X-Git-Url: https://mattmccutchen.net/rsync/rsync-patches.git/blobdiff_plain/d4e89c6a0706b85150f334b9ca6060cb7ee423df..c2f699cc6bfb241c97cc53dfe6bfd0ea0385456c:/link-by-hash.diff diff --git a/link-by-hash.diff b/link-by-hash.diff index cf70d05..ed8bf5b 100644 --- a/link-by-hash.diff +++ b/link-by-hash.diff @@ -1,30 +1,31 @@ -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 *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 +146,7 @@ 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 +211,10 @@ 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 +250,13 @@ 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 (file->length == 0) ++ return robust_rename(fnametmp, fname, NULL, 0644); + + if (do_stat(hashname, &st) == -1) { + char *dirname; @@ -266,29 +265,27 @@ 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 +295,16 @@ 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 +323,7 @@ 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 +336,7 @@ 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 +347,7 @@ 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)); @@ -367,49 +365,51 @@ the file's name. +} + +#endif ---- orig/options.c 2004-07-20 21:36:07 -+++ options.c 2004-07-03 20:20:15 -@@ -125,6 +125,7 @@ char *log_format = NULL; - char *password_file = NULL; - char *rsync_path = RSYNC_PATH; - char *backup_dir = NULL; +--- old/options.c ++++ new/options.c +@@ -145,6 +145,7 @@ 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; -@@ -276,6 +277,7 @@ 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; +@@ -349,6 +350,7 @@ 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"); -@@ -316,7 +318,7 @@ 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_TIMEOUT, -+ OPT_READ_BATCH, OPT_WRITE_BATCH, OPT_TIMEOUT, 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"); +@@ -398,7 +400,7 @@ 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[] = { -@@ -375,6 +377,7 @@ 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 }, -@@ -601,6 +604,19 @@ int parse_arguments(int *argc, const cha - return 0; - #endif +@@ -499,6 +501,7 @@ 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 }, +@@ -1089,6 +1092,21 @@ int parse_arguments(int *argc, const cha + usage(FINFO); + exit_cleanup(0); + case OPT_LINK_BY_HASH: +#if HAVE_LINK -+ link_by_hash_dir = (char *)poptGetOptArg(pc); -+ checksum_seed = FIXED_CHECKSUM_SEED; ++ 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, @@ -421,9 +421,9 @@ the file's name. + default: /* A large opt value means that set_refuse_options() - * turned this option off (opt-BASE is its index). */ -@@ -1007,6 +1023,11 @@ void server_options(char **args,int *arg - args[ac++] = compare_dest; + * turned this option off. */ +@@ -1739,6 +1757,11 @@ void server_options(char **args,int *arg + } } + if (link_by_hash_dir && am_sender) { @@ -431,20 +431,20 @@ the file's name. + args[ac++] = link_by_hash_dir; + } + - if (files_from && (!am_sender || remote_filesfrom_file)) { - if (remote_filesfrom_file) { + if (files_from && (!am_sender || filesfrom_host)) { + if (filesfrom_host) { args[ac++] = "--files-from"; ---- orig/receiver.c 2004-07-20 21:36:07 -+++ receiver.c 2004-07-20 21:44:05 -@@ -37,6 +37,7 @@ extern int cvs_exclude; - extern int io_error; +--- old/receiver.c ++++ new/receiver.c +@@ -50,6 +50,7 @@ extern int delay_updates; + extern struct stats stats; + extern char *stdout_format; extern char *tmpdir; - extern char *compare_dest; +extern char *link_by_hash_dir; - extern int make_backups; - extern int do_progress; - extern char *backup_dir; -@@ -200,12 +201,13 @@ static int get_tmpname(char *fnametmp, c + extern char *partial_dir; + extern char *basis_dir[]; + extern struct file_list *the_file_list; +@@ -124,12 +125,13 @@ static int get_tmpname(char *fnametmp, c static int receive_data(int f_in, char *fname_r, int fd_r, OFF_T size_r, @@ -456,10 +456,10 @@ the file's name. struct map_struct *mapbuf; struct sum_struct sum; + struct mdfour mdfour_data; - unsigned int len; + int32 len; OFF_T offset = 0; OFF_T offset2; -@@ -224,6 +226,9 @@ static int receive_data(int f_in, char * +@@ -149,6 +151,9 @@ static int receive_data(int f_in, char * } else mapbuf = NULL; @@ -468,26 +468,26 @@ the file's name. + sum_init(checksum_seed); - while ((i = recv_token(f_in, &data)) != 0) { -@@ -240,6 +245,8 @@ static int receive_data(int f_in, char * + if (append_mode) { +@@ -191,6 +196,8 @@ 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", -@@ -267,6 +274,8 @@ static int receive_data(int f_in, char * + if (fd != -1 && write_file(fd,data,i) != i) + goto report_write_error; +@@ -217,6 +224,8 @@ 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 (!inplace || offset != offset2) { -@@ -305,6 +314,8 @@ static int receive_data(int f_in, char * + if (updating_basis) { +@@ -259,6 +268,8 @@ static int receive_data(int f_in, char * } sum_end(file_sum1); @@ -496,7 +496,7 @@ the file's name. if (mapbuf) unmap_file(mapbuf); -@@ -320,7 +331,7 @@ static int receive_data(int f_in, char * +@@ -274,7 +285,7 @@ static int receive_data(int f_in, char * static void discard_receive_data(int f_in, OFF_T length) { @@ -504,47 +504,52 @@ the file's name. + receive_data(f_in, NULL, -1, 0, NULL, -1, length, NULL); } - -@@ -508,8 +519,12 @@ int recv_files(int f_in, struct file_lis + static void handle_delayed_updates(struct file_list *flist, char *local_name) +@@ -611,8 +622,12 @@ int recv_files(int f_in, struct file_lis rprintf(FINFO, "%s\n", fname); /* recv file data */ -+#ifdef HAVE_LINK ++#if HAVE_LINK + if (link_by_hash_dir) -+ file->u.sum = (char*)malloc(MD4_SUM_LENGTH); ++ file->u.sum = new_array(char, MD4_SUM_LENGTH); +#endif recv_ok = receive_data(f_in, fnamecmp, fd1, st.st_size, - fname, fd2, file->length); + fname, fd2, file->length, file->u.sum); - log_recv(file, &initial_stats); + log_item(log_code, file, &initial_stats, iflags, NULL); ---- orig/rsync.c 2004-07-16 20:07:23 -+++ rsync.c 2004-07-16 20:16:53 -@@ -33,6 +33,7 @@ extern int preserve_gid; - extern int force_delete; - extern int recurse; +--- old/rsync.c ++++ new/rsync.c +@@ -48,6 +48,7 @@ extern int inplace; + extern int keep_dirlinks; extern int make_backups; + extern mode_t orig_umask; +extern char *link_by_hash_dir; - extern char *backup_dir; - extern int inplace; + extern struct stats stats; + extern struct chmod_mode_struct *daemon_chmod_modes; -@@ -250,6 +251,12 @@ void finish_transfer(char *fname, char * +@@ -271,8 +272,15 @@ void finish_transfer(char *fname, char * /* move tmp file over real file */ if (verbose > 2) rprintf(FINFO, "renaming %s to %s\n", fnametmp, fname); -+#ifdef HAVE_LINK +- ret = robust_rename(fnametmp, fname, partialptr, +- file->mode & INITACCESSPERMS); ++#if HAVE_LINK + if (link_by_hash_dir) + ret = link_by_hash(fnametmp, fname, file); + else +#endif -+ ret = robust_rename(fnametmp, fname, file->mode & INITACCESSPERMS); - 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\"", ---- orig/rsync.h 2004-07-20 21:36:08 -+++ rsync.h 2004-07-03 20:20:15 -@@ -522,6 +522,14 @@ struct stats { + ret == -2 ? "copy" : "rename", +--- old/rsync.h ++++ new/rsync.h +@@ -651,6 +651,14 @@ struct stats { int current_file_index; }; @@ -556,6 +561,16 @@ 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 + #include "byteorder.h" +--- old/rsync.yo ++++ new/rsync.yo +@@ -366,6 +366,7 @@ 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