[rsync/rsync-patches.git] / g2r-basis-filename.diff

Added a pipe from the generator to the receiver that communicates
what basis file we used to generate the file data (if it was not
the default name).  This optimizes away the basis-file search in
the receiver and makes future options that do more basis-file
searching more efficient (such as the --fuzzy option and the
support for multiple --compare-dest options).

You must run "make proto" before compiling.

--- orig/generator.c	2004-07-16 19:35:29
+++ generator.c	2004-07-03 20:08:07
@@ -251,7 +251,7 @@ static void generate_and_send_sums(struc
  * out.  It might be wrong.
  */
 static void recv_generator(char *fname, struct file_struct *file, int i,
-			   int f_out)
+			   int f_out, int f_nameout)
 {
 	int fd;
 	STRUCT_STAT st;
@@ -418,8 +418,10 @@ static void recv_generator(char *fname, 
 		statret = link_stat(fnamecmpbuf, &st, 0);
 		if (!S_ISREG(st.st_mode))
 			statret = -1;
-		if (statret == -1)
+		if (statret < 0) {
 			errno = saveerrno;
+			*fnamecmpbuf = '\0';
+		}
 #if HAVE_LINK
 		else if (link_dest && !dry_run) {
 			if (do_link(fnamecmpbuf, fname) != 0) {
@@ -427,18 +429,22 @@ static void recv_generator(char *fname, 
 					rsyserr(FINFO, errno, "link %s => %s",
 						fnamecmpbuf, fname);
 				}
-			}
-			fnamecmp = fnamecmpbuf;
+				fnamecmp = fnamecmpbuf;
+			} else
+				*fnamecmpbuf = '\0';
 		}
 #endif
 		else
 			fnamecmp = fnamecmpbuf;
-	}
+	} else
+		*fnamecmpbuf = '\0';
 
 	if (statret == -1) {
 		if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 			return;
 		if (errno == ENOENT) {
+			if (f_nameout >= 0)
+				write(f_nameout, "", 1);
 			write_int(f_out,i);
 			if (!dry_run)
 				write_sum_head(f_out, NULL);
@@ -458,19 +464,21 @@ static void recv_generator(char *fname, 
 		/* now pretend the file didn't exist */
 		if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 			return;
+		if (f_nameout >= 0)
+			write(f_nameout, "", 1);
 		write_int(f_out,i);
 		if (!dry_run)
 			write_sum_head(f_out, NULL);
 		return;
 	}
 
-	if (opt_ignore_existing && fnamecmp == fname) {
+	if (opt_ignore_existing && !*fnamecmpbuf) {
 		if (verbose > 1)
 			rprintf(FINFO,"%s exists\n",fname);
 		return;
 	}
 
-	if (update_only && fnamecmp == fname
+	if (update_only && !*fnamecmpbuf
 	    && cmp_modtime(st.st_mtime, file->modtime) > 0) {
 		if (verbose > 1)
 			rprintf(FINFO,"%s is newer\n",fname);
@@ -478,17 +486,21 @@ static void recv_generator(char *fname, 
 	}
 
 	if (skip_file(fname, file, &st)) {
-		if (fnamecmp == fname)
+		if (!*fnamecmpbuf)
 			set_perms(fname, file, &st, PERMS_REPORT);
 		return;
 	}
 
 	if (dry_run) {
+		if (f_nameout >= 0)
+			write(f_nameout, "", 1);
 		write_int(f_out,i);
 		return;
 	}
 
 	if (whole_file > 0) {
+		if (f_nameout >= 0)
+			write(f_nameout, "", 1);
 		write_int(f_out,i);
 		write_sum_head(f_out, NULL);
 		return;
@@ -503,6 +515,8 @@ static void recv_generator(char *fname, 
 		/* pretend the file didn't exist */
 		if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 			return;
+		if (f_nameout >= 0)
+			write(f_nameout, "", 1);
 		write_int(f_out,i);
 		write_sum_head(f_out, NULL);
 		return;
@@ -521,6 +535,8 @@ static void recv_generator(char *fname, 
 	if (verbose > 2)
 		rprintf(FINFO, "generating and sending sums for %d\n", i);
 
+	if (f_nameout >= 0)
+		write(f_nameout, fnamecmpbuf, strlen(fnamecmpbuf) + 1);
 	write_int(f_out,i);
 	generate_and_send_sums(mapbuf, st.st_size, f_out);
 
@@ -530,7 +546,8 @@ static void recv_generator(char *fname, 
 }
 
 
-void generate_files(int f_out, struct file_list *flist, char *local_name)
+void generate_files(int f_out, struct file_list *flist, char *local_name,
+		    int f_nameout)
 {
 	int i;
 	int phase = 0;
@@ -571,7 +588,7 @@ void generate_files(int f_out, struct fi
 		}
 
 		recv_generator(local_name ? local_name : f_name_to(file, fbuf),
-			       file, i, f_out);
+			       file, i, f_out, f_nameout);
 	}
 
 	phase++;
@@ -588,7 +605,7 @@ void generate_files(int f_out, struct fi
 	while ((i = get_redo_num()) != -1) {
 		struct file_struct *file = flist->files[i];
 		recv_generator(local_name ? local_name : f_name_to(file, fbuf),
-			       file, i, f_out);
+			       file, i, f_out, f_nameout);
 	}
 
 	phase++;
@@ -607,7 +624,7 @@ void generate_files(int f_out, struct fi
 		if (!file->basename || !S_ISDIR(file->mode))
 			continue;
 		recv_generator(local_name ? local_name : f_name(file),
-			       file, i, -1);
+			       file, i, -1, -1);
 	}
 
 	if (verbose > 2)
--- orig/main.c	2004-07-15 17:02:03
+++ main.c	2004-07-03 20:08:07
@@ -443,7 +443,7 @@ static int do_recv(int f_in,int f_out,st
 {
 	int pid;
 	int status = 0;
-	int error_pipe[2];
+	int error_pipe[2], name_pipe[2];
 
 	if (preserve_hard_links)
 		init_hard_links(flist);
@@ -455,8 +455,8 @@ static int do_recv(int f_in,int f_out,st
 		}
 	}
 
-	if (fd_pair(error_pipe) < 0) {
-		rprintf(FERROR,"error pipe failed in do_recv\n");
+	if (fd_pair(error_pipe) < 0 || fd_pair(name_pipe) < 0) {
+		rprintf(FERROR, "fd_pair() failed in do_recv\n");
 		exit_cleanup(RERR_SOCKETIO);
 	}
 
@@ -464,8 +464,10 @@ static int do_recv(int f_in,int f_out,st
 
 	if ((pid = do_fork()) == 0) {
 		close(error_pipe[0]);
+		close(name_pipe[1]);
 		if (f_in != f_out)
 			close(f_out);
+		set_blocking(name_pipe[0]);
 
 		/* we can't let two processes write to the socket at one time */
 		io_multiplexing_close();
@@ -473,7 +475,7 @@ static int do_recv(int f_in,int f_out,st
 		/* set place to send errors */
 		set_msg_fd_out(error_pipe[1]);
 
-		recv_files(f_in,flist,local_name);
+		recv_files(f_in, flist, local_name, name_pipe[0]);
 		io_flush(FULL_FLUSH);
 		report(f_in);
 
@@ -491,14 +493,16 @@ static int do_recv(int f_in,int f_out,st
 		stop_write_batch();
 
 	close(error_pipe[1]);
+	close(name_pipe[0]);
 	if (f_in != f_out)
 		close(f_in);
+	set_blocking(name_pipe[1]);
 
 	io_start_buffering_out(f_out);
 
 	set_msg_fd_in(error_pipe[0]);
 
-	generate_files(f_out, flist, local_name);
+	generate_files(f_out, flist, local_name, name_pipe[1]);
 
 	get_redo_num(); /* Read final MSG_DONE and any prior messages. */
 	report(-1);
--- orig/receiver.c	2004-07-16 20:07:22
+++ receiver.c	2004-07-03 20:08:07
@@ -304,14 +304,15 @@ static int receive_data(int f_in,struct 
  * main routine for receiver process.
  *
  * Receiver process runs on the same host as the generator process. */
-int recv_files(int f_in, struct file_list *flist, char *local_name)
+int recv_files(int f_in, struct file_list *flist, char *local_name,
+	       int f_name)
 {
 	int fd1,fd2;
 	STRUCT_STAT st;
 	char *fname, fbuf[MAXPATHLEN];
 	char template[MAXPATHLEN];
 	char fnametmp[MAXPATHLEN];
-	char *fnamecmp;
+	char *fnamecmp, *cp;
 	char fnamecmpbuf[MAXPATHLEN];
 	struct map_struct *mapbuf;
 	struct file_struct *file;
@@ -373,19 +374,19 @@ int recv_files(int f_in, struct file_lis
 		if (verbose > 2)
 			rprintf(FINFO,"recv_files(%s)\n",fname);
 
-		fnamecmp = fname;
+		for (cp = fnamecmpbuf; ; cp++) {
+			if (read(f_name, cp, 1) <= 0) {
+				rsyserr(FERROR, errno, "fname-pipe read failed");
+				exit_cleanup(RERR_PROTOCOL);
+			}
+			if (!*cp)
+				break;
+		}
+		fnamecmp = *fnamecmpbuf ? fnamecmpbuf : fname;
 
 		/* open the file */
 		fd1 = do_open(fnamecmp, O_RDONLY, 0);
 
-		if (fd1 == -1 && compare_dest != NULL) {
-			/* try the file at compare_dest instead */
-			pathjoin(fnamecmpbuf, sizeof fnamecmpbuf,
-				 compare_dest, fname);
-			fnamecmp = fnamecmpbuf;
-			fd1 = do_open(fnamecmp, O_RDONLY, 0);
-		}
-
 		if (fd1 != -1 && do_fstat(fd1,&st) != 0) {
 			rsyserr(FERROR, errno, "fstat %s failed",
 				full_fname(fnamecmp));
@@ -394,7 +395,7 @@ int recv_files(int f_in, struct file_lis
 			continue;
 		}
 
-		if (fd1 != -1 && S_ISDIR(st.st_mode) && fnamecmp == fname) {
+		if (fd1 != -1 && S_ISDIR(st.st_mode) && !*fnamecmpbuf) {
 			/* this special handling for directories
 			 * wouldn't be necessary if robust_rename()
 			 * and the underlying robust_unlink could cope
Commit	Line	Data
77945073 WD	1	Added a pipe from the generator to the receiver that communicates
	2	what basis file we used to generate the file data (if it was not
	3	the default name). This optimizes away the basis-file search in
	4	the receiver and makes future options that do more basis-file
	5	searching more efficient (such as the --fuzzy option and the
	6	support for multiple --compare-dest options).
	7
	8	You must run "make proto" before compiling.
	9
f6c3b300	10	--- orig/generator.c 2004-07-16 19:35:29
13bed3dd	11	+++ generator.c 2004-07-03 20:08:07
f6c3b300	12	@@ -251,7 +251,7 @@ static void generate_and_send_sums(struc
77945073 WD	13	* out. It might be wrong.
	14	*/
	15	static void recv_generator(char fname, struct file_struct file, int i,
	16	- int f_out)
	17	+ int f_out, int f_nameout)
	18	{
	19	int fd;
	20	STRUCT_STAT st;
f6c3b300	21	@@ -418,8 +418,10 @@ static void recv_generator(char *fname,
77945073 WD	22	statret = link_stat(fnamecmpbuf, &st, 0);
	23	if (!S_ISREG(st.st_mode))
	24	statret = -1;
	25	- if (statret == -1)
	26	+ if (statret < 0) {
	27	errno = saveerrno;
	28	+ *fnamecmpbuf = '\0';
	29	+ }
	30	#if HAVE_LINK
	31	else if (link_dest && !dry_run) {
	32	if (do_link(fnamecmpbuf, fname) != 0) {
f6c3b300	33	@@ -427,18 +429,22 @@ static void recv_generator(char *fname,
77945073 WD	34	rsyserr(FINFO, errno, "link %s => %s",
	35	fnamecmpbuf, fname);
	36	}
	37	- }
	38	- fnamecmp = fnamecmpbuf;
	39	+ fnamecmp = fnamecmpbuf;
	40	+ } else
	41	+ *fnamecmpbuf = '\0';
	42	}
	43	#endif
	44	else
	45	fnamecmp = fnamecmpbuf;
	46	- }
	47	+ } else
	48	+ *fnamecmpbuf = '\0';
	49
	50	if (statret == -1) {
	51	if (preserve_hard_links && hard_link_check(file, HL_SKIP))
	52	return;
	53	if (errno == ENOENT) {
	54	+ if (f_nameout >= 0)
	55	+ write(f_nameout, "", 1);
	56	write_int(f_out,i);
	57	if (!dry_run)
	58	write_sum_head(f_out, NULL);
f6c3b300	59	@@ -458,19 +464,21 @@ static void recv_generator(char *fname,
77945073 WD	60	/* now pretend the file didn't exist */
	61	if (preserve_hard_links && hard_link_check(file, HL_SKIP))
	62	return;
	63	+ if (f_nameout >= 0)
	64	+ write(f_nameout, "", 1);
	65	write_int(f_out,i);
	66	if (!dry_run)
	67	write_sum_head(f_out, NULL);
	68	return;
	69	}
	70
	71	- if (opt_ignore_existing && fnamecmp == fname) {
	72	+ if (opt_ignore_existing && !*fnamecmpbuf) {
	73	if (verbose > 1)
	74	rprintf(FINFO,"%s exists\n",fname);
	75	return;
	76	}
	77
	78	- if (update_only && fnamecmp == fname
	79	+ if (update_only && !*fnamecmpbuf
	80	&& cmp_modtime(st.st_mtime, file->modtime) > 0) {
	81	if (verbose > 1)
	82	rprintf(FINFO,"%s is newer\n",fname);
f6c3b300	83	@@ -478,17 +486,21 @@ static void recv_generator(char *fname,
77945073 WD	84	}
	85
	86	if (skip_file(fname, file, &st)) {
	87	- if (fnamecmp == fname)
	88	+ if (!*fnamecmpbuf)
	89	set_perms(fname, file, &st, PERMS_REPORT);
	90	return;
	91	}
	92
	93	if (dry_run) {
	94	+ if (f_nameout >= 0)
	95	+ write(f_nameout, "", 1);
	96	write_int(f_out,i);
	97	return;
	98	}
	99
9be39c35	100	if (whole_file > 0) {
77945073 WD	101	+ if (f_nameout >= 0)
	102	+ write(f_nameout, "", 1);
	103	write_int(f_out,i);
	104	write_sum_head(f_out, NULL);
	105	return;
f6c3b300	106	@@ -503,6 +515,8 @@ static void recv_generator(char *fname,
77945073 WD	107	/* pretend the file didn't exist */
	108	if (preserve_hard_links && hard_link_check(file, HL_SKIP))
	109	return;
	110	+ if (f_nameout >= 0)
	111	+ write(f_nameout, "", 1);
	112	write_int(f_out,i);
	113	write_sum_head(f_out, NULL);
	114	return;
f6c3b300	115	@@ -521,6 +535,8 @@ static void recv_generator(char *fname,
77945073 WD	116	if (verbose > 2)
	117	rprintf(FINFO, "generating and sending sums for %d\n", i);
	118
	119	+ if (f_nameout >= 0)
	120	+ write(f_nameout, fnamecmpbuf, strlen(fnamecmpbuf) + 1);
	121	write_int(f_out,i);
	122	generate_and_send_sums(mapbuf, st.st_size, f_out);
	123
f6c3b300	124	@@ -530,7 +546,8 @@ static void recv_generator(char *fname,
77945073 WD	125	}
	126
	127
9be39c35 WD	128	-void generate_files(int f_out, struct file_list flist, char local_name)
9be39c35 WD	129	+void generate_files(int f_out, struct file_list flist, char local_name,
77945073 WD	130	+ int f_nameout)
	131	{
	132	int i;
8cec1ead	133	int phase = 0;
f6c3b300	134	@@ -571,7 +588,7 @@ void generate_files(int f_out, struct fi
77945073 WD	135	}
	136
	137	recv_generator(local_name ? local_name : f_name_to(file, fbuf),
9be39c35 WD	138	- file, i, f_out);
9be39c35 WD	139	+ file, i, f_out, f_nameout);
77945073 WD	140	}
	141
	142	phase++;
f6c3b300	143	@@ -588,7 +605,7 @@ void generate_files(int f_out, struct fi
77945073 WD	144	while ((i = get_redo_num()) != -1) {
	145	struct file_struct *file = flist->files[i];
	146	recv_generator(local_name ? local_name : f_name_to(file, fbuf),
9be39c35 WD	147	- file, i, f_out);
9be39c35 WD	148	+ file, i, f_out, f_nameout);
77945073 WD	149	}
	150
	151	phase++;
f6c3b300	152	@@ -607,7 +624,7 @@ void generate_files(int f_out, struct fi
77945073 WD	153	if (!file->basename \|\| !S_ISDIR(file->mode))
	154	continue;
	155	recv_generator(local_name ? local_name : f_name(file),
	156	- file, i, -1);
	157	+ file, i, -1, -1);
	158	}
	159
	160	if (verbose > 2)
9be39c35	161	--- orig/main.c 2004-07-15 17:02:03
13bed3dd	162	+++ main.c 2004-07-03 20:08:07
9be39c35	163	@@ -443,7 +443,7 @@ static int do_recv(int f_in,int f_out,st
77945073 WD	164	{
77945073 WD	165	int pid;
8cec1ead	166	int status = 0;
77945073	167	- int error_pipe[2];
77945073 WD	168	+ int error_pipe[2], name_pipe[2];
	169
	170	if (preserve_hard_links)
	171	init_hard_links(flist);
9be39c35	172	@@ -455,8 +455,8 @@ static int do_recv(int f_in,int f_out,st
77945073 WD	173	}
	174	}
	175
	176	- if (fd_pair(error_pipe) < 0) {
	177	- rprintf(FERROR,"error pipe failed in do_recv\n");
	178	+ if (fd_pair(error_pipe) < 0 \|\| fd_pair(name_pipe) < 0) {
	179	+ rprintf(FERROR, "fd_pair() failed in do_recv\n");
	180	exit_cleanup(RERR_SOCKETIO);
	181	}
	182
9be39c35	183	@@ -464,8 +464,10 @@ static int do_recv(int f_in,int f_out,st
77945073	184
8cec1ead	185	if ((pid = do_fork()) == 0) {
77945073 WD	186	close(error_pipe[0]);
	187	+ close(name_pipe[1]);
	188	if (f_in != f_out)
	189	close(f_out);
	190	+ set_blocking(name_pipe[0]);
	191
	192	/* we can't let two processes write to the socket at one time */
	193	io_multiplexing_close();
9be39c35	194	@@ -473,7 +475,7 @@ static int do_recv(int f_in,int f_out,st
77945073 WD	195	/* set place to send errors */
	196	set_msg_fd_out(error_pipe[1]);
	197
	198	- recv_files(f_in,flist,local_name);
	199	+ recv_files(f_in, flist, local_name, name_pipe[0]);
	200	io_flush(FULL_FLUSH);
	201	report(f_in);
	202
9be39c35 WD	203	@@ -491,14 +493,16 @@ static int do_recv(int f_in,int f_out,st
9be39c35 WD	204	stop_write_batch();
77945073 WD	205
	206	close(error_pipe[1]);
	207	+ close(name_pipe[0]);
	208	if (f_in != f_out)
	209	close(f_in);
	210	+ set_blocking(name_pipe[1]);
	211
	212	io_start_buffering_out(f_out);
	213
	214	set_msg_fd_in(error_pipe[0]);
	215
	216	- generate_files(f_out, flist, local_name);
	217	+ generate_files(f_out, flist, local_name, name_pipe[1]);
	218
	219	get_redo_num(); /* Read final MSG_DONE and any prior messages. */
	220	report(-1);
f6c3b300	221	--- orig/receiver.c 2004-07-16 20:07:22
13bed3dd	222	+++ receiver.c 2004-07-03 20:08:07
f6c3b300	223	@@ -304,14 +304,15 @@ static int receive_data(int f_in,struct
77945073 WD	224	* main routine for receiver process.
	225	*
	226	* Receiver process runs on the same host as the generator process. */
9be39c35	227	-int recv_files(int f_in, struct file_list flist, char local_name)
77945073 WD	228	+int recv_files(int f_in, struct file_list flist, char local_name,
	229	+ int f_name)
	230	{
	231	int fd1,fd2;
	232	STRUCT_STAT st;
	233	char *fname, fbuf[MAXPATHLEN];
	234	char template[MAXPATHLEN];
	235	char fnametmp[MAXPATHLEN];
	236	- char *fnamecmp;
	237	+ char fnamecmp, cp;
	238	char fnamecmpbuf[MAXPATHLEN];
	239	struct map_struct *mapbuf;
	240	struct file_struct *file;
f6c3b300	241	@@ -373,19 +374,19 @@ int recv_files(int f_in, struct file_lis
77945073 WD	242	if (verbose > 2)
	243	rprintf(FINFO,"recv_files(%s)\n",fname);
	244
	245	- fnamecmp = fname;
	246	+ for (cp = fnamecmpbuf; ; cp++) {
	247	+ if (read(f_name, cp, 1) <= 0) {
	248	+ rsyserr(FERROR, errno, "fname-pipe read failed");
	249	+ exit_cleanup(RERR_PROTOCOL);
	250	+ }
	251	+ if (!*cp)
	252	+ break;
	253	+ }
	254	+ fnamecmp = *fnamecmpbuf ? fnamecmpbuf : fname;
	255
	256	/* open the file */
	257	fd1 = do_open(fnamecmp, O_RDONLY, 0);
	258
	259	- if (fd1 == -1 && compare_dest != NULL) {
	260	- /* try the file at compare_dest instead */
	261	- pathjoin(fnamecmpbuf, sizeof fnamecmpbuf,
	262	- compare_dest, fname);
	263	- fnamecmp = fnamecmpbuf;
	264	- fd1 = do_open(fnamecmp, O_RDONLY, 0);
	265	- }
	266	-
	267	if (fd1 != -1 && do_fstat(fd1,&st) != 0) {
	268	rsyserr(FERROR, errno, "fstat %s failed",
	269	full_fname(fnamecmp));
f6c3b300	270	@@ -394,7 +395,7 @@ int recv_files(int f_in, struct file_lis
77945073 WD	271	continue;
	272	}
	273
	274	- if (fd1 != -1 && S_ISDIR(st.st_mode) && fnamecmp == fname) {
	275	+ if (fd1 != -1 && S_ISDIR(st.st_mode) && !*fnamecmpbuf) {
	276	/* this special handling for directories
	277	* wouldn't be necessary if robust_rename()
	278	* and the underlying robust_unlink could cope