(spath, NULL);
1056
1057             if (VerboseOpt > 1) {
1058                 if (mres < 0)
1059                     logstd("%-32s md5-update\n", (dpath) ? dpath : spath);
1060                 else
1061                     logstd("%-32s md5-ok\n", (dpath) ? dpath : spath);
1062             } else if (!QuietOpt && mres < 0) {
1063                 logstd("%-32s md5-update\n", (dpath) ? dpath : spath);
1064             }
1065         }
1066 #endif
1067     } else if (S_ISREG(st1.st_mode)) {
1068         char *path;
1069         char *hpath;
1070         int fd1;
1071         int fd2;
1072
1073         if (st2Valid)
1074                 path = mprintf("%s.tmp%d", dpath, (int)getpid());
1075         else
1076                 path = mprintf("%s", dpath);
1077
1078         /*
1079          * Handle check failure message.
1080          */
1081 #ifndef NOMD5
1082         if (mres < 0)
1083             logerr("%-32s md5-CHECK-FAILED\n", (dpath) ? dpath : spath);
1084         else
1085 #endif
1086         if (fres < 0)
1087             logerr("%-32s fsmid-CHECK-FAILED\n", (dpath) ? dpath : spath);
1088
1089         /*
1090          * Not quite ready to do the copy yet.  If UseHLPath is defined,
1091          * see if we can hardlink instead.
1092          *
1093          * If we can hardlink, and the target exists, we have to remove it
1094          * first or the hardlink will fail.  This can occur in a number of
1095          * situations but most typically when the '-f -H' combination is
1096          * used.
1097          */
1098         if (UseHLPath && (hpath = checkHLPath(&st1, spath, dpath)) != NULL) {
1099                 if (st2Valid)
1100                         xremove(&DstHost, dpath);
1101                 if (hc_link(&DstHost, hpath, dpath) == 0) {
1102                         ++CountLinkedItems;
1103                         if (VerboseOpt) {
1104                             logstd("%-32s hardlinked(-H)\n",
1105                                    (dpath ? dpath : spath));
1106                         }
1107                         free(hpath);
1108                         goto skip_copy;
1109                 }
1110                 /*
1111                  * Shucks, we may have hit a filesystem hard linking limit,
1112                  * we have to copy instead.
1113                  */
1114                 free(hpath);
1115         }
1116
1117         if ((fd1 = hc_open(&SrcHost, spath, O_RDONLY, 0)) >= 0) {
1118             if ((fd2 = hc_open(&DstHost, path, O_WRONLY|O_CREAT|O_EXCL, 0600)) < 0) {
1119                 /*
1120                  * There could be a .tmp file from a previously interrupted
1121                  * run, delete and retry.  Fail if we still can't get at it.
1122                  */
1123 #ifdef _ST_FLAGS_PRESENT_
1124                 hc_chflags(&DstHost, path, 0);
1125 #endif
1126                 hc_remove(&DstHost, path);
1127                 fd2 = hc_open(&DstHost, path, O_WRONLY|O_CREAT|O_EXCL|O_TRUNC, 0600);
1128             }
1129             if (fd2 >= 0) {
1130                 const char *op;
1131                 char *iobuf1 = malloc(GETIOSIZE);
1132                 int n;
1133
1134                 /*
1135                  * Matt: What about holes?
1136                  */
1137                 op = "read";
1138                 while ((n = hc_read(&SrcHost, fd1, iobuf1, GETIOSIZE)) > 0) {
1139                     op = "write";
1140                     if (hc_write(&DstHost, fd2, iobuf1, n) != n)
1141                         break;
1142                     op = "read";
1143                 }
1144                 hc_close(&DstHost, fd2);
1145                 if (n == 0) {
1146                     struct timeval tv[2];
1147
1148                     bzero(tv, sizeof(tv));
1149                     tv[0].tv_sec = st1.st_mtime;
1150                     tv[1].tv_sec = st1.st_mtime;
1151
1152                     if (DstRootPrivs || ChgrpAllowed(st1.st_gid))
1153                         hc_chown(&DstHost, path, st1.st_uid, st1.st_gid);
1154                     hc_chmod(&DstHost, path, st1.st_mode);
1155 #ifdef _ST_FLAGS_PRESENT_
1156                     if (st1.st_flags & (UF_IMMUTABLE|SF_IMMUTABLE))
1157                         hc_utimes(&DstHost, path, tv);
1158 #else
1159                     hc_utimes(&DstHost, path, tv);
1160 #endif
1161                     if (st2Valid && xrename(path, dpath, st2_flags) != 0) {
1162                         logerr("%-32s rename-after-copy failed: %s\n",
1163                             (dpath ? dpath : spath), strerror(errno)
1164                         );
1165                         ++r;
1166                     } else {
1167                         if (VerboseOpt)
1168                             logstd("%-32s copy-ok\n", (dpath ? dpath : spath));
1169 #ifdef _ST_FLAGS_PRESENT_
1170                         if (DstRootPrivs ? st1.st_flags : st1.st_flags & UF_SETTABLE)
1171                             hc_chflags(&DstHost, dpath, st1.st_flags);
1172 #endif
1173                     }
1174 #ifdef _ST_FLAGS_PRESENT_
1175                     if ((st1.st_flags & (UF_IMMUTABLE|SF_IMMUTABLE)) == 0)
1176                         hc_utimes(&DstHost, dpath, tv);
1177 #endif
1178                     CountSourceReadBytes += size;
1179                     CountWriteBytes += size;
1180                     CountSourceBytes += size;
1181                     CountSourceItems++;
1182                     CountCopiedItems++;
1183                 } else {
1184                     logerr("%-32s %s failed: %s\n",
1185                         (dpath ? dpath : spath), op, strerror(errno)
1186                     );
1187                     hc_remove(&DstHost, path);
1188                     ++r;
1189                 }
1190                 free(iobuf1);
1191             } else {
1192                 logerr("%-32s create (uid %d, euid %d) failed: %s\n",
1193                     (dpath ? dpath : spath), getuid(), geteuid(),
1194                     strerror(errno)
1195                 );
1196                 ++r;
1197             }
1198             hc_close(&SrcHost, fd1);
1199         } else {
1200             logerr("%-32s copy: open failed: %s\n",
1201                 (dpath ? dpath : spath),
1202                 strerror(errno)
1203             );
1204             ++r;
1205         }
1206 skip_copy:
1207         free(path);
1208
1209         if (hln) {
1210             if (!r && hc_stat(&DstHost, dpath, &st2) == 0) {
1211                 hltsetdino(hln, st2.st_ino);
1212             } else {
1213                 hltdelete(hln);
1214                 hln = NULL;
1215             }
1216         }
1217     } else if (S_ISLNK(st1.st_mode)) {
1218         char *link1 = malloc(GETLINKSIZE);
1219         char *link2 = malloc(GETLINKSIZE);
1220         char *path;
1221         int n1;
1222         int n2;
1223
1224         n1 = hc_readlink(&SrcHost, spath, link1, GETLINKSIZE - 1);
1225         if (st2Valid) {
1226                 path = mprintf("%s.tmp%d", dpath, (int)getpid());
1227                 n2 = hc_readlink(&DstHost, dpath, link2, GETLINKSIZE - 1);
1228         } else {
1229                 path = mprintf("%s", dpath);
1230                 n2 = -1;
1231         }
1232         if (n1 >= 0) {
1233             if (ForceOpt || n1 != n2 || bcmp(link1, link2, n1) != 0) {
1234                 hc_umask(&DstHost, ~st1.st_mode);
1235                 xremove(&DstHost, path);
1236                 link1[n1] = 0;
1237                 if (hc_symlink(&DstHost, link1, path) < 0) {
1238                       logerr("%-32s symlink (%s->%s) failed: %s\n",
1239                           (dpath ? dpath : spath), link1, path,
1240                           strerror(errno)
1241                       );
1242                       ++r;
1243                 } else {
1244                     if (DstRootPrivs || ChgrpAllowed(st1.st_gid))
1245                         hc_lchown(&DstHost, path, st1.st_uid, st1.st_gid);
1246                     /*
1247                      * there is no lchmod() or lchflags(), we
1248                      * cannot chmod or chflags a softlink.
1249                      */
1250                     if (st2Valid && xrename(path, dpath, st2_flags) != 0) {
1251                         logerr("%-32s rename softlink (%s->%s) failed: %s\n",
1252                             (dpath ? dpath : spath),
1253                             path, dpath, strerror(errno));
1254                     } else if (VerboseOpt) {
1255                         logstd("%-32s softlink-ok\n", (dpath ? dpath : spath));
1256                     }
1257                     hc_umask(&DstHost, 000);
1258                     CountWriteBytes += n1;
1259                     CountCopiedItems++;
1260                 }
1261             } else {
1262                 if (VerboseOpt >= 3)
1263                     logstd("%-32s nochange", (dpath ? dpath : spath));
1264                 if (!OwnerMatch(&st1, &st2)) {
1265                     hc_lchown(&DstHost, dpath, st1.st_uid, st1.st_gid);
1266                     if (VerboseOpt >= 3)
1267                         logstd(" (uid/gid differ)");
1268                 }
1269                 if (VerboseOpt >= 3)
1270                     logstd("\n");
1271             }
1272             CountSourceBytes += n1;
1273             CountSourceReadBytes += n1;
1274             if (n2 > 0)
1275                 CountTargetReadBytes += n2;
1276             CountSourceItems++;
1277         } else {
1278             r = 1;
1279             logerr("%-32s softlink-failed\n", (dpath ? dpath : spath));
1280         }
1281         free(link1);
1282         free(link2);
1283         free(path);
1284     } else if ((S_ISCHR(st1.st_mode) || S_ISBLK(st1.st_mode)) && DeviceOpt) {
1285         char *path = NULL;
1286
1287         if (ForceOpt ||
1288             st2Valid == 0 ||
1289             st1.st_mode != st2.st_mode ||
1290             st1.st_rdev != st2.st_rdev ||
1291             !OwnerMatch(&st1, &st2)
1292         ) {
1293             if (st2Valid) {
1294                 path = mprintf("%s.tmp%d", dpath, (int)getpid());
1295                 xremove(&DstHost, path);
1296             } else {
1297                 path = mprintf("%s", dpath);
1298             }
1299
1300             if (hc_mknod(&DstHost, path, st1.st_mode, st1.st_rdev) == 0) {
1301                 hc_chmod(&DstHost, path, st1.st_mode);
1302                 hc_chown(&DstHost, path, st1.st_uid, st1.st_gid);
1303                 if (st2Valid)
1304                         xremove(&DstHost, dpath);
1305                 if (st2Valid && xrename(path, dpath, st2_flags) != 0) {
1306                     logerr("%-32s dev-rename-after-create failed: %s\n",
1307                         (dpath ? dpath : spath),
1308                         strerror(errno)
1309                     );
1310                 } else if (VerboseOpt) {
1311                     logstd("%-32s dev-ok\n", (dpath ? dpath : spath));
1312                 }
1313                 CountCopiedItems++;
1314             } else {
1315                 r = 1;
1316                 logerr("%-32s dev failed: %s\n",
1317                     (dpath ? dpath : spath), strerror(errno)
1318                 );
1319             }
1320         } else {
1321             if (VerboseOpt >= 3)
1322                 logstd("%-32s nochange\n", (dpath ? dpath : spath));
1323         }
1324         if (path)
1325                 free(path);
1326         CountSourceItems++;
1327     }
1328 done:
1329     if (hln) {
1330         if (hln->dino == (ino_t)-1) {
1331             hltdelete(hln);
1332             /*hln = NULL; unneeded */
1333         } else {
1334             hltrels(hln);
1335         }
1336     }
1337     return (r);
1338 }
1339
1340 int
1341 ScanDir(List *list, struct HostConf *host, const char *path,
1342         int64_t *CountReadBytes, int n)
1343 {
1344     DIR *dir;
1345     struct dirent *den;
1346
1347     if ((dir = hc_opendir(host, path)) == NULL)
1348         return (1);
1349
1350     if (n == 0) {
1351         /*
1352          * scan .cpignore file for files/directories to ignore
1353          * (only in the source directory, i.e. if n == 0).
1354          */
1355         if (UseCpFile) {
1356             int fd;
1357             int nread;
1358             int bufused;
1359             char *buf = malloc(GETBUFSIZE);
1360             char *nl, *next;
1361             char *fpath;
1362
1363             if (UseCpFile[0] == '/') {
1364                 fpath = mprintf("%s", UseCpFile);
1365             } else {
1366                 fpath = mprintf("%s/%s", path, UseCpFile);
1367             }
1368             AddList(list, strrchr(fpath, '/') + 1, 1);
1369             if ((fd = hc_open(host, fpath, O_RDONLY, 0)) >= 0) {
1370                 bufused = 0;
1371                 while ((nread = hc_read(host, fd, buf + bufused,
1372                         GETIOSIZE - bufused - 1)) > 0) {
1373                     *CountReadBytes += nread;
1374                     bufused += nread;
1375                     buf[bufused] = 0;
1376                     for (next = buf; (nl = strchr(next, '\n')); next = nl+1) {
1377                         *nl = 0;
1378                         AddList(list, next, 1);
1379                     }
1380                     bufused = strlen(next);
1381                     if (bufused)
1382                         bcopy(next, buf, bufused);
1383                 }
1384                 if (bufused) {
1385                     /* last line has no trailing newline */
1386                     buf[bufused] = 0;
1387                     AddList(list, buf, 1);
1388                 }
1389                 hc_close(host, fd);
1390             }
1391             free(fpath);
1392             free(buf);
1393         }
1394
1395         /*
1396          * Automatically exclude MD5CacheFile that we create on the
1397          * source from the copy to the destination.
1398          *
1399          * Automatically exclude a FSMIDCacheFile on the source that
1400          * would otherwise overwrite the one we maintain on the target.
1401          */
1402         if (UseMD5Opt)
1403             AddList(list, MD5CacheFile, 1);
1404         if (UseFSMIDOpt)
1405             AddList(list, FSMIDCacheFile, 1);
1406     }
1407
1408     while ((den = hc_readdir(host, dir)) != NULL) {
1409         /*
1410          * ignore . and ..
1411          */
1412         if (strcmp(den->d_name, ".") != 0 && strcmp(den->d_name, "..") != 0)
1413              AddList(list, den->d_name, n);
1414     }
1415     hc_closedir(host, dir);
1416
1417     return (0);
1418 }
1419
1420 /*
1421  * RemoveRecur()
1422  */
1423
1424 void
1425 RemoveRecur(const char *dpath, dev_t devNo)
1426 {
1427     struct stat st;
1428
1429     if (hc_lstat(&DstHost, dpath, &st) == 0) {
1430         if (devNo == (dev_t)-1)
1431             devNo = st.st_dev;
1432         if (st.st_dev == devNo) {
1433             if (S_ISDIR(st.st_mode)) {
1434                 DIR *dir;
1435
1436                 if ((dir = hc_opendir(&DstHost, dpath)) != NULL) {
1437                     struct dirent *den;
1438                     while ((den = hc_readdir(&DstHost, dir)) != NULL) {
1439                         char *ndpath;
1440
1441                         if (strcmp(den->d_name, ".") == 0)
1442                             continue;
1443                         if (strcmp(den->d_name, "..") == 0)
1444                             continue;
1445                         ndpath = mprintf("%s/%s", dpath, den->d_name);
1446                         RemoveRecur(ndpath, devNo);
1447                         free(ndpath);
1448                     }
1449                     hc_closedir(&DstHost, dir);
1450                 }
1451                 if (AskConfirmation && NoRemoveOpt == 0) {
1452                     if (YesNo(dpath)) {
1453                         if (hc_rmdir(&DstHost, dpath) < 0) {
1454                             logerr("%-32s rmdir failed: %s\n",
1455                                 dpath, strerror(errno)
1456                             );
1457                         }
1458                         CountRemovedItems++;
1459                     }
1460                 } else {
1461                     if (NoRemoveOpt) {
1462                         if (VerboseOpt)
1463                             logstd("%-32s not-removed\n", dpath);
1464                     } else if (hc_rmdir(&DstHost, dpath) == 0) {
1465                         if (VerboseOpt)
1466                             logstd("%-32s rmdir-ok\n", dpath);
1467                         CountRemovedItems++;
1468                     } else {
1469                         logerr("%-32s rmdir failed: %s\n",
1470                             dpath, strerror(errno)
1471                         );
1472                     }
1473                 }
1474             } else {
1475                 if (AskConfirmation && NoRemoveOpt == 0) {
1476                     if (YesNo(dpath)) {
1477                         if (xremove(&DstHost, dpath) < 0) {
1478                             logerr("%-32s remove failed: %s\n",
1479                                 dpath, strerror(errno)
1480                             );
1481                         }
1482                         CountRemovedItems++;
1483                     }
1484                 } else {
1485                     if (NoRemoveOpt) {
1486                         if (VerboseOpt)
1487                             logstd("%-32s not-removed\n", dpath);
1488                     } else if (xremove(&DstHost, dpath) == 0) {
1489                         if (VerboseOpt)
1490                             logstd("%-32s remove-ok\n", dpath);
1491                         CountRemovedItems++;
1492                     } else {
1493                         logerr("%-32s remove failed: %s\n",
1494                             dpath, strerror(errno)
1495                         );
1496                     }
1497                 }
1498             }
1499         }
1500     }
1501 }
1502
1503 void
1504 InitList(List *list)
1505 {
1506     bzero(list, sizeof(List));
1507     list->li_Node.no_Next = &list->li_Node;
1508 }
1509
1510 void
1511 ResetList(List *list)
1512 {
1513     Node *node;
1514
1515     while ((node = list->li_Node.no_Next) != &list->li_Node) {
1516         list->li_Node.no_Next = node->no_Next;
1517         free(node);
1518     }
1519     InitList(list);
1520 }
1521
1522 char *
1523 IterateList(List *list, Node **nodeptr, int n)
1524 {
1525     Node *node = *nodeptr;
1526
1527     if (node == NULL)
1528         node = list->li_Node.no_Next;
1529     while (node != &list->li_Node) {
1530         if (node->no_Value == n) {
1531             *nodeptr = node->no_Next;
1532             return (node->no_Name);
1533         }
1534         node = node->no_Next;
1535     }
1536     return (NULL);
1537 }
1538
1539 int
1540 AddList(List *list, const char *name, int n)
1541 {
1542     Node *node;
1543     int hv;
1544
1545     hv = shash(name);
1546
1547     /*
1548      * Scan against wildcards.  Only a node value of 1 can be a wildcard
1549      * ( usually scanned from .cpignore )
1550      */
1551
1552     for (node = list->li_Hash[0]; node; node = node->no_HNext) {
1553         if (strcmp(name, node->no_Name) == 0 ||
1554             (n != 1 && node->no_Value == 1 &&
1555             fnmatch(node->no_Name, name, 0) == 0)
1556         ) {
1557             return(node->no_Value);
1558         }
1559     }
1560
1561     /*
1562      * Look for exact match
1563      */
1564
1565     for (node = list->li_Hash[hv]; node; node = node->no_HNext) {
1566         if (strcmp(name, node->no_Name) == 0) {
1567             return(node->no_Value);
1568         }
1569     }
1570     node = malloc(sizeof(Node) + strlen(name) + 1);
1571     if (node == NULL) {
1572         fprintf(stderr, "out of memory\n");
1573         exit(EXIT_FAILURE);
1574     }
1575
1576     node->no_Next = list->li_Node.no_Next;
1577     list->li_Node.no_Next = node;
1578
1579     node->no_HNext = list->li_Hash[hv];
1580     list->li_Hash[hv] = node;
1581
1582     strcpy(node->no_Name, name);
1583     node->no_Value = n;
1584
1585     return(n);
1586 }
1587
1588 static int
1589 shash(const char *s)
1590 {
1591     int hv;
1592
1593     hv = 0xA4FB3255;
1594
1595     while (*s) {
1596         if (*s == '*' || *s == '?' ||
1597             *s == '{' || *s == '}' ||
1598             *s == '[' || *s == ']' ||
1599             *s == '|'
1600         ) {
1601             return(0);
1602         }
1603         hv = (hv << 5) ^ *s ^ (hv >> 23);
1604         ++s;
1605     }
1606     return(((hv >> 16) ^ hv) & HMASK);
1607 }
1608
1609 int
1610 YesNo(const char *path)
1611 {
1612     int ch, first;
1613
1614     fprintf(stderr, "remove %s (Yes/No) [No]? ", path);
1615     fflush(stderr);
1616
1617     first = ch = getchar();
1618     while (ch != '\n' && ch != EOF)
1619         ch = getchar();
1620     return ((first == 'y' || first == 'Y'));
1621 }
1622
1623 /*
1624  * xrename() - rename with override
1625  *
1626  *      If the rename fails, attempt to override st_flags on the
1627  *      destination and rename again.  If that fails too, try to
1628  *      set the flags back the way they were and give up.
1629  */
1630
1631 static int
1632 xrename(const char *src, const char *dst, u_long flags)
1633 {
1634     int r;
1635
1636     if ((r = hc_rename(&DstHost, src, dst)) < 0) {
1637 #ifdef _ST_FLAGS_PRESENT_
1638         hc_chflags(&DstHost, dst, 0);
1639         if ((r = hc_rename(&DstHost, src, dst)) < 0)
1640                 hc_chflags(&DstHost, dst, flags);
1641 #endif
1642     }
1643     return(r);
1644 }
1645
1646 static int
1647 xlink(const char *src, const char *dst, u_long flags)
1648 {
1649     int r;
1650 #ifdef _ST_FLAGS_PRESENT_
1651     int e;
1652 #endif
1653
1654     if ((r = hc_link(&DstHost, src, dst)) < 0) {
1655 #ifdef _ST_FLAGS_PRESENT_
1656         hc_chflags(&DstHost, src, 0);
1657         r = hc_link(&DstHost, src, dst);
1658         e = errno;
1659         hc_chflags(&DstHost, src, flags);
1660         errno = e;
1661 #endif
1662     }
1663     if (r == 0)
1664             ++CountLinkedItems;
1665     return(r);
1666 }
1667
1668 static int
1669 xremove(struct HostConf *host, const char *path)
1670 {
1671     int res;
1672
1673     res = hc_remove(host, path);
1674 #ifdef _ST_FLAGS_PRESENT_
1675     if (res == -EPERM) {
1676         hc_chflags(host, path, 0);
1677         res = hc_remove(host, path);
1678     }
1679 #endif
1680     return(res);
1681 }
1682