/* -*- c-file-style: "linux" -*- * * Copyright (C) 1996-2000 by Andrew Tridgell * Copyright (C) Paul Mackerras 1996 * Copyright (C) 2001, 2002 by Martin Pool * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ /** * @file * * Utilities used in rsync **/ #include "rsync.h" extern int verbose; int sanitize_paths = 0; /** * Set a fd into nonblocking mode **/ void set_nonblocking(int fd) { int val; if ((val = fcntl(fd, F_GETFL, 0)) == -1) return; if (!(val & NONBLOCK_FLAG)) { val |= NONBLOCK_FLAG; fcntl(fd, F_SETFL, val); } } /** * Set a fd into blocking mode **/ void set_blocking(int fd) { int val; if ((val = fcntl(fd, F_GETFL, 0)) == -1) return; if (val & NONBLOCK_FLAG) { val &= ~NONBLOCK_FLAG; fcntl(fd, F_SETFL, val); } } /** * Create a file descriptor pair - like pipe() but use socketpair if * possible (because of blocking issues on pipes). * * Always set non-blocking. */ int fd_pair(int fd[2]) { int ret; #if HAVE_SOCKETPAIR ret = socketpair(AF_UNIX, SOCK_STREAM, 0, fd); #else ret = pipe(fd); #endif if (ret == 0) { set_nonblocking(fd[0]); set_nonblocking(fd[1]); } return ret; } void print_child_argv(char **cmd) { rprintf(FINFO, "opening connection using "); for (; *cmd; cmd++) { /* Look for characters that ought to be quoted. This * is not a great quoting algorithm, but it's * sufficient for a log message. */ if (strspn(*cmd, "abcdefghijklmnopqrstuvwxyz" "ABCDEFGHIJKLMNOPQRSTUVWXYZ" "0123456789" ",.-_=+@/") != strlen(*cmd)) { rprintf(FINFO, "\"%s\" ", *cmd); } else { rprintf(FINFO, "%s ", *cmd); } } rprintf(FINFO, "\n"); } void out_of_memory(char *str) { rprintf(FERROR,"ERROR: out of memory in %s\n",str); exit_cleanup(RERR_MALLOC); } void overflow(char *str) { rprintf(FERROR,"ERROR: buffer overflow in %s\n",str); exit_cleanup(RERR_MALLOC); } int set_modtime(char *fname, time_t modtime) { extern int dry_run; if (dry_run) return 0; if (verbose > 2) { rprintf(FINFO, "set modtime of %s to (%ld) %s", fname, (long) modtime, asctime(localtime(&modtime))); } { #ifdef HAVE_UTIMBUF struct utimbuf tbuf; tbuf.actime = time(NULL); tbuf.modtime = modtime; return utime(fname,&tbuf); #elif defined(HAVE_UTIME) time_t t[2]; t[0] = time(NULL); t[1] = modtime; return utime(fname,t); #else struct timeval t[2]; t[0].tv_sec = time(NULL); t[0].tv_usec = 0; t[1].tv_sec = modtime; t[1].tv_usec = 0; return utimes(fname,t); #endif } } /** Create any necessary directories in fname. Unfortunately we don't know what perms to give the directory when this is called so we need to rely on the umask **/ int create_directory_path(char *fname, int base_umask) { char *p; while (*fname == '/') fname++; while (strncmp(fname,"./",2)==0) fname += 2; p = fname; while ((p=strchr(p,'/'))) { *p = 0; do_mkdir(fname, 0777 & ~base_umask); *p = '/'; p++; } return 0; } /** * Write @p len bytes at @p ptr to descriptor @p desc, retrying if * interrupted. * * @retval len upon success * * @retval <0 write's (negative) error code * * Derived from GNU C's cccp.c. */ static int full_write(int desc, char *ptr, size_t len) { int total_written; total_written = 0; while (len > 0) { int written = write (desc, ptr, len); if (written < 0) { #ifdef EINTR if (errno == EINTR) continue; #endif return written; } total_written += written; ptr += written; len -= written; } return total_written; } /** * Read @p len bytes at @p ptr from descriptor @p desc, retrying if * interrupted. * * @retval >0 the actual number of bytes read * * @retval 0 for EOF * * @retval <0 for an error. * * Derived from GNU C's cccp.c. */ static int safe_read(int desc, char *ptr, size_t len) { int n_chars; if (len == 0) return len; #ifdef EINTR do { n_chars = read(desc, ptr, len); } while (n_chars < 0 && errno == EINTR); #else n_chars = read(desc, ptr, len); #endif return n_chars; } /** Copy a file. * * This is used in conjunction with the --temp-dir option */ int copy_file(char *source, char *dest, mode_t mode) { int ifd; int ofd; char buf[1024 * 8]; int len; /* Number of bytes read into `buf'. */ ifd = do_open(source, O_RDONLY, 0); if (ifd == -1) { rprintf(FERROR,"open %s: %s\n", source,strerror(errno)); return -1; } if (robust_unlink(dest) && errno != ENOENT) { rprintf(FERROR,"unlink %s: %s\n", dest,strerror(errno)); return -1; } ofd = do_open(dest, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL, mode); if (ofd == -1) { rprintf(FERROR,"open %s: %s\n", dest,strerror(errno)); close(ifd); return -1; } while ((len = safe_read(ifd, buf, sizeof(buf))) > 0) { if (full_write(ofd, buf, len) < 0) { rprintf(FERROR,"write %s: %s\n", dest,strerror(errno)); close(ifd); close(ofd); return -1; } } close(ifd); close(ofd); if (len < 0) { rprintf(FERROR,"read %s: %s\n", source,strerror(errno)); return -1; } return 0; } /* MAX_RENAMES should be 10**MAX_RENAMES_DIGITS */ #define MAX_RENAMES_DIGITS 3 #define MAX_RENAMES 1000 /** * Robust unlink: some OS'es (HPUX) refuse to unlink busy files, so rename to /.rsyncNNN instead. Note that successive rsync runs will shuffle the filenames around a bit as long as the file is still busy; this is because this function does not know if the unlink call is due to a new file coming in, or --delete trying to remove old .rsyncNNN files, hence it renames it each time. */ int robust_unlink(char *fname) { #ifndef ETXTBSY return do_unlink(fname); #else static int counter = 1; int rc, pos, start; char path[MAXPATHLEN]; rc = do_unlink(fname); if ((rc == 0) || (errno != ETXTBSY)) return rc; strlcpy(path, fname, MAXPATHLEN); pos = strlen(path); while((path[--pos] != '/') && (pos >= 0)) ; ++pos; strlcpy(&path[pos], ".rsync", MAXPATHLEN-pos); pos += sizeof(".rsync")-1; if (pos > (MAXPATHLEN-MAX_RENAMES_DIGITS-1)) { errno = ETXTBSY; return -1; } /* start where the last one left off to reduce chance of clashes */ start = counter; do { sprintf(&path[pos], "%03d", counter); if (++counter >= MAX_RENAMES) counter = 1; } while (((rc = access(path, 0)) == 0) && (counter != start)); if (verbose > 0) rprintf(FINFO,"renaming %s to %s because of text busy\n", fname, path); /* maybe we should return rename()'s exit status? Nah. */ if (do_rename(fname, path) != 0) { errno = ETXTBSY; return -1; } return 0; #endif } int robust_rename(char *from, char *to) { #ifndef ETXTBSY return do_rename(from, to); #else int rc = do_rename(from, to); if ((rc == 0) || (errno != ETXTBSY)) return rc; if (robust_unlink(to) != 0) return -1; return do_rename(from, to); #endif } static pid_t all_pids[10]; static int num_pids; /** Fork and record the pid of the child. **/ pid_t do_fork(void) { pid_t newpid = fork(); if (newpid != 0 && newpid != -1) { all_pids[num_pids++] = newpid; } return newpid; } /** * Kill all children. * * @todo It would be kind of nice to make sure that they are actually * all our children before we kill them, because their pids may have * been recycled by some other process. Perhaps when we wait for a * child, we should remove it from this array. Alternatively we could * perhaps use process groups, but I think that would not work on * ancient Unix versions that don't support them. **/ void kill_all(int sig) { int i; for (i = 0; i < num_pids; i++) { /* Let's just be a little careful where we * point that gun, hey? See kill(2) for the * magic caused by negative values. */ pid_t p = all_pids[i]; if (p == getpid()) continue; if (p <= 0) continue; kill(p, sig); } } /** Turn a user name into a uid */ int name_to_uid(char *name, uid_t *uid) { struct passwd *pass; if (!name || !*name) return 0; pass = getpwnam(name); if (pass) { *uid = pass->pw_uid; return 1; } return 0; } /** Turn a group name into a gid */ int name_to_gid(char *name, gid_t *gid) { struct group *grp; if (!name || !*name) return 0; grp = getgrnam(name); if (grp) { *gid = grp->gr_gid; return 1; } return 0; } /** Lock a byte range in a open file */ int lock_range(int fd, int offset, int len) { struct flock lock; lock.l_type = F_WRLCK; lock.l_whence = SEEK_SET; lock.l_start = offset; lock.l_len = len; lock.l_pid = 0; return fcntl(fd,F_SETLK,&lock) == 0; } static void glob_expand_one(char *s, char **argv, int *argc, int maxargs) { #if !(defined(HAVE_GLOB) && defined(HAVE_GLOB_H)) if (!*s) s = "."; argv[*argc] = strdup(s); (*argc)++; return; #else extern int sanitize_paths; glob_t globbuf; int i; if (!*s) s = "."; argv[*argc] = strdup(s); if (sanitize_paths) { sanitize_path(argv[*argc], NULL); } memset(&globbuf, 0, sizeof(globbuf)); glob(argv[*argc], 0, NULL, &globbuf); if (globbuf.gl_pathc == 0) { (*argc)++; globfree(&globbuf); return; } for (i=0; i<(maxargs - (*argc)) && i < (int) globbuf.gl_pathc;i++) { if (i == 0) free(argv[*argc]); argv[(*argc) + i] = strdup(globbuf.gl_pathv[i]); if (!argv[(*argc) + i]) out_of_memory("glob_expand"); } globfree(&globbuf); (*argc) += i; #endif } void glob_expand(char *base1, char **argv, int *argc, int maxargs) { char *s = argv[*argc]; char *p, *q; char *base = base1; if (!s || !*s) return; if (strncmp(s, base, strlen(base)) == 0) { s += strlen(base); } s = strdup(s); if (!s) out_of_memory("glob_expand"); if (asprintf(&base," %s/", base1) <= 0) out_of_memory("glob_expand"); q = s; while ((p = strstr(q,base)) && ((*argc) < maxargs)) { /* split it at this point */ *p = 0; glob_expand_one(q, argv, argc, maxargs); q = p+strlen(base); } if (*q && (*argc < maxargs)) glob_expand_one(q, argv, argc, maxargs); free(s); free(base); } /** * Convert a string to lower case **/ void strlower(char *s) { while (*s) { if (isupper(*s)) *s = tolower(*s); s++; } } void *Realloc(void *p, int size) { if (!p) return (void *)malloc(size); return (void *)realloc(p, size); } void clean_fname(char *name) { char *p; int l; int modified = 1; if (!name) return; while (modified) { modified = 0; if ((p=strstr(name,"/./"))) { modified = 1; while (*p) { p[0] = p[2]; p++; } } if ((p=strstr(name,"//"))) { modified = 1; while (*p) { p[0] = p[1]; p++; } } if (strncmp(p=name,"./",2) == 0) { modified = 1; do { p[0] = p[2]; } while (*p++); } l = strlen(p=name); if (l > 1 && p[l-1] == '/') { modified = 1; p[l-1] = 0; } } } /** * Make path appear as if a chroot had occurred: * * 1. remove leading "/" (or replace with "." if at end) * 2. remove leading ".." components (except those allowed by "reldir") * 3. delete any other "/.." (recursively) * * Can only shrink paths, so sanitizes in place. * * While we're at it, remove double slashes and "." components like * clean_fname does(), but DON'T remove a trailing slash because that * is sometimes significant on command line arguments. * * If "reldir" is non-null, it is a sanitized directory that the path will be * relative to, so allow as many ".." at the beginning of the path as * there are components in reldir. This is used for symbolic link targets. * If reldir is non-null and the path began with "/", to be completely like * a chroot we should add in depth levels of ".." at the beginning of the * path, but that would blow the assumption that the path doesn't grow and * it is not likely to end up being a valid symlink anyway, so just do * the normal removal of the leading "/" instead. * * Contributed by Dave Dykstra */ void sanitize_path(char *p, char *reldir) { char *start, *sanp; int depth = 0; int allowdotdot = 0; if (reldir) { depth++; while (*reldir) { if (*reldir++ == '/') { depth++; } } } start = p; sanp = p; while (*p == '/') { /* remove leading slashes */ p++; } while (*p != '\0') { /* this loop iterates once per filename component in p. * both p (and sanp if the original had a slash) should * always be left pointing after a slash */ if ((*p == '.') && ((*(p+1) == '/') || (*(p+1) == '\0'))) { /* skip "." component */ while (*++p == '/') { /* skip following slashes */ ; } continue; } allowdotdot = 0; if ((*p == '.') && (*(p+1) == '.') && ((*(p+2) == '/') || (*(p+2) == '\0'))) { /* ".." component followed by slash or end */ if ((depth > 0) && (sanp == start)) { /* allow depth levels of .. at the beginning */ --depth; allowdotdot = 1; } else { p += 2; if (*p == '/') p++; if (sanp != start) { /* back up sanp one level */ --sanp; /* now pointing at slash */ while ((sanp > start) && (*(sanp - 1) != '/')) { /* skip back up to slash */ sanp--; } } continue; } } while (1) { /* copy one component through next slash */ *sanp++ = *p++; if ((*p == '\0') || (*(p-1) == '/')) { while (*p == '/') { /* skip multiple slashes */ p++; } break; } } if (allowdotdot) { /* move the virtual beginning to leave the .. alone */ start = sanp; } } if ((sanp == start) && !allowdotdot) { /* ended up with nothing, so put in "." component */ /* * note that the !allowdotdot doesn't prevent this from * happening in all allowed ".." situations, but I didn't * think it was worth putting in an extra variable to ensure * it since an extra "." won't hurt in those situations. */ *sanp++ = '.'; } *sanp = '\0'; } static char curr_dir[MAXPATHLEN]; /** like chdir() but can be reversed with pop_dir() if save is set. It is also much faster as it remembers where we have been */ char *push_dir(char *dir, int save) { char *ret = curr_dir; static int initialised; if (!initialised) { initialised = 1; getcwd(curr_dir, sizeof(curr_dir)-1); } if (!dir) return NULL; /* this call was probably just to initialize */ if (chdir(dir)) return NULL; if (save) { ret = strdup(curr_dir); } if (*dir == '/') { strlcpy(curr_dir, dir, sizeof(curr_dir)); } else { strlcat(curr_dir,"/", sizeof(curr_dir)); strlcat(curr_dir,dir, sizeof(curr_dir)); } clean_fname(curr_dir); return ret; } /** Reverse a push_dir call */ int pop_dir(char *dir) { int ret; ret = chdir(dir); if (ret) { free(dir); return ret; } strlcpy(curr_dir, dir, sizeof(curr_dir)); free(dir); return 0; } /** We need to supply our own strcmp function for file list comparisons to ensure that signed/unsigned usage is consistent between machines. */ int u_strcmp(const char *cs1, const char *cs2) { const uchar *s1 = (const uchar *)cs1; const uchar *s2 = (const uchar *)cs2; while (*s1 && *s2 && (*s1 == *s2)) { s1++; s2++; } return (int)*s1 - (int)*s2; } /** * Determine if a symlink points outside the current directory tree. * This is considered "unsafe" because e.g. when mirroring somebody * else's machine it might allow them to establish a symlink to * /etc/passwd, and then read it through a web server. * * @param dest Target of the symlink in question. * * @param src Top source directory currently applicable. Basically this * is the first parameter to rsync in a simple invocation, but it's * modified by flist.c in slightly complex ways. * * @retval True if unsafe * @retval False is unsafe **/ int unsafe_symlink(char *dest, char *src) { char *tok; int depth = 0; /* all absolute and null symlinks are unsafe */ if (!dest || !(*dest) || (*dest == '/')) return 1; src = strdup(src); if (!src) out_of_memory("unsafe_symlink"); /* find out what our safety margin is */ for (tok=strtok(src,"/"); tok; tok=strtok(NULL,"/")) { if (strcmp(tok,"..") == 0) { depth=0; } else if (strcmp(tok,".") == 0) { /* nothing */ } else { depth++; } } free(src); /* drop by one to account for the filename portion */ depth--; dest = strdup(dest); if (!dest) out_of_memory("unsafe_symlink"); for (tok=strtok(dest,"/"); tok; tok=strtok(NULL,"/")) { if (strcmp(tok,"..") == 0) { depth--; } else if (strcmp(tok,".") == 0) { /* nothing */ } else { depth++; } /* if at any point we go outside the current directory then stop - it is unsafe */ if (depth < 0) break; } free(dest); return (depth < 0); } /** Return the date and time as a string */ char *timestring(time_t t) { static char TimeBuf[200]; struct tm *tm = localtime(&t); #ifdef HAVE_STRFTIME strftime(TimeBuf,sizeof(TimeBuf)-1,"%Y/%m/%d %T",tm); #else strlcpy(TimeBuf, asctime(tm), sizeof(TimeBuf)); #endif if (TimeBuf[strlen(TimeBuf)-1] == '\n') { TimeBuf[strlen(TimeBuf)-1] = 0; } return(TimeBuf); } /** * Sleep for a specified number of milliseconds. * * Always returns TRUE. (In the future it might return FALSE if * interrupted.) **/ int msleep(int t) { int tdiff=0; struct timeval tval,t1,t2; gettimeofday(&t1, NULL); gettimeofday(&t2, NULL); while (tdiff < t) { tval.tv_sec = (t-tdiff)/1000; tval.tv_usec = 1000*((t-tdiff)%1000); errno = 0; select(0,NULL,NULL, NULL, &tval); gettimeofday(&t2, NULL); tdiff = (t2.tv_sec - t1.tv_sec)*1000 + (t2.tv_usec - t1.tv_usec)/1000; } return True; } /** * Determine if two file modification times are equivalent (either * exact or in the modification timestamp window established by * --modify-window). * * @retval 0 if the times should be treated as the same * * @retval +1 if the first is later * * @retval -1 if the 2nd is later **/ int cmp_modtime(time_t file1, time_t file2) { extern int modify_window; if (file2 > file1) { if (file2 - file1 <= modify_window) return 0; return -1; } if (file1 - file2 <= modify_window) return 0; return 1; } #ifdef __INSURE__XX #include /** This routine is a trick to immediately catch errors when debugging with insure. A xterm with a gdb is popped up when insure catches a error. It is Linux specific. **/ int _Insure_trap_error(int a1, int a2, int a3, int a4, int a5, int a6) { static int (*fn)(); int ret; char *cmd; asprintf(&cmd, "/usr/X11R6/bin/xterm -display :0 -T Panic -n Panic -e /bin/sh -c 'cat /tmp/ierrs.*.%d ; gdb /proc/%d/exe %d'", getpid(), getpid(), getpid()); if (!fn) { static void *h; h = dlopen("/usr/local/parasoft/insure++lite/lib.linux2/libinsure.so", RTLD_LAZY); fn = dlsym(h, "_Insure_trap_error"); } ret = fn(a1, a2, a3, a4, a5, a6); system(cmd); free(cmd); return ret; } #endif