8e543a6360eb43f628336b20a85002e26f92163b
[rsync/rsync.git] / flist.c
1 /*
2  * Generate and receive file lists.
3  *
4  * Copyright (C) 1996 Andrew Tridgell
5  * Copyright (C) 1996 Paul Mackerras
6  * Copyright (C) 2001, 2002 Martin Pool <mbp@samba.org>
7  * Copyright (C) 2002, 2003, 2004, 2005, 2006 Wayne Davison
8  *
9  * This program is free software; you can redistribute it and/or modify
10  * it under the terms of the GNU General Public License as published by
11  * the Free Software Foundation; either version 2 of the License, or
12  * (at your option) any later version.
13  *
14  * This program is distributed in the hope that it will be useful,
15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
17  * GNU General Public License for more details.
18  *
19  * You should have received a copy of the GNU General Public License along
20  * with this program; if not, write to the Free Software Foundation, Inc.,
21  * 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.
22  */
23
24 #include "rsync.h"
25
26 extern int verbose;
27 extern int list_only;
28 extern int am_root;
29 extern int am_server;
30 extern int am_daemon;
31 extern int am_sender;
32 extern int do_progress;
33 extern int always_checksum;
34 extern int module_id;
35 extern int ignore_errors;
36 extern int numeric_ids;
37 extern int recurse;
38 extern int xfer_dirs;
39 extern int filesfrom_fd;
40 extern int one_file_system;
41 extern int copy_dirlinks;
42 extern int keep_dirlinks;
43 extern int preserve_links;
44 extern int preserve_hard_links;
45 extern int preserve_devices;
46 extern int preserve_specials;
47 extern int preserve_uid;
48 extern int preserve_gid;
49 extern int relative_paths;
50 extern int implied_dirs;
51 extern int ignore_perishable;
52 extern int non_perishable_cnt;
53 extern int prune_empty_dirs;
54 extern int copy_links;
55 extern int copy_unsafe_links;
56 extern int protocol_version;
57 extern int sanitize_paths;
58 extern struct stats stats;
59 extern struct file_list *the_file_list;
60
61 extern char curr_dir[MAXPATHLEN];
62
63 extern struct chmod_mode_struct *chmod_modes;
64
65 extern struct filter_list_struct filter_list;
66 extern struct filter_list_struct server_filter_list;
67
68 int io_error;
69 int checksum_len;
70 dev_t filesystem_dev; /* used to implement -x */
71 unsigned int file_struct_len;
72
73 static char empty_sum[MD4_SUM_LENGTH];
74 static int flist_count_offset;
75
76 static void clean_flist(struct file_list *flist, int strip_root, int no_dups);
77 static void output_flist(struct file_list *flist);
78
79 void init_flist(void)
80 {
81         struct file_struct f;
82
83         /* Figure out how big the file_struct is without trailing padding */
84         file_struct_len = offsetof(struct file_struct, flags) + sizeof f.flags;
85         checksum_len = protocol_version < 21 ? 2 : MD4_SUM_LENGTH;
86 }
87
88 static int show_filelist_p(void)
89 {
90         return verbose && xfer_dirs && !am_server;
91 }
92
93 static void start_filelist_progress(char *kind)
94 {
95         rprintf(FCLIENT, "%s ... ", kind);
96         if (verbose > 1 || do_progress)
97                 rprintf(FCLIENT, "\n");
98         rflush(FINFO);
99 }
100
101 static void emit_filelist_progress(int count)
102 {
103         rprintf(FCLIENT, " %d files...\r", count);
104 }
105
106 static void maybe_emit_filelist_progress(int count)
107 {
108         if (do_progress && show_filelist_p() && (count % 100) == 0)
109                 emit_filelist_progress(count);
110 }
111
112 static void finish_filelist_progress(const struct file_list *flist)
113 {
114         if (do_progress) {
115                 /* This overwrites the progress line */
116                 rprintf(FINFO, "%d file%sto consider\n",
117                         flist->count, flist->count == 1 ? " " : "s ");
118         } else
119                 rprintf(FINFO, "done\n");
120 }
121
122 void show_flist_stats(void)
123 {
124         /* Nothing yet */
125 }
126
127 static void list_file_entry(struct file_struct *f)
128 {
129         char permbuf[PERMSTRING_SIZE];
130
131         if (!f->basename) {
132                 /* this can happen if duplicate names were removed */
133                 return;
134         }
135
136         permstring(permbuf, f->mode);
137
138 #ifdef SUPPORT_LINKS
139         if (preserve_links && S_ISLNK(f->mode)) {
140                 rprintf(FINFO, "%s %11.0f %s %s -> %s\n",
141                         permbuf,
142                         (double)f->length, timestring(f->modtime),
143                         f_name(f, NULL), f->u.link);
144         } else
145 #endif
146         {
147                 rprintf(FINFO, "%s %11.0f %s %s\n",
148                         permbuf,
149                         (double)f->length, timestring(f->modtime),
150                         f_name(f, NULL));
151         }
152 }
153
154 /* Stat either a symlink or its referent, depending on the settings of
155  * copy_links, copy_unsafe_links, etc.  Returns -1 on error, 0 on success.
156  *
157  * If path is the name of a symlink, then the linkbuf buffer (which must hold
158  * MAXPATHLEN chars) will be set to the symlink's target string.
159  *
160  * The stat structure pointed to by stp will contain information about the
161  * link or the referent as appropriate, if they exist. */
162 static int readlink_stat(const char *path, STRUCT_STAT *stp, char *linkbuf)
163 {
164 #ifdef SUPPORT_LINKS
165         if (link_stat(path, stp, copy_dirlinks) < 0)
166                 return -1;
167         if (S_ISLNK(stp->st_mode)) {
168                 int llen = readlink(path, linkbuf, MAXPATHLEN - 1);
169                 if (llen < 0)
170                         return -1;
171                 linkbuf[llen] = '\0';
172                 if (copy_unsafe_links && unsafe_symlink(linkbuf, path)) {
173                         if (verbose > 1) {
174                                 rprintf(FINFO,"copying unsafe symlink \"%s\" -> \"%s\"\n",
175                                         path, linkbuf);
176                         }
177                         return do_stat(path, stp);
178                 }
179         }
180         return 0;
181 #else
182         return do_stat(path, stp);
183 #endif
184 }
185
186 int link_stat(const char *path, STRUCT_STAT *stp, int follow_dirlinks)
187 {
188 #ifdef SUPPORT_LINKS
189         if (copy_links)
190                 return do_stat(path, stp);
191         if (do_lstat(path, stp) < 0)
192                 return -1;
193         if (follow_dirlinks && S_ISLNK(stp->st_mode)) {
194                 STRUCT_STAT st;
195                 if (do_stat(path, &st) == 0 && S_ISDIR(st.st_mode))
196                         *stp = st;
197         }
198         return 0;
199 #else
200         return do_stat(path, stp);
201 #endif
202 }
203
204 /* This function is used to check if a file should be included/excluded
205  * from the list of files based on its name and type etc.  The value of
206  * filter_level is set to either SERVER_FILTERS or ALL_FILTERS. */
207 static int is_excluded(char *fname, int is_dir, int filter_level)
208 {
209 #if 0 /* This currently never happens, so avoid a useless compare. */
210         if (filter_level == NO_FILTERS)
211                 return 0;
212 #endif
213         if (fname) {
214                 /* never exclude '.', even if somebody does --exclude '*' */
215                 if (fname[0] == '.' && !fname[1])
216                         return 0;
217                 /* Handle the -R version of the '.' dir. */
218                 if (fname[0] == '/') {
219                         int len = strlen(fname);
220                         if (fname[len-1] == '.' && fname[len-2] == '/')
221                                 return 0;
222                 }
223         }
224         if (server_filter_list.head
225             && check_filter(&server_filter_list, fname, is_dir) < 0)
226                 return 1;
227         if (filter_level != ALL_FILTERS)
228                 return 0;
229         if (filter_list.head
230             && check_filter(&filter_list, fname, is_dir) < 0)
231                 return 1;
232         return 0;
233 }
234
235 static int to_wire_mode(mode_t mode)
236 {
237 #ifdef SUPPORT_LINKS
238 #if _S_IFLNK != 0120000
239         if (S_ISLNK(mode))
240                 return (mode & ~(_S_IFMT)) | 0120000;
241 #endif
242 #endif
243         return mode;
244 }
245
246 static mode_t from_wire_mode(int mode)
247 {
248 #if _S_IFLNK != 0120000
249         if ((mode & (_S_IFMT)) == 0120000)
250                 return (mode & ~(_S_IFMT)) | _S_IFLNK;
251 #endif
252         return mode;
253 }
254
255 static void send_directory(int f, struct file_list *flist,
256                            char *fbuf, int len);
257
258 static const char *flist_dir;
259 static int flist_dir_len;
260
261
262 /**
263  * Make sure @p flist is big enough to hold at least @p flist->count
264  * entries.
265  **/
266 void flist_expand(struct file_list *flist)
267 {
268         struct file_struct **new_ptr;
269
270         if (flist->count < flist->malloced)
271                 return;
272
273         if (flist->malloced < FLIST_START)
274                 flist->malloced = FLIST_START;
275         else if (flist->malloced >= FLIST_LINEAR)
276                 flist->malloced += FLIST_LINEAR;
277         else
278                 flist->malloced *= 2;
279
280         /*
281          * In case count jumped or we are starting the list
282          * with a known size just set it.
283          */
284         if (flist->malloced < flist->count)
285                 flist->malloced = flist->count;
286
287         new_ptr = realloc_array(flist->files, struct file_struct *,
288                                 flist->malloced);
289
290         if (verbose >= 2 && flist->malloced != FLIST_START) {
291                 rprintf(FCLIENT, "[%s] expand file_list to %.0f bytes, did%s move\n",
292                     who_am_i(),
293                     (double)sizeof flist->files[0] * flist->malloced,
294                     (new_ptr == flist->files) ? " not" : "");
295         }
296
297         flist->files = new_ptr;
298
299         if (!flist->files)
300                 out_of_memory("flist_expand");
301 }
302
303 static void send_file_entry(struct file_struct *file, int f)
304 {
305         unsigned short flags;
306         static time_t modtime;
307         static mode_t mode;
308         static int64 dev;
309         static dev_t rdev;
310         static uint32 rdev_major;
311         static uid_t uid;
312         static gid_t gid;
313         static char lastname[MAXPATHLEN];
314         char fname[MAXPATHLEN];
315         int l1, l2;
316
317         if (f < 0)
318                 return;
319
320         if (!file) {
321                 write_byte(f, 0);
322                 modtime = 0, mode = 0;
323                 dev = 0, rdev = MAKEDEV(0, 0);
324                 rdev_major = 0;
325                 uid = 0, gid = 0;
326                 *lastname = '\0';
327                 return;
328         }
329
330         f_name(file, fname);
331
332         flags = file->flags & XMIT_TOP_DIR;
333
334         if (file->mode == mode)
335                 flags |= XMIT_SAME_MODE;
336         else
337                 mode = file->mode;
338         if ((preserve_devices && IS_DEVICE(mode))
339          || (preserve_specials && IS_SPECIAL(mode))) {
340                 if (protocol_version < 28) {
341                         if (file->u.rdev == rdev)
342                                 flags |= XMIT_SAME_RDEV_pre28;
343                         else
344                                 rdev = file->u.rdev;
345                 } else {
346                         rdev = file->u.rdev;
347                         if ((uint32)major(rdev) == rdev_major)
348                                 flags |= XMIT_SAME_RDEV_MAJOR;
349                         else
350                                 rdev_major = major(rdev);
351                         if ((uint32)minor(rdev) <= 0xFFu)
352                                 flags |= XMIT_RDEV_MINOR_IS_SMALL;
353                 }
354         } else if (protocol_version < 28)
355                 rdev = MAKEDEV(0, 0);
356         if (file->uid == uid)
357                 flags |= XMIT_SAME_UID;
358         else
359                 uid = file->uid;
360         if (file->gid == gid)
361                 flags |= XMIT_SAME_GID;
362         else
363                 gid = file->gid;
364         if (file->modtime == modtime)
365                 flags |= XMIT_SAME_TIME;
366         else
367                 modtime = file->modtime;
368
369 #ifdef SUPPORT_HARD_LINKS
370         if (file->link_u.idev) {
371                 if (file->F_DEV == dev) {
372                         if (protocol_version >= 28)
373                                 flags |= XMIT_SAME_DEV;
374                 } else
375                         dev = file->F_DEV;
376                 flags |= XMIT_HAS_IDEV_DATA;
377         }
378 #endif
379
380         for (l1 = 0;
381             lastname[l1] && (fname[l1] == lastname[l1]) && (l1 < 255);
382             l1++) {}
383         l2 = strlen(fname+l1);
384
385         if (l1 > 0)
386                 flags |= XMIT_SAME_NAME;
387         if (l2 > 255)
388                 flags |= XMIT_LONG_NAME;
389
390         /* We must make sure we don't send a zero flag byte or the
391          * other end will terminate the flist transfer.  Note that
392          * the use of XMIT_TOP_DIR on a non-dir has no meaning, so
393          * it's harmless way to add a bit to the first flag byte. */
394         if (protocol_version >= 28) {
395                 if (!flags && !S_ISDIR(mode))
396                         flags |= XMIT_TOP_DIR;
397                 if ((flags & 0xFF00) || !flags) {
398                         flags |= XMIT_EXTENDED_FLAGS;
399                         write_byte(f, flags);
400                         write_byte(f, flags >> 8);
401                 } else
402                         write_byte(f, flags);
403         } else {
404                 if (!(flags & 0xFF))
405                         flags |= S_ISDIR(mode) ? XMIT_LONG_NAME : XMIT_TOP_DIR;
406                 write_byte(f, flags);
407         }
408         if (flags & XMIT_SAME_NAME)
409                 write_byte(f, l1);
410         if (flags & XMIT_LONG_NAME)
411                 write_int(f, l2);
412         else
413                 write_byte(f, l2);
414         write_buf(f, fname + l1, l2);
415
416         write_longint(f, file->length);
417         if (!(flags & XMIT_SAME_TIME))
418                 write_int(f, modtime);
419         if (!(flags & XMIT_SAME_MODE))
420                 write_int(f, to_wire_mode(mode));
421         if (preserve_uid && !(flags & XMIT_SAME_UID)) {
422                 if (!numeric_ids)
423                         add_uid(uid);
424                 write_int(f, uid);
425         }
426         if (preserve_gid && !(flags & XMIT_SAME_GID)) {
427                 if (!numeric_ids)
428                         add_gid(gid);
429                 write_int(f, gid);
430         }
431         if ((preserve_devices && IS_DEVICE(mode))
432          || (preserve_specials && IS_SPECIAL(mode))) {
433                 if (protocol_version < 28) {
434                         if (!(flags & XMIT_SAME_RDEV_pre28))
435                                 write_int(f, (int)rdev);
436                 } else {
437                         if (!(flags & XMIT_SAME_RDEV_MAJOR))
438                                 write_int(f, major(rdev));
439                         if (flags & XMIT_RDEV_MINOR_IS_SMALL)
440                                 write_byte(f, minor(rdev));
441                         else
442                                 write_int(f, minor(rdev));
443                 }
444         }
445
446 #ifdef SUPPORT_LINKS
447         if (preserve_links && S_ISLNK(mode)) {
448                 int len = strlen(file->u.link);
449                 write_int(f, len);
450                 write_buf(f, file->u.link, len);
451         }
452 #endif
453
454 #ifdef SUPPORT_HARD_LINKS
455         if (file->link_u.idev) {
456                 if (protocol_version < 26) {
457                         /* 32-bit dev_t and ino_t */
458                         write_int(f, dev);
459                         write_int(f, file->F_INODE);
460                 } else {
461                         /* 64-bit dev_t and ino_t */
462                         if (!(flags & XMIT_SAME_DEV))
463                                 write_longint(f, dev);
464                         write_longint(f, file->F_INODE);
465                 }
466         }
467 #endif
468
469         if (always_checksum && (S_ISREG(mode) || protocol_version < 28)) {
470                 char *sum;
471                 if (S_ISREG(mode))
472                         sum = file->u.sum;
473                 else {
474                         /* Prior to 28, we sent a useless set of nulls. */
475                         sum = empty_sum;
476                 }
477                 write_buf(f, sum, checksum_len);
478         }
479
480         strlcpy(lastname, fname, MAXPATHLEN);
481 }
482
483 static struct file_struct *receive_file_entry(struct file_list *flist,
484                                               unsigned short flags, int f)
485 {
486         static time_t modtime;
487         static mode_t mode;
488         static int64 dev;
489         static dev_t rdev;
490         static uint32 rdev_major;
491         static uid_t uid;
492         static gid_t gid;
493         static char lastname[MAXPATHLEN], *lastdir;
494         static int lastdir_depth, lastdir_len = -1;
495         static unsigned int del_hier_name_len = 0;
496         static int in_del_hier = 0;
497         char thisname[MAXPATHLEN];
498         unsigned int l1 = 0, l2 = 0;
499         int alloc_len, basename_len, dirname_len, linkname_len, sum_len;
500         OFF_T file_length;
501         char *basename, *dirname, *bp;
502         struct file_struct *file;
503
504         if (!flist) {
505                 modtime = 0, mode = 0;
506                 dev = 0, rdev = MAKEDEV(0, 0);
507                 rdev_major = 0;
508                 uid = 0, gid = 0;
509                 *lastname = '\0';
510                 lastdir_len = -1;
511                 in_del_hier = 0;
512                 return NULL;
513         }
514
515         if (flags & XMIT_SAME_NAME)
516                 l1 = read_byte(f);
517
518         if (flags & XMIT_LONG_NAME)
519                 l2 = read_int(f);
520         else
521                 l2 = read_byte(f);
522
523         if (l2 >= MAXPATHLEN - l1) {
524                 rprintf(FERROR,
525                         "overflow: flags=0x%x l1=%d l2=%d lastname=%s\n",
526                         flags, l1, l2, lastname);
527                 overflow_exit("receive_file_entry");
528         }
529
530         strlcpy(thisname, lastname, l1 + 1);
531         read_sbuf(f, &thisname[l1], l2);
532         thisname[l1 + l2] = 0;
533
534         strlcpy(lastname, thisname, MAXPATHLEN);
535
536         clean_fname(thisname, 0);
537
538         if (sanitize_paths)
539                 sanitize_path(thisname, thisname, "", 0, NULL);
540
541         if ((basename = strrchr(thisname, '/')) != NULL) {
542                 dirname_len = ++basename - thisname; /* counts future '\0' */
543                 if (lastdir_len == dirname_len - 1
544                     && strncmp(thisname, lastdir, lastdir_len) == 0) {
545                         dirname = lastdir;
546                         dirname_len = 0; /* indicates no copy is needed */
547                 } else
548                         dirname = thisname;
549         } else {
550                 basename = thisname;
551                 dirname = NULL;
552                 dirname_len = 0;
553         }
554         basename_len = strlen(basename) + 1; /* count the '\0' */
555
556         file_length = read_longint(f);
557         if (!(flags & XMIT_SAME_TIME))
558                 modtime = (time_t)read_int(f);
559         if (!(flags & XMIT_SAME_MODE))
560                 mode = from_wire_mode(read_int(f));
561
562         if (chmod_modes && !S_ISLNK(mode))
563                 mode = tweak_mode(mode, chmod_modes);
564
565         if (preserve_uid && !(flags & XMIT_SAME_UID))
566                 uid = (uid_t)read_int(f);
567         if (preserve_gid && !(flags & XMIT_SAME_GID))
568                 gid = (gid_t)read_int(f);
569
570         if ((preserve_devices && IS_DEVICE(mode))
571          || (preserve_specials && IS_SPECIAL(mode))) {
572                 if (protocol_version < 28) {
573                         if (!(flags & XMIT_SAME_RDEV_pre28))
574                                 rdev = (dev_t)read_int(f);
575                 } else {
576                         uint32 rdev_minor;
577                         if (!(flags & XMIT_SAME_RDEV_MAJOR))
578                                 rdev_major = read_int(f);
579                         if (flags & XMIT_RDEV_MINOR_IS_SMALL)
580                                 rdev_minor = read_byte(f);
581                         else
582                                 rdev_minor = read_int(f);
583                         rdev = MAKEDEV(rdev_major, rdev_minor);
584                 }
585         } else if (protocol_version < 28)
586                 rdev = MAKEDEV(0, 0);
587
588 #ifdef SUPPORT_LINKS
589         if (preserve_links && S_ISLNK(mode)) {
590                 linkname_len = read_int(f) + 1; /* count the '\0' */
591                 if (linkname_len <= 0 || linkname_len > MAXPATHLEN) {
592                         rprintf(FERROR, "overflow: linkname_len=%d\n",
593                                 linkname_len - 1);
594                         overflow_exit("receive_file_entry");
595                 }
596         }
597         else
598 #endif
599                 linkname_len = 0;
600
601         sum_len = always_checksum && S_ISREG(mode) ? MD4_SUM_LENGTH : 0;
602
603         alloc_len = file_struct_len + dirname_len + basename_len
604                   + linkname_len + sum_len;
605         bp = pool_alloc(flist->file_pool, alloc_len, "receive_file_entry");
606
607         file = (struct file_struct *)bp;
608         memset(bp, 0, file_struct_len);
609         bp += file_struct_len;
610
611         file->modtime = modtime;
612         file->length = file_length;
613         file->mode = mode;
614         file->uid = uid;
615         file->gid = gid;
616
617         if (dirname_len) {
618                 file->dirname = lastdir = bp;
619                 lastdir_len = dirname_len - 1;
620                 memcpy(bp, dirname, dirname_len - 1);
621                 bp += dirname_len;
622                 bp[-1] = '\0';
623                 lastdir_depth = count_dir_elements(lastdir);
624                 file->dir.depth = lastdir_depth + 1;
625         } else if (dirname) {
626                 file->dirname = dirname; /* we're reusing lastname */
627                 file->dir.depth = lastdir_depth + 1;
628         } else
629                 file->dir.depth = 1;
630
631         if (S_ISDIR(mode)) {
632                 if (basename_len == 1+1 && *basename == '.') /* +1 for '\0' */
633                         file->dir.depth--;
634                 if (flags & XMIT_TOP_DIR) {
635                         in_del_hier = recurse;
636                         del_hier_name_len = file->dir.depth == 0 ? 0 : l1 + l2;
637                         if (relative_paths && del_hier_name_len > 2
638                             && lastname[del_hier_name_len-1] == '.'
639                             && lastname[del_hier_name_len-2] == '/')
640                                 del_hier_name_len -= 2;
641                         file->flags |= FLAG_TOP_DIR | FLAG_DEL_HERE;
642                 } else if (in_del_hier) {
643                         if (!relative_paths || !del_hier_name_len
644                          || (l1 >= del_hier_name_len
645                           && lastname[del_hier_name_len] == '/'))
646                                 file->flags |= FLAG_DEL_HERE;
647                         else
648                                 in_del_hier = 0;
649                 }
650         }
651
652         file->basename = bp;
653         memcpy(bp, basename, basename_len);
654         bp += basename_len;
655
656         if ((preserve_devices && IS_DEVICE(mode))
657          || (preserve_specials && IS_SPECIAL(mode)))
658                 file->u.rdev = rdev;
659
660 #ifdef SUPPORT_LINKS
661         if (linkname_len) {
662                 file->u.link = bp;
663                 read_sbuf(f, bp, linkname_len - 1);
664                 if (sanitize_paths)
665                         sanitize_path(bp, bp, "", lastdir_depth, NULL);
666                 bp += linkname_len;
667         }
668 #endif
669
670 #ifdef SUPPORT_HARD_LINKS
671         if (preserve_hard_links && protocol_version < 28 && S_ISREG(mode))
672                 flags |= XMIT_HAS_IDEV_DATA;
673         if (flags & XMIT_HAS_IDEV_DATA) {
674                 int64 inode;
675                 if (protocol_version < 26) {
676                         dev = read_int(f);
677                         inode = read_int(f);
678                 } else {
679                         if (!(flags & XMIT_SAME_DEV))
680                                 dev = read_longint(f);
681                         inode = read_longint(f);
682                 }
683                 if (flist->hlink_pool) {
684                         file->link_u.idev = pool_talloc(flist->hlink_pool,
685                             struct idev, 1, "inode_table");
686                         file->F_INODE = inode;
687                         file->F_DEV = dev;
688                 }
689         }
690 #endif
691
692         if (always_checksum && (sum_len || protocol_version < 28)) {
693                 char *sum;
694                 if (sum_len) {
695                         file->u.sum = sum = bp;
696                         /*bp += sum_len;*/
697                 } else {
698                         /* Prior to 28, we get a useless set of nulls. */
699                         sum = empty_sum;
700                 }
701                 read_buf(f, sum, checksum_len);
702         }
703
704         return file;
705 }
706
707 /**
708  * Create a file_struct for a named file by reading its stat()
709  * information and performing extensive checks against global
710  * options.
711  *
712  * @return the new file, or NULL if there was an error or this file
713  * should be excluded.
714  *
715  * @todo There is a small optimization opportunity here to avoid
716  * stat()ing the file in some circumstances, which has a certain cost.
717  * We are called immediately after doing readdir(), and so we may
718  * already know the d_type of the file.  We could for example avoid
719  * statting directories if we're not recursing, but this is not a very
720  * important case.  Some systems may not have d_type.
721  **/
722 struct file_struct *make_file(char *fname, struct file_list *flist,
723                               STRUCT_STAT *stp, unsigned short flags,
724                               int filter_level)
725 {
726         static char *lastdir;
727         static int lastdir_len = -1;
728         struct file_struct *file;
729         STRUCT_STAT st;
730         char sum[SUM_LENGTH];
731         char thisname[MAXPATHLEN];
732         char linkname[MAXPATHLEN];
733         int alloc_len, basename_len, dirname_len, linkname_len, sum_len;
734         char *basename, *dirname, *bp;
735
736         if (!flist || !flist->count)    /* Ignore lastdir when invalid. */
737                 lastdir_len = -1;
738
739         if (strlcpy(thisname, fname, sizeof thisname)
740             >= sizeof thisname - flist_dir_len) {
741                 rprintf(FINFO, "skipping overly long name: %s\n", fname);
742                 return NULL;
743         }
744         clean_fname(thisname, 0);
745         if (sanitize_paths)
746                 sanitize_path(thisname, thisname, "", 0, NULL);
747
748         memset(sum, 0, SUM_LENGTH);
749
750         if (stp && S_ISDIR(stp->st_mode)) {
751                 st = *stp; /* Needed for "symlink/." with --relative. */
752                 *linkname = '\0'; /* make IBM code checker happy */
753         } else if (readlink_stat(thisname, &st, linkname) != 0) {
754                 int save_errno = errno;
755                 /* See if file is excluded before reporting an error. */
756                 if (filter_level != NO_FILTERS
757                  && (is_excluded(thisname, 0, filter_level)
758                   || is_excluded(thisname, 1, filter_level))) {
759                         if (ignore_perishable && save_errno != ENOENT)
760                                 non_perishable_cnt++;
761                         return NULL;
762                 }
763                 if (save_errno == ENOENT) {
764 #ifdef SUPPORT_LINKS
765                         /* Avoid "vanished" error if symlink points nowhere. */
766                         if (copy_links && do_lstat(thisname, &st) == 0
767                             && S_ISLNK(st.st_mode)) {
768                                 io_error |= IOERR_GENERAL;
769                                 rprintf(FERROR, "symlink has no referent: %s\n",
770                                         full_fname(thisname));
771                         } else
772 #endif
773                         {
774                                 enum logcode c = am_daemon && protocol_version < 28
775                                     ? FERROR : FINFO;
776                                 io_error |= IOERR_VANISHED;
777                                 rprintf(c, "file has vanished: %s\n",
778                                         full_fname(thisname));
779                         }
780                 } else {
781                         io_error |= IOERR_GENERAL;
782                         rsyserr(FERROR, save_errno, "readlink %s failed",
783                                 full_fname(thisname));
784                 }
785                 return NULL;
786         }
787
788         /* backup.c calls us with filter_level set to NO_FILTERS. */
789         if (filter_level == NO_FILTERS)
790                 goto skip_filters;
791
792         if (S_ISDIR(st.st_mode) && !xfer_dirs) {
793                 rprintf(FINFO, "skipping directory %s\n", thisname);
794                 return NULL;
795         }
796
797         /* We only care about directories because we need to avoid recursing
798          * into a mount-point directory, not to avoid copying a symlinked
799          * file if -L (or similar) was specified. */
800         if (one_file_system && st.st_dev != filesystem_dev
801          && S_ISDIR(st.st_mode)) {
802                 if (one_file_system > 1) {
803                         if (verbose > 2) {
804                                 rprintf(FINFO, "skipping mount-point dir %s\n",
805                                         thisname);
806                         }
807                         return NULL;
808                 }
809                 flags |= FLAG_MOUNT_POINT;
810         }
811
812         if (is_excluded(thisname, S_ISDIR(st.st_mode) != 0, filter_level)) {
813                 if (ignore_perishable)
814                         non_perishable_cnt++;
815                 return NULL;
816         }
817
818         if (lp_ignore_nonreadable(module_id)) {
819 #ifdef SUPPORT_LINKS
820                 if (!S_ISLNK(st.st_mode))
821 #endif
822                         if (access(thisname, R_OK) != 0)
823                                 return NULL;
824         }
825
826   skip_filters:
827
828         if (verbose > 2) {
829                 rprintf(FINFO, "[%s] make_file(%s,*,%d)\n",
830                         who_am_i(), thisname, filter_level);
831         }
832
833         if ((basename = strrchr(thisname, '/')) != NULL) {
834                 dirname_len = ++basename - thisname; /* counts future '\0' */
835                 if (lastdir_len == dirname_len - 1
836                     && strncmp(thisname, lastdir, lastdir_len) == 0) {
837                         dirname = lastdir;
838                         dirname_len = 0; /* indicates no copy is needed */
839                 } else
840                         dirname = thisname;
841         } else {
842                 basename = thisname;
843                 dirname = NULL;
844                 dirname_len = 0;
845         }
846         basename_len = strlen(basename) + 1; /* count the '\0' */
847
848 #ifdef SUPPORT_LINKS
849         linkname_len = S_ISLNK(st.st_mode) ? strlen(linkname) + 1 : 0;
850 #else
851         linkname_len = 0;
852 #endif
853
854         sum_len = always_checksum && am_sender && S_ISREG(st.st_mode)
855                 ? MD4_SUM_LENGTH : 0;
856
857         alloc_len = file_struct_len + dirname_len + basename_len
858                   + linkname_len + sum_len;
859         if (flist)
860                 bp = pool_alloc(flist->file_pool, alloc_len, "make_file");
861         else {
862                 if (!(bp = new_array(char, alloc_len)))
863                         out_of_memory("make_file");
864         }
865
866         file = (struct file_struct *)bp;
867         memset(bp, 0, file_struct_len);
868         bp += file_struct_len;
869
870         file->flags = flags;
871         file->modtime = st.st_mtime;
872         file->length = st.st_size;
873         file->mode = st.st_mode;
874         file->uid = st.st_uid;
875         file->gid = st.st_gid;
876
877 #ifdef SUPPORT_HARD_LINKS
878         if (flist && flist->hlink_pool) {
879                 if (protocol_version < 28) {
880                         if (S_ISREG(st.st_mode))
881                                 file->link_u.idev = pool_talloc(
882                                     flist->hlink_pool, struct idev, 1,
883                                     "inode_table");
884                 } else {
885                         if (!S_ISDIR(st.st_mode) && st.st_nlink > 1)
886                                 file->link_u.idev = pool_talloc(
887                                     flist->hlink_pool, struct idev, 1,
888                                     "inode_table");
889                 }
890         }
891         if (file->link_u.idev) {
892                 file->F_DEV = st.st_dev;
893                 file->F_INODE = st.st_ino;
894         }
895 #endif
896
897         if (dirname_len) {
898                 file->dirname = lastdir = bp;
899                 lastdir_len = dirname_len - 1;
900                 memcpy(bp, dirname, dirname_len - 1);
901                 bp += dirname_len;
902                 bp[-1] = '\0';
903         } else if (dirname)
904                 file->dirname = dirname;
905
906         file->basename = bp;
907         memcpy(bp, basename, basename_len);
908         bp += basename_len;
909
910 #ifdef HAVE_STRUCT_STAT_ST_RDEV
911         if ((preserve_devices && IS_DEVICE(st.st_mode))
912          || (preserve_specials && IS_SPECIAL(st.st_mode)))
913                 file->u.rdev = st.st_rdev;
914 #endif
915
916 #ifdef SUPPORT_LINKS
917         if (linkname_len) {
918                 file->u.link = bp;
919                 memcpy(bp, linkname, linkname_len);
920                 bp += linkname_len;
921         }
922 #endif
923
924         if (sum_len) {
925                 file->u.sum = bp;
926                 file_checksum(thisname, bp, st.st_size);
927                 /*bp += sum_len;*/
928         }
929
930         file->dir.root = flist_dir;
931
932         /* This code is only used by the receiver when it is building
933          * a list of files for a delete pass. */
934         if (keep_dirlinks && linkname_len && flist) {
935                 STRUCT_STAT st2;
936                 int save_mode = file->mode;
937                 file->mode = S_IFDIR; /* Find a directory with our name. */
938                 if (flist_find(the_file_list, file) >= 0
939                     && do_stat(thisname, &st2) == 0 && S_ISDIR(st2.st_mode)) {
940                         file->modtime = st2.st_mtime;
941                         file->length = st2.st_size;
942                         file->mode = st2.st_mode;
943                         file->uid = st2.st_uid;
944                         file->gid = st2.st_gid;
945                         file->u.link = NULL;
946                 } else
947                         file->mode = save_mode;
948         }
949
950         if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))
951                 stats.total_size += st.st_size;
952
953         return file;
954 }
955
956 static struct file_struct *send_file_name(int f, struct file_list *flist,
957                                           char *fname, STRUCT_STAT *stp,
958                                           unsigned short flags)
959 {
960         struct file_struct *file;
961
962         file = make_file(fname, flist, stp, flags,
963                          f == -2 ? SERVER_FILTERS : ALL_FILTERS);
964         if (!file)
965                 return NULL;
966
967         if (chmod_modes && !S_ISLNK(file->mode))
968                 file->mode = tweak_mode(file->mode, chmod_modes);
969
970         maybe_emit_filelist_progress(flist->count + flist_count_offset);
971
972         flist_expand(flist);
973
974         if (file->basename[0]) {
975                 flist->files[flist->count++] = file;
976                 send_file_entry(file, f);
977         }
978         return file;
979 }
980
981 static void send_if_directory(int f, struct file_list *flist,
982                               struct file_struct *file,
983                               char *fbuf, unsigned int ol)
984 {
985         char is_dot_dir = fbuf[ol-1] == '.' && (ol == 1 || fbuf[ol-2] == '/');
986
987         if (S_ISDIR(file->mode)
988             && !(file->flags & FLAG_MOUNT_POINT) && f_name(file, fbuf)) {
989                 void *save_filters;
990                 unsigned int len = strlen(fbuf);
991                 if (len > 1 && fbuf[len-1] == '/')
992                         fbuf[--len] = '\0';
993                 if (len >= MAXPATHLEN - 1) {
994                         io_error |= IOERR_GENERAL;
995                         rprintf(FERROR, "skipping long-named directory: %s\n",
996                                 full_fname(fbuf));
997                         return;
998                 }
999                 save_filters = push_local_filters(fbuf, len);
1000                 send_directory(f, flist, fbuf, len);
1001                 pop_local_filters(save_filters);
1002                 fbuf[ol] = '\0';
1003                 if (is_dot_dir)
1004                         fbuf[ol-1] = '.';
1005         }
1006 }
1007
1008 /* This function is normally called by the sender, but the receiving side also
1009  * calls it from get_dirlist() with f set to -1 so that we just construct the
1010  * file list in memory without sending it over the wire.  Also, get_dirlist()
1011  * might call this with f set to -2, which also indicates that local filter
1012  * rules should be ignored. */
1013 static void send_directory(int f, struct file_list *flist,
1014                            char *fbuf, int len)
1015 {
1016         struct dirent *di;
1017         unsigned remainder;
1018         char *p;
1019         DIR *d;
1020         int start = flist->count;
1021
1022         if (!(d = opendir(fbuf))) {
1023                 io_error |= IOERR_GENERAL;
1024                 rsyserr(FERROR, errno, "opendir %s failed", full_fname(fbuf));
1025                 return;
1026         }
1027
1028         p = fbuf + len;
1029         if (len != 1 || *fbuf != '/')
1030                 *p++ = '/';
1031         *p = '\0';
1032         remainder = MAXPATHLEN - (p - fbuf);
1033
1034         for (errno = 0, di = readdir(d); di; errno = 0, di = readdir(d)) {
1035                 char *dname = d_name(di);
1036                 if (dname[0] == '.' && (dname[1] == '\0'
1037                     || (dname[1] == '.' && dname[2] == '\0')))
1038                         continue;
1039                 if (strlcpy(p, dname, remainder) >= remainder) {
1040                         io_error |= IOERR_GENERAL;
1041                         rprintf(FINFO,
1042                                 "cannot send long-named file %s\n",
1043                                 full_fname(fbuf));
1044                         continue;
1045                 }
1046
1047                 send_file_name(f, flist, fbuf, NULL, 0);
1048         }
1049
1050         fbuf[len] = '\0';
1051
1052         if (errno) {
1053                 io_error |= IOERR_GENERAL;
1054                 rsyserr(FERROR, errno, "readdir(%s)", full_fname(fbuf));
1055         }
1056
1057         closedir(d);
1058
1059         if (recurse) {
1060                 int i, end = flist->count - 1;
1061                 for (i = start; i <= end; i++)
1062                         send_if_directory(f, flist, flist->files[i], fbuf, len);
1063         }
1064 }
1065
1066 struct file_list *send_file_list(int f, int argc, char *argv[])
1067 {
1068         int len;
1069         STRUCT_STAT st;
1070         char *p, *dir, olddir[sizeof curr_dir];
1071         char lastpath[MAXPATHLEN] = "";
1072         struct file_list *flist;
1073         struct timeval start_tv, end_tv;
1074         int64 start_write;
1075         int use_ff_fd = 0;
1076
1077         rprintf(FLOG, "building file list\n");
1078         if (show_filelist_p())
1079                 start_filelist_progress("building file list");
1080
1081         start_write = stats.total_written;
1082         gettimeofday(&start_tv, NULL);
1083
1084         flist = flist_new(WITH_HLINK, "send_file_list");
1085
1086         io_start_buffering_out();
1087         if (filesfrom_fd >= 0) {
1088                 if (argv[0] && !push_dir(argv[0], 0)) {
1089                         rsyserr(FERROR, errno, "push_dir %s failed",
1090                                 full_fname(argv[0]));
1091                         exit_cleanup(RERR_FILESELECT);
1092                 }
1093                 use_ff_fd = 1;
1094         }
1095
1096         while (1) {
1097                 char fbuf[MAXPATHLEN];
1098                 char *fn;
1099                 int is_dot_dir;
1100
1101                 if (use_ff_fd) {
1102                         if (read_filesfrom_line(filesfrom_fd, fbuf) == 0)
1103                                 break;
1104                         sanitize_path(fbuf, fbuf, "", 0, NULL);
1105                 } else {
1106                         if (argc-- == 0)
1107                                 break;
1108                         strlcpy(fbuf, *argv++, MAXPATHLEN);
1109                         if (sanitize_paths)
1110                                 sanitize_path(fbuf, fbuf, "", 0, NULL);
1111                 }
1112
1113                 len = strlen(fbuf);
1114                 if (relative_paths) {
1115                         /* We clean up fbuf below. */
1116                         is_dot_dir = 0;
1117                 } else if (!len || fbuf[len - 1] == '/') {
1118                         if (len == 2 && fbuf[0] == '.') {
1119                                 /* Turn "./" into just "." rather than "./." */
1120                                 fbuf[1] = '\0';
1121                         } else {
1122                                 if (len + 1 >= MAXPATHLEN)
1123                                         overflow_exit("send_file_list");
1124                                 fbuf[len++] = '.';
1125                                 fbuf[len] = '\0';
1126                         }
1127                         is_dot_dir = 1;
1128                 } else if (len > 1 && fbuf[len-1] == '.' && fbuf[len-2] == '.'
1129                     && (len == 2 || fbuf[len-3] == '/')) {
1130                         if (len + 2 >= MAXPATHLEN)
1131                                 overflow_exit("send_file_list");
1132                         fbuf[len++] = '/';
1133                         fbuf[len++] = '.';
1134                         fbuf[len] = '\0';
1135                         is_dot_dir = 1;
1136                 } else {
1137                         is_dot_dir = fbuf[len-1] == '.'
1138                                    && (len == 1 || fbuf[len-2] == '/');
1139                 }
1140
1141                 if (link_stat(fbuf, &st, copy_dirlinks) != 0) {
1142                         io_error |= IOERR_GENERAL;
1143                         rsyserr(FERROR, errno, "link_stat %s failed",
1144                                 full_fname(fbuf));
1145                         continue;
1146                 }
1147
1148                 if (S_ISDIR(st.st_mode) && !xfer_dirs) {
1149                         rprintf(FINFO, "skipping directory %s\n", fbuf);
1150                         continue;
1151                 }
1152
1153                 dir = NULL;
1154                 olddir[0] = '\0';
1155
1156                 if (!relative_paths) {
1157                         p = strrchr(fbuf, '/');
1158                         if (p) {
1159                                 *p = '\0';
1160                                 if (p == fbuf)
1161                                         dir = "/";
1162                                 else
1163                                         dir = fbuf;
1164                                 len -= p - fbuf + 1;
1165                                 fn = p + 1;
1166                         } else
1167                                 fn = fbuf;
1168                 } else {
1169                         if ((p = strstr(fbuf, "/./")) != NULL) {
1170                                 *p = '\0';
1171                                 if (p == fbuf)
1172                                         dir = "/";
1173                                 else
1174                                         dir = fbuf;
1175                                 len -= p - fbuf + 3;
1176                                 fn = p + 3;
1177                         } else
1178                                 fn = fbuf;
1179                         /* Get rid of trailing "/" and "/.". */
1180                         while (len) {
1181                                 if (fn[len - 1] == '/') {
1182                                         is_dot_dir = 1;
1183                                         if (!--len && !dir) {
1184                                                 len++;
1185                                                 break;
1186                                         }
1187                                 }
1188                                 else if (len >= 2 && fn[len - 1] == '.'
1189                                                   && fn[len - 2] == '/') {
1190                                         is_dot_dir = 1;
1191                                         if (!(len -= 2) && !dir) {
1192                                                 len++;
1193                                                 break;
1194                                         }
1195                                 } else
1196                                         break;
1197                         }
1198                         if (len == 1 && fn[0] == '/')
1199                                 fn[len++] = '.';
1200                         fn[len] = '\0';
1201                         /* Reject a ".." dir in the active part of the path. */
1202                         for (p = fn; (p = strstr(p, "..")) != NULL; p += 2) {
1203                                 if ((p[2] == '/' || p[2] == '\0')
1204                                  && (p == fn || p[-1] == '/')) {
1205                                         rprintf(FERROR,
1206                                             "found \"..\" dir in relative path: %s\n",
1207                                             fbuf);
1208                                         exit_cleanup(RERR_SYNTAX);
1209                                 }
1210                         }
1211                 }
1212
1213                 if (!*fn) {
1214                         len = 1;
1215                         fn = ".";
1216                 }
1217
1218                 if (dir && *dir) {
1219                         static const char *lastdir;
1220                         static int lastdir_len;
1221
1222                         strlcpy(olddir, curr_dir, sizeof olddir);
1223
1224                         if (!push_dir(dir, 0)) {
1225                                 io_error |= IOERR_GENERAL;
1226                                 rsyserr(FERROR, errno, "push_dir %s failed",
1227                                         full_fname(dir));
1228                                 continue;
1229                         }
1230
1231                         if (lastdir && strcmp(lastdir, dir) == 0) {
1232                                 flist_dir = lastdir;
1233                                 flist_dir_len = lastdir_len;
1234                         } else {
1235                                 flist_dir = lastdir = strdup(dir);
1236                                 flist_dir_len = lastdir_len = strlen(dir);
1237                         }
1238                 }
1239
1240                 if (fn != fbuf)
1241                         memmove(fbuf, fn, len + 1);
1242
1243                 if (implied_dirs && (p=strrchr(fbuf,'/')) && p != fbuf) {
1244                         /* Send the implied directories at the start of the
1245                          * source spec, so we get their permissions right. */
1246                         char *lp = lastpath, *slash = fbuf;
1247                         *p = '\0';
1248                         /* Skip any initial directories in our path that we
1249                          * have in common with lastpath. */
1250                         for (fn = fbuf; *fn && *lp == *fn; lp++, fn++) {
1251                                 if (*fn == '/')
1252                                         slash = fn;
1253                         }
1254                         *p = '/';
1255                         if (fn != p || (*lp && *lp != '/')) {
1256                                 int save_copy_links = copy_links;
1257                                 int save_xfer_dirs = xfer_dirs;
1258                                 copy_links |= copy_unsafe_links;
1259                                 xfer_dirs = 1;
1260                                 while ((slash = strchr(slash+1, '/')) != 0) {
1261                                         *slash = '\0';
1262                                         send_file_name(f, flist, fbuf, NULL, 0);
1263                                         *slash = '/';
1264                                 }
1265                                 copy_links = save_copy_links;
1266                                 xfer_dirs = save_xfer_dirs;
1267                                 *p = '\0';
1268                                 strlcpy(lastpath, fbuf, sizeof lastpath);
1269                                 *p = '/';
1270                         }
1271                 }
1272
1273                 if (one_file_system)
1274                         filesystem_dev = st.st_dev;
1275
1276                 if (recurse || (xfer_dirs && is_dot_dir)) {
1277                         struct file_struct *file;
1278                         file = send_file_name(f, flist, fbuf, &st, FLAG_TOP_DIR);
1279                         if (file)
1280                                 send_if_directory(f, flist, file, fbuf, len);
1281                 } else
1282                         send_file_name(f, flist, fbuf, &st, 0);
1283
1284                 if (olddir[0]) {
1285                         flist_dir = NULL;
1286                         flist_dir_len = 0;
1287                         if (!pop_dir(olddir)) {
1288                                 rsyserr(FERROR, errno, "pop_dir %s failed",
1289                                         full_fname(olddir));
1290                                 exit_cleanup(RERR_FILESELECT);
1291                         }
1292                 }
1293         }
1294
1295         gettimeofday(&end_tv, NULL);
1296         stats.flist_buildtime = (int64)(end_tv.tv_sec - start_tv.tv_sec) * 1000
1297                               + (end_tv.tv_usec - start_tv.tv_usec) / 1000;
1298         if (stats.flist_buildtime == 0)
1299                 stats.flist_buildtime = 1;
1300         start_tv = end_tv;
1301
1302         send_file_entry(NULL, f);
1303
1304         if (show_filelist_p())
1305                 finish_filelist_progress(flist);
1306
1307         gettimeofday(&end_tv, NULL);
1308         stats.flist_xfertime = (int64)(end_tv.tv_sec - start_tv.tv_sec) * 1000
1309                              + (end_tv.tv_usec - start_tv.tv_usec) / 1000;
1310
1311         if (flist->hlink_pool) {
1312                 pool_destroy(flist->hlink_pool);
1313                 flist->hlink_pool = NULL;
1314         }
1315
1316         /* Sort the list without removing any duplicates.  This allows the
1317          * receiving side to ask for any name they like, which gives us the
1318          * flexibility to change the way we unduplicate names in the future
1319          * without causing a compatibility problem with older versions. */
1320         clean_flist(flist, 0, 0);
1321
1322         send_uid_list(f);
1323
1324         /* send the io_error flag */
1325         write_int(f, lp_ignore_errors(module_id) ? 0 : io_error);
1326
1327         io_end_buffering();
1328         stats.flist_size = stats.total_written - start_write;
1329         stats.num_files = flist->count;
1330
1331         if (verbose > 3)
1332                 output_flist(flist);
1333
1334         if (verbose > 2)
1335                 rprintf(FINFO, "send_file_list done\n");
1336
1337         return flist;
1338 }
1339
1340 struct file_list *recv_file_list(int f)
1341 {
1342         struct file_list *flist;
1343         unsigned short flags;
1344         int64 start_read;
1345
1346         rprintf(FLOG, "receiving file list\n");
1347         if (show_filelist_p())
1348                 start_filelist_progress("receiving file list");
1349
1350         start_read = stats.total_read;
1351
1352         flist = flist_new(WITH_HLINK, "recv_file_list");
1353
1354         flist->count = 0;
1355         flist->malloced = 1000;
1356         flist->files = new_array(struct file_struct *, flist->malloced);
1357         if (!flist->files)
1358                 goto oom;
1359
1360         while ((flags = read_byte(f)) != 0) {
1361                 struct file_struct *file;
1362
1363                 flist_expand(flist);
1364
1365                 if (protocol_version >= 28 && (flags & XMIT_EXTENDED_FLAGS))
1366                         flags |= read_byte(f) << 8;
1367                 file = receive_file_entry(flist, flags, f);
1368
1369                 if (S_ISREG(file->mode) || S_ISLNK(file->mode))
1370                         stats.total_size += file->length;
1371
1372                 flist->files[flist->count++] = file;
1373
1374                 maybe_emit_filelist_progress(flist->count);
1375
1376                 if (verbose > 2) {
1377                         rprintf(FINFO, "recv_file_name(%s)\n",
1378                                 f_name(file, NULL));
1379                 }
1380         }
1381         receive_file_entry(NULL, 0, 0); /* Signal that we're done. */
1382
1383         if (verbose > 2)
1384                 rprintf(FINFO, "received %d names\n", flist->count);
1385
1386         if (show_filelist_p())
1387                 finish_filelist_progress(flist);
1388
1389         clean_flist(flist, relative_paths, 1);
1390
1391         if (f >= 0) {
1392                 recv_uid_list(f, flist);
1393
1394                 /* Recv the io_error flag */
1395                 if (lp_ignore_errors(module_id) || ignore_errors)
1396                         read_int(f);
1397                 else
1398                         io_error |= read_int(f);
1399         }
1400
1401         if (verbose > 3)
1402                 output_flist(flist);
1403
1404         if (list_only) {
1405                 int i;
1406                 for (i = 0; i < flist->count; i++)
1407                         list_file_entry(flist->files[i]);
1408         }
1409
1410         if (verbose > 2)
1411                 rprintf(FINFO, "recv_file_list done\n");
1412
1413         stats.flist_size = stats.total_read - start_read;
1414         stats.num_files = flist->count;
1415
1416         return flist;
1417
1418   oom:
1419         out_of_memory("recv_file_list");
1420         return NULL;            /* not reached */
1421 }
1422
1423 static int file_compare(struct file_struct **file1, struct file_struct **file2)
1424 {
1425         return f_name_cmp(*file1, *file2);
1426 }
1427
1428 /* Search for an identically-named item in the file list.  Note that the
1429  * items must agree in their directory-ness, or no match is returned. */
1430 int flist_find(struct file_list *flist, struct file_struct *f)
1431 {
1432         int low = flist->low, high = flist->high;
1433         int diff, mid, mid_up;
1434
1435         while (low <= high) {
1436                 mid = (low + high) / 2;
1437                 if (flist->files[mid]->basename)
1438                         mid_up = mid;
1439                 else {
1440                         /* Scan for the next non-empty entry using the cached
1441                          * distance values.  If the value isn't fully up-to-
1442                          * date, update it. */
1443                         mid_up = mid + flist->files[mid]->dir.depth;
1444                         if (!flist->files[mid_up]->basename) {
1445                                 do {
1446                                     mid_up += flist->files[mid_up]->dir.depth;
1447                                 } while (!flist->files[mid_up]->basename);
1448                                 flist->files[mid]->dir.depth = mid_up - mid;
1449                         }
1450                         if (mid_up > high) {
1451                                 /* If there's nothing left above us, set high to
1452                                  * a non-empty entry below us and continue. */
1453                                 high = mid - flist->files[mid]->length;
1454                                 if (!flist->files[high]->basename) {
1455                                         do {
1456                                             high -= flist->files[high]->length;
1457                                         } while (!flist->files[high]->basename);
1458                                         flist->files[mid]->length = mid - high;
1459                                 }
1460                                 continue;
1461                         }
1462                 }
1463                 diff = f_name_cmp(flist->files[mid_up], f);
1464                 if (diff == 0) {
1465                         if (protocol_version < 29
1466                             && S_ISDIR(flist->files[mid_up]->mode)
1467                             != S_ISDIR(f->mode))
1468                                 return -1;
1469                         return mid_up;
1470                 }
1471                 if (diff < 0)
1472                         low = mid_up + 1;
1473                 else
1474                         high = mid - 1;
1475         }
1476         return -1;
1477 }
1478
1479 /*
1480  * Free up any resources a file_struct has allocated
1481  * and clear the file.
1482  */
1483 void clear_file(struct file_struct *file, struct file_list *flist)
1484 {
1485         if (flist->hlink_pool && file->link_u.idev)
1486                 pool_free(flist->hlink_pool, 0, file->link_u.idev);
1487         memset(file, 0, file_struct_len);
1488         /* In an empty entry, dir.depth is an offset to the next non-empty
1489          * entry.  Likewise for length in the opposite direction.  We assume
1490          * that we're alone for now since flist_find() will adjust the counts
1491          * it runs into that aren't up-to-date. */
1492         file->length = file->dir.depth = 1;
1493 }
1494
1495 /*
1496  * allocate a new file list
1497  */
1498 struct file_list *flist_new(int with_hlink, char *msg)
1499 {
1500         struct file_list *flist;
1501
1502         flist = new(struct file_list);
1503         if (!flist)
1504                 out_of_memory(msg);
1505
1506         memset(flist, 0, sizeof (struct file_list));
1507
1508         if (!(flist->file_pool = pool_create(FILE_EXTENT, 0,
1509             out_of_memory, POOL_INTERN)))
1510                 out_of_memory(msg);
1511
1512 #ifdef SUPPORT_HARD_LINKS
1513         if (with_hlink && preserve_hard_links) {
1514                 if (!(flist->hlink_pool = pool_create(HLINK_EXTENT,
1515                     sizeof (struct idev), out_of_memory, POOL_INTERN)))
1516                         out_of_memory(msg);
1517         }
1518 #endif
1519
1520         return flist;
1521 }
1522
1523 /*
1524  * free up all elements in a flist
1525  */
1526 void flist_free(struct file_list *flist)
1527 {
1528         pool_destroy(flist->file_pool);
1529         pool_destroy(flist->hlink_pool);
1530         free(flist->files);
1531         free(flist);
1532 }
1533
1534 /*
1535  * This routine ensures we don't have any duplicate names in our file list.
1536  * duplicate names can cause corruption because of the pipelining
1537  */
1538 static void clean_flist(struct file_list *flist, int strip_root, int no_dups)
1539 {
1540         char fbuf[MAXPATHLEN];
1541         int i, prev_i = 0;
1542
1543         if (!flist)
1544                 return;
1545         if (flist->count == 0) {
1546                 flist->high = -1;
1547                 return;
1548         }
1549
1550         qsort(flist->files, flist->count,
1551             sizeof flist->files[0], (int (*)())file_compare);
1552
1553         for (i = no_dups? 0 : flist->count; i < flist->count; i++) {
1554                 if (flist->files[i]->basename) {
1555                         prev_i = i;
1556                         break;
1557                 }
1558         }
1559         flist->low = prev_i;
1560         while (++i < flist->count) {
1561                 int j;
1562                 struct file_struct *file = flist->files[i];
1563
1564                 if (!file->basename)
1565                         continue;
1566                 if (f_name_cmp(file, flist->files[prev_i]) == 0)
1567                         j = prev_i;
1568                 else if (protocol_version >= 29 && S_ISDIR(file->mode)) {
1569                         int save_mode = file->mode;
1570                         /* Make sure that this directory doesn't duplicate a
1571                          * non-directory earlier in the list. */
1572                         flist->high = prev_i;
1573                         file->mode = S_IFREG;
1574                         j = flist_find(flist, file);
1575                         file->mode = save_mode;
1576                 } else
1577                         j = -1;
1578                 if (j >= 0) {
1579                         struct file_struct *fp = flist->files[j];
1580                         int keep, drop;
1581                         /* If one is a dir and the other is not, we want to
1582                          * keep the dir because it might have contents in the
1583                          * list. */
1584                         if (S_ISDIR(file->mode) != S_ISDIR(fp->mode)) {
1585                                 if (S_ISDIR(file->mode))
1586                                         keep = i, drop = j;
1587                                 else
1588                                         keep = j, drop = i;
1589                         } else
1590                                 keep = j, drop = i;
1591                         if (verbose > 1 && !am_server) {
1592                                 rprintf(FINFO,
1593                                         "removing duplicate name %s from file list (%d)\n",
1594                                         f_name(file, fbuf), drop);
1595                         }
1596                         /* Make sure we don't lose track of a user-specified
1597                          * top directory. */
1598                         flist->files[keep]->flags |= flist->files[drop]->flags
1599                                                    & (FLAG_TOP_DIR|FLAG_DEL_HERE);
1600
1601                         clear_file(flist->files[drop], flist);
1602
1603                         if (keep == i) {
1604                                 if (flist->low == drop) {
1605                                         for (j = drop + 1;
1606                                              j < i && !flist->files[j]->basename;
1607                                              j++) {}
1608                                         flist->low = j;
1609                                 }
1610                                 prev_i = i;
1611                         }
1612                 } else
1613                         prev_i = i;
1614         }
1615         flist->high = no_dups ? prev_i : flist->count - 1;
1616
1617         if (strip_root) {
1618                 /* We need to strip off the leading slashes for relative
1619                  * paths, but this must be done _after_ the sorting phase. */
1620                 for (i = flist->low; i <= flist->high; i++) {
1621                         struct file_struct *file = flist->files[i];
1622
1623                         if (!file->dirname)
1624                                 continue;
1625                         while (*file->dirname == '/')
1626                                 file->dirname++;
1627                         if (!*file->dirname)
1628                                 file->dirname = NULL;
1629                 }
1630         }
1631
1632         if (prune_empty_dirs && no_dups) {
1633                 int j, prev_depth = 0;
1634
1635                 prev_i = 0; /* It's OK that this isn't really true. */
1636
1637                 for (i = flist->low; i <= flist->high; i++) {
1638                         struct file_struct *fp, *file = flist->files[i];
1639
1640                         /* This temporarily abuses the dir.depth value for a
1641                          * directory that is in a chain that might get pruned.
1642                          * We restore the old value if it gets a reprieve. */
1643                         if (S_ISDIR(file->mode) && file->dir.depth) {
1644                                 /* Dump empty dirs when coming back down. */
1645                                 for (j = prev_depth; j >= file->dir.depth; j--) {
1646                                         fp = flist->files[prev_i];
1647                                         if (fp->dir.depth >= 0)
1648                                                 break;
1649                                         prev_i = -fp->dir.depth-1;
1650                                         clear_file(fp, flist);
1651                                 }
1652                                 prev_depth = file->dir.depth;
1653                                 if (is_excluded(f_name(file, fbuf), 1,
1654                                                        ALL_FILTERS)) {
1655                                         /* Keep dirs through this dir. */
1656                                         for (j = prev_depth-1; ; j--) {
1657                                                 fp = flist->files[prev_i];
1658                                                 if (fp->dir.depth >= 0)
1659                                                         break;
1660                                                 prev_i = -fp->dir.depth-1;
1661                                                 fp->dir.depth = j;
1662                                         }
1663                                 } else
1664                                         file->dir.depth = -prev_i-1;
1665                                 prev_i = i;
1666                         } else {
1667                                 /* Keep dirs through this non-dir. */
1668                                 for (j = prev_depth; ; j--) {
1669                                         fp = flist->files[prev_i];
1670                                         if (fp->dir.depth >= 0)
1671                                                 break;
1672                                         prev_i = -fp->dir.depth-1;
1673                                         fp->dir.depth = j;
1674                                 }
1675                         }
1676                 }
1677                 /* Dump empty all remaining empty dirs. */
1678                 while (1) {
1679                         struct file_struct *fp = flist->files[prev_i];
1680                         if (fp->dir.depth >= 0)
1681                                 break;
1682                         prev_i = -fp->dir.depth-1;
1683                         clear_file(fp, flist);
1684                 }
1685
1686                 for (i = flist->low; i <= flist->high; i++) {
1687                         if (flist->files[i]->basename)
1688                                 break;
1689                 }
1690                 flist->low = i;
1691                 for (i = flist->high; i >= flist->low; i--) {
1692                         if (flist->files[i]->basename)
1693                                 break;
1694                 }
1695                 flist->high = i;
1696         }
1697 }
1698
1699 static void output_flist(struct file_list *flist)
1700 {
1701         char uidbuf[16], gidbuf[16], depthbuf[16];
1702         struct file_struct *file;
1703         const char *who = who_am_i();
1704         int i;
1705
1706         for (i = 0; i < flist->count; i++) {
1707                 file = flist->files[i];
1708                 if ((am_root || am_sender) && preserve_uid)
1709                         snprintf(uidbuf, sizeof uidbuf, " uid=%ld", (long)file->uid);
1710                 else
1711                         *uidbuf = '\0';
1712                 if (preserve_gid && file->gid != GID_NONE)
1713                         snprintf(gidbuf, sizeof gidbuf, " gid=%ld", (long)file->gid);
1714                 else
1715                         *gidbuf = '\0';
1716                 if (!am_sender)
1717                         snprintf(depthbuf, sizeof depthbuf, "%d", file->dir.depth);
1718                 rprintf(FINFO, "[%s] i=%d %s %s%s%s%s mode=0%o len=%.0f%s%s flags=%x\n",
1719                         who, i, am_sender ? NS(file->dir.root) : depthbuf,
1720                         file->dirname ? file->dirname : "",
1721                         file->dirname ? "/" : "", NS(file->basename),
1722                         S_ISDIR(file->mode) ? "/" : "", (int)file->mode,
1723                         (double)file->length, uidbuf, gidbuf, file->flags);
1724         }
1725 }
1726
1727 enum fnc_state { s_DIR, s_SLASH, s_BASE, s_TRAILING };
1728 enum fnc_type { t_PATH, t_ITEM };
1729
1730 /* Compare the names of two file_struct entities, similar to how strcmp()
1731  * would do if it were operating on the joined strings.
1732  *
1733  * Some differences beginning with protocol_version 29: (1) directory names
1734  * are compared with an assumed trailing slash so that they compare in a
1735  * way that would cause them to sort immediately prior to any content they
1736  * may have; (2) a directory of any name compares after a non-directory of
1737  * any name at the same depth; (3) a directory with name "." compares prior
1738  * to anything else.  These changes mean that a directory and a non-dir
1739  * with the same name will not compare as equal (protocol_version >= 29).
1740  *
1741  * The dirname component can be an empty string, but the basename component
1742  * cannot (and never is in the current codebase).  The basename component
1743  * may be NULL (for a removed item), in which case it is considered to be
1744  * after any existing item. */
1745 int f_name_cmp(struct file_struct *f1, struct file_struct *f2)
1746 {
1747         int dif;
1748         const uchar *c1, *c2;
1749         enum fnc_state state1, state2;
1750         enum fnc_type type1, type2;
1751         enum fnc_type t_path = protocol_version >= 29 ? t_PATH : t_ITEM;
1752
1753         if (!f1 || !f1->basename) {
1754                 if (!f2 || !f2->basename)
1755                         return 0;
1756                 return -1;
1757         }
1758         if (!f2 || !f2->basename)
1759                 return 1;
1760
1761         c1 = (uchar*)f1->dirname;
1762         c2 = (uchar*)f2->dirname;
1763         if (c1 == c2)
1764                 c1 = c2 = NULL;
1765         if (!c1) {
1766                 type1 = S_ISDIR(f1->mode) ? t_path : t_ITEM;
1767                 c1 = (uchar*)f1->basename;
1768                 if (type1 == t_PATH && *c1 == '.' && !c1[1]) {
1769                         type1 = t_ITEM;
1770                         state1 = s_TRAILING;
1771                         c1 = (uchar*)"";
1772                 } else
1773                         state1 = s_BASE;
1774         } else {
1775                 type1 = t_path;
1776                 state1 = s_DIR;
1777         }
1778         if (!c2) {
1779                 type2 = S_ISDIR(f2->mode) ? t_path : t_ITEM;
1780                 c2 = (uchar*)f2->basename;
1781                 if (type2 == t_PATH && *c2 == '.' && !c2[1]) {
1782                         type2 = t_ITEM;
1783                         state2 = s_TRAILING;
1784                         c2 = (uchar*)"";
1785                 } else
1786                         state2 = s_BASE;
1787         } else {
1788                 type2 = t_path;
1789                 state2 = s_DIR;
1790         }
1791
1792         if (type1 != type2)
1793                 return type1 == t_PATH ? 1 : -1;
1794
1795         do {
1796                 if (!*c1) {
1797                         switch (state1) {
1798                         case s_DIR:
1799                                 state1 = s_SLASH;
1800                                 c1 = (uchar*)"/";
1801                                 break;
1802                         case s_SLASH:
1803                                 type1 = S_ISDIR(f1->mode) ? t_path : t_ITEM;
1804                                 c1 = (uchar*)f1->basename;
1805                                 if (type1 == t_PATH && *c1 == '.' && !c1[1]) {
1806                                         type1 = t_ITEM;
1807                                         state1 = s_TRAILING;
1808                                         c1 = (uchar*)"";
1809                                 } else
1810                                         state1 = s_BASE;
1811                                 break;
1812                         case s_BASE:
1813                                 state1 = s_TRAILING;
1814                                 if (type1 == t_PATH) {
1815                                         c1 = (uchar*)"/";
1816                                         break;
1817                                 }
1818                                 /* FALL THROUGH */
1819                         case s_TRAILING:
1820                                 type1 = t_ITEM;
1821                                 break;
1822                         }
1823                         if (*c2 && type1 != type2)
1824                                 return type1 == t_PATH ? 1 : -1;
1825                 }
1826                 if (!*c2) {
1827                         switch (state2) {
1828                         case s_DIR:
1829                                 state2 = s_SLASH;
1830                                 c2 = (uchar*)"/";
1831                                 break;
1832                         case s_SLASH:
1833                                 type2 = S_ISDIR(f2->mode) ? t_path : t_ITEM;
1834                                 c2 = (uchar*)f2->basename;
1835                                 if (type2 == t_PATH && *c2 == '.' && !c2[1]) {
1836                                         type2 = t_ITEM;
1837                                         state2 = s_TRAILING;
1838                                         c2 = (uchar*)"";
1839                                 } else
1840                                         state2 = s_BASE;
1841                                 break;
1842                         case s_BASE:
1843                                 state2 = s_TRAILING;
1844                                 if (type2 == t_PATH) {
1845                                         c2 = (uchar*)"/";
1846                                         break;
1847                                 }
1848                                 /* FALL THROUGH */
1849                         case s_TRAILING:
1850                                 if (!*c1)
1851                                         return 0;
1852                                 type2 = t_ITEM;
1853                                 break;
1854                         }
1855                         if (type1 != type2)
1856                                 return type1 == t_PATH ? 1 : -1;
1857                 }
1858         } while ((dif = (int)*c1++ - (int)*c2++) == 0);
1859
1860         return dif;
1861 }
1862
1863 /* Return a copy of the full filename of a flist entry, using the indicated
1864  * buffer or one of 5 static buffers if fbuf is NULL.  No size-checking is
1865  * done because we checked the size when creating the file_struct entry.
1866  */
1867 char *f_name(struct file_struct *f, char *fbuf)
1868 {
1869         if (!f || !f->basename)
1870                 return NULL;
1871
1872         if (!fbuf) {
1873                 static char names[5][MAXPATHLEN];
1874                 static unsigned int n;
1875
1876                 n = (n + 1) % (sizeof names / sizeof names[0]);
1877
1878                 fbuf = names[n];
1879         }
1880
1881         if (f->dirname) {
1882                 int len = strlen(f->dirname);
1883                 memcpy(fbuf, f->dirname, len);
1884                 fbuf[len] = '/';
1885                 strlcpy(fbuf + len + 1, f->basename, MAXPATHLEN - (len + 1));
1886         } else
1887                 strlcpy(fbuf, f->basename, MAXPATHLEN);
1888
1889         return fbuf;
1890 }
1891
1892 /* Do a non-recursive scan of the named directory, possibly ignoring all
1893  * exclude rules except for the daemon's.  If "dlen" is >=0, it is the length
1894  * of the dirname string, and also indicates that "dirname" is a MAXPATHLEN
1895  * buffer (the functions we call will append names onto the end, but the old
1896  * dir value will be restored on exit). */
1897 struct file_list *get_dirlist(char *dirname, int dlen,
1898                               int ignore_filter_rules)
1899 {
1900         struct file_list *dirlist;
1901         char dirbuf[MAXPATHLEN];
1902         int save_recurse = recurse;
1903         int save_xfer_dirs = xfer_dirs;
1904
1905         if (dlen < 0) {
1906                 dlen = strlcpy(dirbuf, dirname, MAXPATHLEN);
1907                 if (dlen >= MAXPATHLEN)
1908                         return NULL;
1909                 dirname = dirbuf;
1910         }
1911
1912         dirlist = flist_new(WITHOUT_HLINK, "get_dirlist");
1913
1914         recurse = 0;
1915         xfer_dirs = 1;
1916         send_directory(ignore_filter_rules ? -2 : -1, dirlist, dirname, dlen);
1917         xfer_dirs = save_xfer_dirs;
1918         recurse = save_recurse;
1919         if (do_progress)
1920                 flist_count_offset += dirlist->count;
1921
1922         clean_flist(dirlist, 0, 0);
1923
1924         if (verbose > 3)
1925                 output_flist(dirlist);
1926
1927         return dirlist;
1928 }