-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:
---- orig/Makefile.in 2004-05-15 00:53:53
-+++ Makefile.in 2004-07-03 20:20:15
-@@ -35,7 +35,7 @@ OBJS1=rsync.o generator.o receiver.o cle
- main.o checksum.o match.o syscall.o log.o backup.o
- OBJS2=options.o flist.o io.o compat.o hlink.o token.o uidlist.o socket.o \
- fileio.o batch.o clientname.o
+ patch -p1 <patches/link-by-hash.diff
+ ./prepare-source
+ ./configure
+ make
+
+--- old/Makefile.in
++++ new/Makefile.in
+@@ -35,7 +35,7 @@ OBJS1=flist.o rsync.o generator.o receiv
+ util.o main.o checksum.o match.o syscall.o log.o backup.o
+ OBJS2=options.o io.o compat.o hlink.o token.o uidlist.o socket.o \
+ fileio.o batch.o clientname.o chmod.o
-OBJS3=progress.o pipe.o
+OBJS3=progress.o pipe.o hashlink.o
DAEMON_OBJ = params.o loadparm.o clientserver.o access.o connection.o authenticate.o
popt_OBJS=popt/findme.o popt/popt.o popt/poptconfig.o \
popt/popthelp.o popt/poptparse.o
---- orig/hashlink.c 2004-07-02 21:41:54
-+++ hashlink.c 2004-07-02 21:41:54
-@@ -0,0 +1,342 @@
+--- old/flist.c
++++ new/flist.c
+@@ -59,6 +59,7 @@ extern int copy_unsafe_links;
+ extern int protocol_version;
+ extern int sanitize_paths;
+ extern struct stats stats;
++extern char *link_by_hash_dir;
+
+ extern char curr_dir[MAXPATHLEN];
+
+@@ -712,7 +713,7 @@ static struct file_struct *recv_file_ent
+ }
+ #endif
+
+- if (always_checksum && S_ISREG(mode))
++ if ((always_checksum || link_by_hash_dir) && S_ISREG(mode))
+ extra_len += SUM_EXTRA_CNT * EXTRA_LEN;
+
+ if (file_length > 0xFFFFFFFFu && S_ISREG(mode))
+--- old/hashlink.c
++++ new/hashlink.c
+@@ -0,0 +1,336 @@
+/*
+ Copyright (C) Cronosys, LLC 2004
+
+
+#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');
+ 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;
+ }
+ 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;
+ }
+ 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;
+ }
+ /* 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;
+ }
+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;
+ 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);
+ 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;
+ 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) {
+ } 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);
+ 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));
+ free(hashname);
+ return rc;
+}
-+
+#endif
---- orig/options.c 2004-06-20 19:30:00
-+++ options.c 2004-07-03 20:20:15
-@@ -124,6 +124,7 @@ char *log_format = NULL;
- char *password_file = NULL;
- char *rsync_path = RSYNC_PATH;
- char *backup_dir = NULL;
+--- old/options.c
++++ new/options.c
+@@ -144,6 +144,7 @@ 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;
-@@ -270,6 +271,7 @@ 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;
+@@ -346,6 +347,7 @@ 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");
-@@ -310,7 +312,7 @@ 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");
+@@ -395,7 +397,7 @@ 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[] = {
-@@ -368,6 +370,7 @@ 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 },
-@@ -596,6 +599,19 @@ int parse_arguments(int *argc, const cha
- return 0;
- #endif
+@@ -497,6 +499,7 @@ 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 },
+@@ -1093,6 +1096,21 @@ 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;
++#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,
+
default:
/* A large opt value means that set_refuse_options()
- * turned this option off (opt-BASE is its index). */
-@@ -977,6 +993,11 @@ void server_options(char **args,int *arg
- args[ac++] = compare_dest;
+ * turned this option off. */
+@@ -1755,6 +1773,11 @@ void server_options(char **args,int *arg
+ }
}
+ if (link_by_hash_dir && am_sender) {
+ 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 2004-07-02 18:23:01
-+++ receiver.c 2004-07-03 20:20:15
-@@ -48,6 +48,7 @@ extern int ignore_errors;
- extern int orig_umask;
- extern int keep_partial;
- extern int checksum_seed;
-+extern char *link_by_hash_dir;
-
- static void delete_one(char *fn, int is_dir)
- {
-@@ -196,10 +197,11 @@ static int get_tmpname(char *fnametmp, c
+--- old/receiver.c
++++ new/receiver.c
+@@ -124,12 +124,14 @@ 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,
+- 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[MD4_SUM_LENGTH];
+ static char file_sum2[MD4_SUM_LENGTH];
+ struct map_struct *mapbuf;
struct sum_struct sum;
+ struct mdfour mdfour_data;
- unsigned int len;
+ int32 len;
OFF_T offset = 0;
OFF_T offset2;
-@@ -210,6 +212,9 @@ static int receive_data(int f_in,struct
-
- read_sum_head(f_in, &sum);
+@@ -149,6 +151,9 @@ static int receive_data(int f_in, char *
+ } else
+ mapbuf = NULL;
+ if (md4)
+ mdfour_begin(&mdfour_data);
+
sum_init(checksum_seed);
- while ((i = recv_token(f_in, &data)) != 0) {
-@@ -226,6 +231,8 @@ static int receive_data(int f_in,struct
+ if (append_mode > 0) {
+@@ -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",
-@@ -253,6 +260,8 @@ static int receive_data(int f_in,struct
+ 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 (fd != -1 && write_file(fd, map, len) != (int)len) {
-@@ -275,6 +284,8 @@ static int receive_data(int f_in,struct
+ if (updating_basis) {
+@@ -259,6 +268,8 @@ static int receive_data(int f_in, char *
}
sum_end(file_sum1);
+ if (md4)
+ mdfour_result(&mdfour_data, (unsigned char*)md4);
- read_buf(f_in,file_sum2,MD4_SUM_LENGTH);
- if (verbose > 2)
-@@ -374,7 +385,7 @@ 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 +398,7 @@ 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;
- }
-@@ -443,7 +454,7 @@ 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)
-@@ -457,7 +468,11 @@ int recv_files(int f_in,struct file_list
- rprintf(FINFO, "%s\n", fname);
+ if (mapbuf)
+ unmap_file(mapbuf);
+@@ -274,7 +285,7 @@ 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)
+@@ -621,7 +632,7 @@ 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);
---- orig/rsync.c 2004-07-02 18:06:32
-+++ rsync.c 2004-07-03 20:20:15
-@@ -34,6 +34,7 @@ extern int force_delete;
- extern int recurse;
+--- old/rsync.c
++++ new/rsync.c
+@@ -51,6 +51,7 @@ extern int flist_eof;
+ extern int keep_dirlinks;
extern int make_backups;
- extern char *backup_dir;
+ extern mode_t orig_umask;
+extern char *link_by_hash_dir;
-
-
- /*
-@@ -242,6 +243,12 @@ void finish_transfer(char *fname, char *
+ extern struct stats stats;
+ extern struct file_list *cur_flist, *first_flist, *dir_flist;
+ extern struct chmod_mode_struct *daemon_chmod_modes;
+@@ -374,8 +375,15 @@ void finish_transfer(char *fname, char *
/* move tmp file over real file */
if (verbose > 2)
rprintf(FINFO, "renaming %s to %s\n", fnametmp, fname);
+- ret = robust_rename(fnametmp, fname, partialptr,
+- file->mode & INITACCESSPERMS);
+#ifdef 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 2004-05-16 07:58:12
-+++ rsync.h 2004-07-03 20:20:15
-@@ -522,6 +522,14 @@ struct stats {
+ ret == -2 ? "copy" : "rename",
+--- old/rsync.h
++++ new/rsync.h
+@@ -750,6 +750,14 @@ struct stats {
int current_file_index;
};
+ 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
+@@ -367,6 +367,7 @@ 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