X-Git-Url: https://mattmccutchen.net/rsync/rsync-patches.git/blobdiff_plain/fe6407b5a9f04c88f74ea95e959aaa38334e41a6..d705d4dd33c2cd5a83d772df98c7b3ff5319af34:/link-by-hash.diff diff --git a/link-by-hash.diff b/link-by-hash.diff index daffd6f..5d72765 100644 --- a/link-by-hash.diff +++ b/link-by-hash.diff @@ -1,30 +1,50 @@ -After applying this patch and running configure, you MUST run this -command before "make": - - make proto - -Jason M. Felice writes: +Jason M. Felice wrote: This patch adds the --link-by-hash=DIR option, which hard links received files in a link farm arranged by MD4 file hash. The result is that the system will only store one copy of the unique contents of each file, regardless of the file's name. +To use this patch, run these commands for a successful build: + + patch -p1 0xFFFFFFFFu && S_ISREG(mode)) +--- old/hashlink.c ++++ new/hashlink.c +@@ -0,0 +1,336 @@ +/* + Copyright (C) Cronosys, LLC 2004 + @@ -51,14 +71,12 @@ the file's name. + +#ifdef HAVE_LINK + -+char* make_hash_name(struct file_struct *file) ++char *make_hash_name(struct file_struct *file) +{ + char hash[33], *dst; -+ unsigned char *src; -+ unsigned char c; ++ uchar c, *src = (uchar*)F_SUM(file); + int i; + -+ src = (unsigned char*)file->u.sum; + for (dst = hash, i = 0; i < 4; i++, src++) { + c = *src >> 4; + *(dst++) = (c >= 10) ? (c - 10 + 'a') : (c + '0'); @@ -111,11 +129,11 @@ the file's name. + long this_fnbr; + + *fnbr = 0; -+ ++ + /* Build a list of potential candidates and open + * them. */ + if ((d = opendir(hashname)) == NULL) { -+ rsyserr(FERROR, errno, "opendir \"%s\"", hashname); ++ rsyserr(FERROR, errno, "opendir failed: \"%s\"", hashname); + free(hashname); + return NULL; + } @@ -130,11 +148,11 @@ the file's name. + if (this_fnbr > *fnbr) + *fnbr = this_fnbr; + -+ hashfile = (struct hashfile_struct*)malloc(sizeof(struct hashfile_struct)); ++ hashfile = new_array(struct hashfile_struct, 1); + asprintf(&hashfile->name,"%s/%s",hashname, + di->d_name); + if (do_stat(hashfile->name,&st) == -1) { -+ rsyserr(FERROR, errno, "%s: %s", hashfile->name); ++ rsyserr(FERROR, errno, "stat failed: %s", hashfile->name); + kill_hashfile(hashfile); + continue; + } @@ -145,7 +163,7 @@ the file's name. + hashfile->nlink = st.st_nlink; + hashfile->fd = open(hashfile->name,O_RDONLY|O_BINARY); + if (hashfile->fd == -1) { -+ rsyserr(FERROR, errno, "%s", hashfile->name); ++ rsyserr(FERROR, errno, "open failed: %s", hashfile->name); + kill_hashfile(hashfile); + continue; + } @@ -210,11 +228,10 @@ the file's name. + /* There are no matches. */ + return NULL; + } -+ + } + + if (amt == -1) { -+ rsyserr(FERROR, errno, "%s"); ++ rsyserr(FERROR, errno, "read failed in compare_hashfiles()"); + kill_hashfiles(files); + return NULL; + } @@ -250,14 +267,13 @@ the file's name. +int link_by_hash(char *fnametmp,char *fname,struct file_struct *file) +{ + STRUCT_STAT st; -+ char *hashname = make_hash_name(file); ++ char *hashname = make_hash_name(file); + int first = 0, rc; + char *linkname; + long last_fnbr; + -+ if (file->length == 0) { -+ return robust_rename(fnametmp,fname,0644); -+ } ++ if (F_LENGTH(file) == 0) ++ return robust_rename(fnametmp, fname, NULL, 0644); + + if (do_stat(hashname, &st) == -1) { + char *dirname; @@ -266,29 +282,27 @@ the file's name. + dirname = strdup(hashname); + *strrchr(dirname,'/') = 0; + if (do_mkdir(dirname, 0755) == -1 && errno != EEXIST) { -+ rsyserr(FERROR, errno, "mkdir %s", dirname); ++ rsyserr(FERROR, errno, "mkdir failed: %s", dirname); + free(hashname); + free(dirname); -+ return robust_rename(fnametmp,fname,0644); ++ return robust_rename(fnametmp, fname, NULL, 0644); + } + free(dirname); + + if (do_mkdir(hashname, 0755) == -1 && errno != EEXIST) { -+ rsyserr(FERROR, errno, "mkdir %s", hashname); ++ rsyserr(FERROR, errno, "mkdir failed: %s", hashname); + free(hashname); -+ return robust_rename(fnametmp,fname,0644); ++ return robust_rename(fnametmp, fname, NULL, 0644); + } + + first = 1; + asprintf(&linkname,"%s/0",hashname); + rprintf(FINFO, "(1) linkname = %s\n", linkname); -+ + } else { + struct hashfile_struct *hashfiles, *hashfile; -+ int fd; + + if (do_stat(fnametmp,&st) == -1) { -+ rsyserr(FERROR, errno, "%s", fname); ++ rsyserr(FERROR, errno, "stat failed: %s", fname); + return -1; + } + hashfiles = find_hashfiles(hashname, st.st_size, &last_fnbr); @@ -298,15 +312,16 @@ the file's name. + asprintf(&linkname,"%s/0",hashname); + rprintf(FINFO, "(2) linkname = %s\n", linkname); + } else { -+ ++ int fd; + /* Search for one identical to us. */ + if ((fd = open(fnametmp,O_RDONLY|O_BINARY)) == -1) { -+ rsyserr(FERROR, errno, "%s", fnametmp); ++ rsyserr(FERROR, errno, "open failed: %s", fnametmp); + kill_hashfiles(hashfiles); + return -1; + } + hashfile = compare_hashfiles(fd, hashfiles); + hashfiles = NULL; ++ close(fd); + + if (hashfile) { + first = 0; @@ -325,6 +340,7 @@ the file's name. + if (!first) { + rprintf(FINFO, "link-by-hash (existing): \"%s\" -> %s\n", + linkname, full_fname(fname)); ++ robust_unlink(fname); + rc = do_link(linkname, fname); + if (rc == -1) { + if (errno == EMLINK) { @@ -337,8 +353,7 @@ the file's name. + } else { + rsyserr(FERROR, errno, "link \"%s\" -> \"%s\"", + linkname, full_fname(fname)); -+ robust_unlink(fname); -+ rc = robust_rename(fnametmp,fname,0644); ++ rc = robust_rename(fnametmp, fname, NULL, 0644); + } + } else { + do_unlink(fnametmp); @@ -349,7 +364,7 @@ the file's name. + rprintf(FINFO, "link-by-hash (new): %s -> \"%s\"\n", + full_fname(fname),linkname); + -+ rc = robust_rename(fnametmp,fname,0644); ++ rc = robust_rename(fnametmp, fname, NULL, 0644); + if (rc != 0) { + rsyserr(FERROR, errno, "rename \"%s\" -> \"%s\"", + full_fname(fnametmp), full_fname(fname)); @@ -365,51 +380,52 @@ the file's name. + free(hashname); + return rc; +} -+ +#endif ---- options.c 6 May 2004 21:08:01 -0000 1.148 -+++ options.c 15 May 2004 20:29:17 -0000 -@@ -121,6 +121,7 @@ char *log_format = NULL; - char *password_file = NULL; - char *rsync_path = RSYNC_PATH; - char *backup_dir = NULL; +--- old/options.c ++++ new/options.c +@@ -146,6 +146,7 @@ char *backup_suffix = NULL; + char *tmpdir = NULL; + char *partial_dir = NULL; + char *basis_dir[MAX_BASIS_DIRS+1]; +char *link_by_hash_dir = NULL; - char backup_dir_buf[MAXPATHLEN]; - int rsync_port = RSYNC_PORT; - int link_dest = 0; -@@ -266,6 +267,7 @@ void usage(enum logcode F) - rprintf(F," -T --temp-dir=DIR create temporary files in directory DIR\n"); + char *config_file = NULL; + char *shell_cmd = NULL; + char *logfile_name = NULL; +@@ -362,6 +363,7 @@ void usage(enum logcode F) rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n"); - rprintf(F," --link-dest=DIR create hardlinks to DIR for unchanged files\n"); -+ rprintf(F," --link-by-hash=DIR create hardlinks by hash to DIR for regular files\n"); - rprintf(F," -P equivalent to --partial --progress\n"); - rprintf(F," -z, --compress compress file data\n"); - rprintf(F," -C, --cvs-exclude auto ignore files in the same way CVS does\n"); -@@ -305,7 +307,7 @@ void usage(enum logcode F) - enum {OPT_VERSION = 1000, OPT_SENDER, OPT_EXCLUDE, OPT_EXCLUDE_FROM, - OPT_DELETE_AFTER, OPT_DELETE_EXCLUDED, OPT_LINK_DEST, - OPT_INCLUDE, OPT_INCLUDE_FROM, OPT_MODIFY_WINDOW, -- OPT_READ_BATCH, OPT_WRITE_BATCH, -+ OPT_READ_BATCH, OPT_WRITE_BATCH, OPT_LINK_BY_HASH, - OPT_REFUSED_BASE = 9000}; + rprintf(F," --copy-dest=DIR ... and include copies of unchanged files\n"); + rprintf(F," --link-dest=DIR hardlink to files in DIR when unchanged\n"); ++ rprintf(F," --link-by-hash=DIR create hardlinks by hash into DIR\n"); + rprintf(F," -z, --compress compress file data during the transfer\n"); + rprintf(F," --compress-level=NUM explicitly set compression level\n"); + rprintf(F," -C, --cvs-exclude auto-ignore files the same way CVS does\n"); +@@ -411,7 +413,7 @@ enum {OPT_VERSION = 1000, OPT_DAEMON, OP + OPT_FILTER, OPT_COMPARE_DEST, OPT_COPY_DEST, OPT_LINK_DEST, OPT_HELP, + OPT_INCLUDE, OPT_INCLUDE_FROM, OPT_MODIFY_WINDOW, OPT_MIN_SIZE, OPT_CHMOD, + OPT_READ_BATCH, OPT_WRITE_BATCH, OPT_ONLY_WRITE_BATCH, OPT_MAX_SIZE, +- OPT_NO_D, ++ OPT_NO_D, OPT_LINK_BY_HASH, + OPT_SERVER, OPT_REFUSED_BASE = 9000}; static struct poptOption long_options[] = { -@@ -362,6 +364,7 @@ static struct poptOption long_options[] - {"temp-dir", 'T', POPT_ARG_STRING, &tmpdir, 0, 0, 0 }, - {"compare-dest", 0, POPT_ARG_STRING, &compare_dest, 0, 0, 0 }, - {"link-dest", 0, POPT_ARG_STRING, &compare_dest, OPT_LINK_DEST, 0, 0 }, -+ {"link-by-hash", 0, POPT_ARG_STRING, 0, OPT_LINK_BY_HASH, 0, 0}, - /* TODO: Should this take an optional int giving the compression level? */ - {"compress", 'z', POPT_ARG_NONE, &do_compression, 0, 0, 0 }, - {"daemon", 0, POPT_ARG_NONE, &daemon_opt, 0, 0, 0 }, -@@ -584,6 +587,19 @@ int parse_arguments(int *argc, const cha +@@ -523,6 +525,7 @@ static struct poptOption long_options[] + {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 }, + {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 }, + {"link-dest", 0, POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 }, ++ {"link-by-hash", 0, POPT_ARG_STRING, 0, OPT_LINK_BY_HASH, 0, 0}, + {"fuzzy", 'y', POPT_ARG_NONE, &fuzzy_basis, 0, 0, 0 }, + {"compress", 'z', POPT_ARG_NONE, 0, 'z', 0, 0 }, + {"compress-level", 0, POPT_ARG_INT, &def_compress_level, 'z', 0, 0 }, +@@ -1148,6 +1151,21 @@ int parse_arguments(int *argc, const cha return 0; #endif + case OPT_LINK_BY_HASH: -+#if HAVE_LINK -+ link_by_hash_dir = (char *)poptGetOptArg(pc); -+ checksum_seed = FIXED_CHECKSUM_SEED; ++#ifdef HAVE_LINK ++ arg = poptGetOptArg(pc); ++ if (sanitize_paths) ++ arg = sanitize_path(NULL, arg, NULL, 0, NULL); ++ link_by_hash_dir = (char *)arg; + break; +#else + snprintf(err_buf, sizeof err_buf, @@ -421,149 +437,123 @@ the file's name. + default: /* A large opt value means that set_refuse_options() - * turned this option off (opt-BASE is its index). */ -@@ -951,6 +967,11 @@ void server_options(char **args,int *arg - */ - args[ac++] = link_dest ? "--link-dest" : "--compare-dest"; - args[ac++] = compare_dest; -+ } -+ + * turned this option off. */ +@@ -1814,6 +1832,11 @@ void server_options(char **args,int *arg + } + } + + if (link_by_hash_dir && am_sender) { + args[ac++] = "--link-by-hash"; + args[ac++] = link_by_hash_dir; - } - - if (files_from && (!am_sender || remote_filesfrom_file)) { ---- receiver.c 15 May 2004 19:31:10 -0000 1.78 -+++ receiver.c 15 May 2004 20:29:17 -0000 -@@ -46,6 +46,7 @@ extern int module_id; - extern int ignore_errors; - extern int orig_umask; - extern int keep_partial; -+extern char *link_by_hash_dir; - - static void delete_one(char *fn, int is_dir) - { -@@ -192,10 +193,11 @@ static int get_tmpname(char *fnametmp, c ++ } ++ + if (files_from && (!am_sender || filesfrom_host)) { + if (filesfrom_host) { + args[ac++] = "--files-from"; +--- old/receiver.c ++++ new/receiver.c +@@ -125,12 +125,14 @@ int get_tmpname(char *fnametmp, char *fn - static int receive_data(int f_in,struct map_struct *mapbuf,int fd,char *fname, -- OFF_T total_size) -+ OFF_T total_size,char *md4) + static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r, +- const char *fname, int fd, OFF_T total_size) ++ const char *fname, int fd, OFF_T total_size, ++ const char *md4) { - int i; + static char file_sum1[MAX_DIGEST_LEN]; + static char file_sum2[MAX_DIGEST_LEN]; + struct map_struct *mapbuf; struct sum_struct sum; -+ struct mdfour mdfour_data; - unsigned int len; ++ md_context mdfour_data; + int32 len, sum_len; OFF_T offset = 0; OFF_T offset2; -@@ -205,7 +207,9 @@ static int receive_data(int f_in,struct - char *map=NULL; +@@ -150,6 +152,9 @@ static int receive_data(int f_in, char * + } else + mapbuf = NULL; - read_sum_head(f_in, &sum); -- + if (md4) + mdfour_begin(&mdfour_data); -+ - sum_init(); ++ + sum_init(checksum_seed); - while ((i = recv_token(f_in, &data)) != 0) { -@@ -222,6 +226,8 @@ static int receive_data(int f_in,struct + if (append_mode > 0) { +@@ -192,6 +197,8 @@ static int receive_data(int f_in, char * cleanup_got_literal = 1; - sum_update(data,i); + sum_update(data, i); + if (md4) -+ mdfour_update(&mdfour_data,data,i); ++ mdfour_update(&mdfour_data, (uchar*)data, i); - if (fd != -1 && write_file(fd,data,i) != i) { - rsyserr(FERROR, errno, "write failed on %s", -@@ -249,6 +255,8 @@ static int receive_data(int f_in,struct + if (fd != -1 && write_file(fd,data,i) != i) + goto report_write_error; +@@ -218,6 +225,8 @@ static int receive_data(int f_in, char * see_token(map, len); - sum_update(map,len); + sum_update(map, len); + if (md4) -+ mdfour_update(&mdfour_data,map,len); ++ mdfour_update(&mdfour_data, (uchar*)map, len); } - if (fd != -1 && write_file(fd,map,len) != (int) len) { -@@ -271,6 +279,8 @@ static int receive_data(int f_in,struct + if (updating_basis) { +@@ -260,6 +269,8 @@ static int receive_data(int f_in, char * } - sum_end(file_sum1); + sum_len = sum_end(file_sum1); + if (md4) -+ mdfour_result(&mdfour_data, (unsigned char*)md4); ++ mdfour_result(&mdfour_data, (uchar*)md4); - read_buf(f_in,file_sum2,MD4_SUM_LENGTH); - if (verbose > 2) { -@@ -374,7 +384,7 @@ int recv_files(int f_in,struct file_list - if (fd1 != -1 && do_fstat(fd1,&st) != 0) { - rsyserr(FERROR, errno, "fstat %s failed", - full_fname(fnamecmp)); -- receive_data(f_in,NULL,-1,NULL,file->length); -+ receive_data(f_in,NULL,-1,NULL,file->length,NULL); - close(fd1); - continue; - } -@@ -387,7 +397,7 @@ int recv_files(int f_in,struct file_list - */ - rprintf(FERROR,"recv_files: %s is a directory\n", - full_fname(fnamecmp)); -- receive_data(f_in, NULL, -1, NULL, file->length); -+ receive_data(f_in,NULL,-1,NULL,file->length,NULL); - close(fd1); - continue; - } -@@ -439,7 +449,7 @@ int recv_files(int f_in,struct file_list - if (fd2 == -1) { - rsyserr(FERROR, errno, "mkstemp %s failed", - full_fname(fnametmp)); -- receive_data(f_in,mapbuf,-1,NULL,file->length); -+ receive_data(f_in,mapbuf,-1,NULL,file->length,NULL); - if (mapbuf) unmap_file(mapbuf); - if (fd1 != -1) close(fd1); - continue; -@@ -452,7 +462,12 @@ int recv_files(int f_in,struct file_list - } + if (mapbuf) + unmap_file(mapbuf); +@@ -275,7 +286,7 @@ static int receive_data(int f_in, char * + + static void discard_receive_data(int f_in, OFF_T length) + { +- receive_data(f_in, NULL, -1, 0, NULL, -1, length); ++ receive_data(f_in, NULL, -1, 0, NULL, -1, length, NULL); + } + + static void handle_delayed_updates(char *local_name) +@@ -646,7 +657,7 @@ int recv_files(int f_in, char *local_nam /* recv file data */ -- recv_ok = receive_data(f_in,mapbuf,fd2,fname,file->length); -+#ifdef HAVE_LINK -+ if (link_by_hash_dir) { -+ file->u.sum = (char*)malloc (MD4_SUM_LENGTH); -+ } -+#endif -+ recv_ok = receive_data(f_in,mapbuf,fd2,fname,file->length,file->u.sum); + recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size, +- fname, fd2, F_LENGTH(file)); ++ fname, fd2, F_LENGTH(file), F_SUM(file)); - log_recv(file, &initial_stats); + log_item(log_code, file, &initial_stats, iflags, NULL); ---- rsync.c 15 May 2004 19:31:10 -0000 1.139 -+++ rsync.c 15 May 2004 20:29:17 -0000 -@@ -31,6 +31,7 @@ extern int am_generator; - extern int preserve_uid; - extern int preserve_gid; +--- old/rsync.c ++++ new/rsync.c +@@ -51,6 +51,7 @@ extern int inplace; + extern int flist_eof; + extern int keep_dirlinks; extern int make_backups; +extern char *link_by_hash_dir; + extern struct file_list *cur_flist, *first_flist, *dir_flist; + extern struct chmod_mode_struct *daemon_chmod_modes; - - /* -@@ -236,8 +237,12 @@ void finish_transfer(char *fname, char * - if (make_backups && !make_backup(fname)) - return; - -- /* move tmp file over real file */ -- ret = robust_rename(fnametmp, fname, file->mode & INITACCESSPERMS); +@@ -410,8 +411,15 @@ void finish_transfer(const char *fname, + /* move tmp file over real file */ + if (verbose > 2) + rprintf(FINFO, "renaming %s to %s\n", fnametmp, fname); +- ret = robust_rename(fnametmp, fname, partialptr, +- file->mode & INITACCESSPERMS); +#ifdef HAVE_LINK + if (link_by_hash_dir) -+ ret = link_by_hash(fnametmp,fname,file); ++ ret = link_by_hash(fnametmp, fname, file); + else +#endif -+ ret = robust_rename(fnametmp, fname, file->mode & INITACCESSPERMS); ++ { ++ ret = robust_rename(fnametmp, fname, partialptr, ++ file->mode & INITACCESSPERMS); ++ } if (ret < 0) { rsyserr(FERROR, errno, "%s %s -> \"%s\"", - ret == -2 ? "copy" : "rename", ---- rsync.h 13 May 2004 18:51:22 -0000 1.203 -+++ rsync.h 15 May 2004 20:29:18 -0000 -@@ -521,6 +521,14 @@ struct stats { + ret == -2 ? "copy" : "rename", +--- old/rsync.h ++++ new/rsync.h +@@ -758,6 +758,14 @@ struct stats { int current_file_index; }; @@ -575,6 +565,16 @@ the file's name. + uint32 nlink; +}; + + struct chmod_mode_struct; - /* we need this function because of the silly way in which duplicate - entries are handled in the file lists - we can't change this + #define EMPTY_ITEM_LIST {NULL, 0, 0} +--- old/rsync.yo ++++ new/rsync.yo +@@ -369,6 +369,7 @@ to the detailed description below for a + --compare-dest=DIR also compare received files relative to DIR + --copy-dest=DIR ... and include copies of unchanged files + --link-dest=DIR hardlink to files in DIR when unchanged ++ --link-by-hash=DIR create hardlinks by hash into DIR + -z, --compress compress file data during the transfer + --compress-level=NUM explicitly set compression level + -C, --cvs-exclude auto-ignore files in the same way CVS does