From 58118c25b50ecc953096d3569aea7bbc6e0d3f25 Mon Sep 17 00:00:00 2001 From: Wayne Davison Date: Wed, 30 Jun 2004 07:38:53 +0000 Subject: [PATCH] Incorporated the g2r-basis-filename.diff changes so that we don't duplicate the fuzzy scan in the receiver that the generator already performed. --- fuzzy.diff | 364 ++++++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 290 insertions(+), 74 deletions(-) diff --git a/fuzzy.diff b/fuzzy.diff index 460d3cb..0dbc1ba 100644 --- a/fuzzy.diff +++ b/fuzzy.diff @@ -2,39 +2,22 @@ Wayne Davison wrote: I greatly simplified the changes to generator.c, making the patch easier to maintain and fixing the failing test in the testsuite. -Very lightly tested (by me). +Very lightly tested. ---- Makefile.in 15 May 2004 00:48:11 -0000 1.101 -+++ Makefile.in 29 Jun 2004 17:46:12 -0000 -@@ -32,7 +32,7 @@ ZLIBOBJ=zlib/deflate.o zlib/infblock.o z - zlib/inflate.o zlib/inftrees.o zlib/infutil.o zlib/trees.o \ - zlib/zutil.o zlib/adler32.o - OBJS1=rsync.o generator.o receiver.o cleanup.o sender.o exclude.o util.o \ -- main.o checksum.o match.o syscall.o log.o backup.o -+ main.o checksum.o match.o syscall.o log.o backup.o alternate.o - OBJS2=options.o flist.o io.o compat.o hlink.o token.o uidlist.o socket.o \ - fileio.o batch.o clientname.o - OBJS3=progress.o pipe.o ---- /dev/null 1 Jan 1970 00:00:00 -0000 -+++ alternate.c 29 Jun 2004 17:46:12 -0000 -@@ -0,0 +1,105 @@ -+#include "rsync.h" -+ -+extern char *compare_dest; -+extern int verbose; -+ -+/* Alternate methods for opening files, if local doesn't exist */ -+/* Sanity check that we are about to open regular file */ -+static int do_open_regular(char *fname) -+{ -+ STRUCT_STAT st; -+ -+ if (do_stat(fname, &st) == 0 && S_ISREG(st.st_mode)) -+ return do_open(fname, O_RDONLY, 0); -+ -+ return -1; -+} -+ +--- generator.c 29 Jun 2004 19:19:00 -0000 1.92 ++++ generator.c 30 Jun 2004 07:35:56 -0000 +@@ -41,6 +41,7 @@ extern int ignore_times; + extern int size_only; + extern int io_timeout; + extern int protocol_version; ++extern int fuzzy; + extern int always_checksum; + extern char *compare_dest; + extern int link_dest; +@@ -257,6 +258,94 @@ static void generate_and_send_sums(struc + } + + +static void split_names(char *fname, char **dirname, char **basename) +{ + char *slash = strrchr(fname, '/'); @@ -48,6 +31,7 @@ Very lightly tested (by me). + } +} + ++ +static unsigned int measure_name(const char *name, const char *basename, + const char *ext) +{ @@ -65,8 +49,8 @@ Very lightly tested (by me). + return score; +} + -+int check_alternate_base_fuzzy(char **fname_ptr, char *buf, -+ STRUCT_STAT *st_ptr) ++ ++static int find_fuzzy(char **fname_ptr, char *buf, STRUCT_STAT *st_ptr) +{ + DIR *d; + struct dirent *di; @@ -117,23 +101,21 @@ Very lightly tested (by me). + who_am_i(), *fname_ptr, buf); + } + *fname_ptr = buf; -+ if (st_ptr) -+ return link_stat(buf, st_ptr, 0); -+ return do_open_regular(buf); ++ return link_stat(buf, st_ptr, 0); + } + return -1; +} ---- generator.c 29 Jun 2004 16:22:54 -0000 1.91 -+++ generator.c 29 Jun 2004 17:46:12 -0000 -@@ -41,6 +41,7 @@ extern int ignore_times; - extern int size_only; - extern int io_timeout; - extern int protocol_version; -+extern int fuzzy; - extern int always_checksum; - extern char *compare_dest; - extern int link_dest; -@@ -271,7 +272,7 @@ void recv_generator(char *fname, struct ++ + + /* + * Acts on file number @p i from @p flist, whose name is @p fname. +@@ -267,12 +356,12 @@ 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; struct map_struct *mapbuf; @@ -142,33 +124,223 @@ Very lightly tested (by me). char *fnamecmp; char fnamecmpbuf[MAXPATHLEN]; -@@ -447,6 +448,15 @@ void recv_generator(char *fname, struct +@@ -431,8 +520,10 @@ 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) { +@@ -440,18 +531,30 @@ 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 && fuzzy) { -+ statret = check_alternate_base_fuzzy(&fnamecmp, fnamecmpbuf, -+ &st); ++ statret = find_fuzzy(&fnamecmp, fnamecmpbuf, &st); + if (!S_ISREG(st.st_mode)) + statret = -1; + else + fuzzy_file = 1; -+ } -+ + } + if (statret == -1) { if (preserve_hard_links && hard_link_check(file, HL_SKIP)) return; -@@ -489,7 +499,7 @@ void recv_generator(char *fname, struct + 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); +@@ -471,37 +574,43 @@ 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); return; } - if (skip_file(fname, file, &st)) { +- if (fnamecmp == fname) + if (!fuzzy_file && 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 (disable_deltas_p()) { ++ if (f_nameout >= 0) ++ write(f_nameout, "", 1); + write_int(f_out,i); + write_sum_head(f_out, NULL); + return; +@@ -516,6 +625,8 @@ 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; +@@ -534,6 +645,8 @@ 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); + +@@ -543,10 +656,11 @@ static void recv_generator(char *fname, + } + + +-void generate_files(int f, struct file_list *flist, char *local_name) ++void generate_files(int f, struct file_list *flist, char *local_name, ++ int f_nameout) + { + int i; +- int phase=0; ++ int phase = 0; + char fbuf[MAXPATHLEN]; + + if (verbose > 2) { +@@ -584,7 +698,7 @@ void generate_files(int f, struct file_l + } + + recv_generator(local_name ? local_name : f_name_to(file, fbuf), +- file, i, f); ++ file, i, f, f_nameout); + } + + phase++; +@@ -601,7 +715,7 @@ void generate_files(int f, struct file_l + 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); ++ file, i, f, f_nameout); + } + + phase++; +@@ -620,7 +734,7 @@ void generate_files(int f, struct file_l + 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) +--- main.c 28 Jun 2004 17:45:40 -0000 1.201 ++++ main.c 30 Jun 2004 07:35:57 -0000 +@@ -428,8 +428,8 @@ static void do_server_sender(int f_in, i + static int do_recv(int f_in,int f_out,struct file_list *flist,char *local_name) + { + int pid; +- int status=0; +- int error_pipe[2]; ++ int status = 0; ++ int error_pipe[2], name_pipe[2]; + + if (preserve_hard_links) + init_hard_links(flist); +@@ -441,17 +441,19 @@ 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); + } + + io_flush(NORMAL_FLUSH); + +- if ((pid=do_fork()) == 0) { ++ 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(); +@@ -459,7 +461,7 @@ 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); + +@@ -475,14 +477,16 @@ static int do_recv(int f_in,int f_out,st + am_generator = 1; + + 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); --- options.c 20 Jun 2004 19:47:05 -0000 1.157 -+++ options.c 29 Jun 2004 17:46:13 -0000 ++++ options.c 30 Jun 2004 07:35:57 -0000 @@ -94,6 +94,7 @@ int ignore_errors = 0; int modify_window = 0; int blocking_io = -1; @@ -204,29 +376,73 @@ Very lightly tested (by me). return; --- receiver.c 29 Jun 2004 15:12:01 -0000 1.83 -+++ receiver.c 29 Jun 2004 17:46:13 -0000 -@@ -48,6 +48,7 @@ extern int ignore_errors; - extern int orig_umask; - extern int keep_partial; - extern int checksum_seed; -+extern int fuzzy; - - static void delete_one(char *fn, int is_dir) ++++ receiver.c 30 Jun 2004 07:35:57 -0000 +@@ -36,7 +36,6 @@ extern int preserve_perms; + extern int cvs_exclude; + extern int io_error; + extern char *tmpdir; +-extern char *compare_dest; + extern int make_backups; + extern int do_progress; + extern char *backup_dir; +@@ -293,14 +292,15 @@ 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) { -@@ -377,6 +378,11 @@ int recv_files(int f_in,struct file_list - fd1 = do_open(fnamecmp, O_RDONLY, 0); - } + 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; +@@ -364,19 +364,19 @@ int recv_files(int f_in,struct file_list + if (verbose > 2) + rprintf(FINFO,"recv_files(%s)\n",fname); -+ if (fd1 == -1 && fuzzy) { -+ fd1 = check_alternate_base_fuzzy(&fnamecmp, fnamecmpbuf, -+ NULL); +- 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)); +@@ -385,7 +385,7 @@ int recv_files(int f_in,struct file_list + 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 --- rsync.yo 5 Jun 2004 16:16:30 -0000 1.171 -+++ rsync.yo 29 Jun 2004 17:46:14 -0000 ++++ rsync.yo 30 Jun 2004 07:35:58 -0000 @@ -325,6 +325,7 @@ verb( -T --temp-dir=DIR create temporary files in directory DIR --compare-dest=DIR also compare received files relative to DIR -- 2.34.1