Use sys_clone under Linux to share I/O contexts between dump processes.
[dump.git] / CHANGES
1 $Id: CHANGES,v 1.300 2009/06/18 09:50:54 stelian Exp $
2
3 Changes between versions 0.4b41 and 0.4b42 (released xxxxxxxxxxxxxxx)
4 =====================================================================
5
6 1. Fix printout of the 'Connection to' message in dump/restore.
7 Thanks to Dale Wiles <dwiles@users.sourceforge.net> for the
8 bug report and its fix.
9
10 2. Fix dump -w/-W output for filesystems having the last backup
11 level equal to 0. Thanks to Pascal Bouchareine
12 <kalou@users.sourceforge.net> for reporting the bug.
13
14 3. Enable restore to handle restoring onto a different SELinux
15 policy from the dump. Thanks to Tony Nelson
16 <tony_nelson@users.sourceforge.net> for the patch.
17 (see http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=189845
18 for details)
19
20 4. Fix EA set failures when restoring immutable files. Thanks
21 to Andrew Kroeger for the patch.
22
23 5. Fix "mode file too large" errors when restoring huge backups.
24 Thanks to Kenneth Porter <shiva@sewingwitch.com> for the bug
25 report.
26
27 6. Add the ability to link against libtinfo library if present,
28 instead of libtermcap or libncurses libraries. Thanks to
29 Adam Tkac <atkac@redhat.com> for the patch.
30
31 7. When comparing a backup, do not consider to be an error the fact
32 that we encounter files created while dump was in progress.
33 Thanks to Kenneth Porter <shiva@sewingwitch.com> for the bug
34 report.
35
36 8. Force '-a' to be the default value when running 'restore -C'.
37 Thanks to Kenneth Porter <shiva@sewingwitch.com> for the bug
38 report.
39
40 9. Many improvements to the 'cron_dump_to_disk' example.
41 Thanks to Aaron S. Hawley <ashawley@users.sourceforge.net> for
42 the patch.
43
44 10. Fix the QFA generation when extended attributes are backuped
45 along with the inodes. Prior to this fix, the entries for
46 some inodes may get corrupted (duplicated entries or incorrect
47 ones), making the QFA file unusable for repositionning in
48 restore. Many thanks to Kenneth Porter <shiva@sewingwitch.com>
49 for the bug report and the testing of the fix.
50
51 11. Fix EA restoration in debug mode (Sourceforge bug #1986216).
52
53 12. Stefan Auracher <star7@onlinehome.de> noticed that the Tower
54 of Hanoi backup strategy documented in the man page wasn't
55 actually based on the Tower of Hanoi algorithm. The man page
56 was updated accordingly.
57
58 13. Fix missing level information from dump output when using
59 the default level (RedHat bug #493635). Thanks to Adam Tkac
60 <atkac@redhat.com> for the patch.
61
62 14. Fix the dump man page to reflect the fact that the default
63 dump level is 0 not 9 (RedHat bug #356121). Thanks to Adam Tkac
64 <atkac@redhat.com> for the patch.
65
66 15. Fixed a few spelling errors in the man pages (RedHat bug #489853).
67 Thanks to Adam Tkac <atkac@redhat.com> for the patch.
68
69 16. Fixed restore man page and program usage which incorrectly stated
70 that -P and -A were both allowed in the same invocation (RedHat
71 bug #490627). Thanks to Adam Tkac <atkac@redhat.com> for the patch.
72
73 17. Use sys_clone under Linux to share I/O contexts between dump
74 processes, thus drastically increasing the performance of dump
75 under CFQ (which is the default I/O scheduler used in a number of
76 distributions). Thanks to Jeff Moyer <jmoyer@redhat.com> for the
77 patch.
78
79 Changes between versions 0.4b40 and 0.4b41 (released January 2, 2006)
80 =====================================================================
81
82 1. Fix restore of dumped Access Control Lists. The previous
83 code for EA works fine, but ACLs needed conversion from the
84 ext2/3 disk format to posix_acl format before restoring.
85
86 2. Fix some issues with restoration of EA on big endian
87 platforms.
88
89 3. Fix restore when the symtab is over 2GB in size.
90
91 4. Made the directory hash indexing an optional feature,
92 accessible by the '-H' option of restore, and disable it
93 by default.
94
95 5. Fixed dump to not include extended attributes information
96 in the toc (archive) file which confused restore -t.
97
98 Changes between versions 0.4b39 and 0.4b40 (released May 2, 2005)
99 =================================================================
100
101 1. Changed restore to emit warnings (instead of emitting a fatal
102 error) if a file (or a directory) is unavailable for a
103 comparision (if the user doesn't have the necessary permissions
104 to access it for example). Thanks to Kenneth Porter
105 <shiva@sewingwitch.com> for the bug report.
106
107 2. Re-done the 'do not save directory entries to non-dumped inodes
108 (excluded from dump)' feature. The previous implementation
109 worked well for excluded directories but not for regular files.
110 Thanks to Kenneth Porter <shiva@sewingwitch.com> for the bug
111 report.
112
113 3. Fixed a bug in dump where the tape size was miscalculated when
114 the user used -d/-s to specify the tape characteristics. Thanks
115 to Philip Goisman <goisman@physics.arizona.edu> for reporting
116 the bug.
117
118 4. Fixed another bug introduced in restore with the hashtree
119 implementation. This one caused restore to stop saying
120 "removenode: non-empty directory" in some cases.
121
122 5. Added support for dumping and restoring ext2/3 extended
123 attributes (EA), like the access control lists (ACL) or
124 the security labels used by SELinux.
125
126 Changes between versions 0.4b38 and 0.4b39 (released January 21, 2005)
127 ======================================================================
128
129 1. The newly added dump_on_cd_3 example was buggy, replace it
130 with an updated version from Andrew Basterfield
131 <bob@cemetery.homeunix.org>.
132
133 2. Made restore to chdir() back into the initial directory when
134 dumping core while aborting a comparision operation. The
135 previous behaviour was to write the corefile at the root of
136 the directory being compared, which could very well be
137 read only and preventing the corefile generation. Thanks
138 to Kenneth Porter <shiva@sewingwitch.com> for the bug report.
139
140 3. Silenced the failure to call fgetflags() when comparing an
141 entry which has no ext2 attributes (as in lsattr()).
142
143 4. Fix a brown paper bug in restore -C which broke restore and
144 caused modifications on the filesystem being compared
145 (directories containing a file with the same name as the
146 directory get renamed to RSTTMP...). Thanks to Kenneth Porter
147 <shiva@sewingwitch.com> for finding the bug and helping me
148 reproduce it.
149
150 5. Made restore -C force the -N flag (no writing allowed on
151 the disk) in order to prevent more bugs like the above one.
152
153 Changes between versions 0.4b37 and 0.4b38 (released January 7, 2005)
154 =====================================================================
155
156 1. Fix a couple of troff syntax bugs in the man pages.
157 Thanks to Eric Raymond <esr@thyrsus.com> for the patch.
158
159 2. Made restore use either libncurses or libtermcap, depending
160 on which one is available at configure time.
161
162 3. Fixed restore negative size display bug when comparing a
163 dump containing files over 2GB. Thanks to Steve Bonds
164 <sbonds@users.sourceforge.net> for the bug report.
165
166 4. Do not save directory entries to non-dumped inodes
167 (excluded from dump). This will eliminate the 'missing
168 file' warnings when doing 'restore -C'.
169
170 5. Fix dump crash when backuping a huge (2TB) filesystem,
171 due to a bogus calculation on the inode map size.
172 Thanks to Kevin B. Haines <K.B.Haines@rl.ac.uk> for
173 submitting the bug and testing the fix.
174
175 6. Fix a problem in restore where the final \0 in the symbolic
176 link names could have been lost, generating corrupt filenames.
177 Thanks to Kyle Wilson <kyle.wilson@amd.com> for reporting the
178 bug.
179
180 7. Implemented a hash list for the directory names in restore.
181 The linear list used before caused problems in interactive
182 restores when dealing with directories having thousands of
183 entries. Thanks to Brian Ristuccia <bristuccia@starentnetworks.com>
184 for reporting the bug.
185
186 8. Improved restore -C, this time including the directory
187 attributes into the comparision.
188
189 9. Made restore understand tapes containing EA/ACLs (which will
190 be dumped by the next version of dump). In this version
191 extended attributes on the tape are ignored, for full EA/ACL
192 support wait for the next version or try the experimental EA
193 patch.
194
195 Changes between versions 0.4b36 and 0.4b37 (released July 7, 2004)
196 ==================================================================
197
198 1. Added the --enable-staticz configure option which enables
199 dump and restore to be linked with static versions of
200 libz and libbz2 (and dynamic versions of all other libraries).
201 This will make Debian users happy, because libz and libbz2
202 were the only needed libraries living in /usr, all the
203 others live in /lib. In case of system emergency, it is
204 better not to have to rely on an extra filesystem.
205 Thanks to Bdale Garbee <bdale@gag.com> for the suggestion.
206
207 2. Fix compilation on (at least the Linux Debian port to) AMD64.
208 (<ext2fs/ext2_types.h> defines some types (__s64 and __u64)
209 that are also defined by <linux/types.h> (<asm/types.h>) and
210 they conflict).
211
212 3. Make dump's reading of the dumpdates file a bit more robust,
213 preventing dump from crashing when the dumpdates file has
214 been modified by hand.
215
216 4. Fixed some offset calculations in dump code which could
217 lead to "bread lseek errors" on large filesystems. Thanks
218 to Bruce Lowekamp <lowekamp@users.sourceforge.net> for
219 reporting this bug and debugging the issue.
220
221 5. Made dump use the blkid library when searching for devices
222 by label or uuid instead of dump's own routines.
223
224 6. Corrected a bug in dump where a wrong LABEL=... line in
225 /etc/fstab could prevent dump from dumping unrelated
226 filesystems. Thanks to Bruce Lowekamp
227 <lowekamp@users.sourceforge.net> for reporting the bug.
228
229 Changes between versions 0.4b35 and 0.4b36 (released April 21, 2004)
230 ====================================================================
231
232 1. Fixed dump compilation with old gcc versions. Thanks to
233 Mike Castle <dalgoda@users.sourceforge.net> for the patch.
234
235 2. Fixed some warnings (howmany, roundup, powerof2 redefined)
236 when compiling against a recent glibc version.
237
238 3. Fixed a bug in restore preventing the read of a dump tape
239 written with Solaris 7 ufsdump. Thanks to Patrick Higgins
240 <phiggins@transzap.com> for reporting the bug and providing
241 the test case.
242
243 4. Changed dump to enable the creation of volumes of different
244 sizes in a single run (make -B accept a list of values).
245 Patch contributed by Florian Zumbiehl <florz@gmx.de>.
246
247 5. Use the glibc provided minor() and major() macros instead
248 of our own bitmask implementation. This should be safe for
249 when the major/minor namespace will migrate to 32 bits.
250 Thanks to Zhang Jun <zhangjun@nanjing-fnst.com> for reporting
251 the bug.
252
253 6. Made explicit in the dump man page that dump will not create
254 a remote file, it will only write to an already existing one.
255
256 7. Another try at making size estimates better again.
257
258 8. Put back the inconditional running of the end-of-tape
259 script which was changed in 0.434 to be run only when -M
260 or multiple -f were NOT specified. Some users rely on this
261 feature even when it is combined with -M/-f.
262
263 9. Fixed restore when restoring huge backups (where rstdir...
264 temporary files are over 2GB). Thanks to
265 Raphael Jaffey <rjaffey@artic.edu> for reporting this,
266 debugging the issue and making the patch.
267
268 10. Made restore understand FreeBSD UFS2 tapes. Thanks to
269 David <vrtin@users.sourceforge.net> for submitting the bug
270 and providing a test case.
271
272 11. Made dump work with an arbitrary integer level (no more 10
273 levels only limitation). Thanks to Matthew
274 <msvincen@midway.uchicago.edu> for the patch.
275
276 Changes between versions 0.4b34 and 0.4b35 (released December 21, 2003)
277 =======================================================================
278
279 1. Added a note in the dump man page saying that the default
280 blocksize can be 32 if -d is used with a high density argument.
281 Thanks to Antonios Christofides <A.Christofides@itia.ntua.gr>
282 for the patch.
283
284 2. Fixed configure to correctly understand CPPFLAGS, CFLAGS,
285 LDFLAGS environment variables. Thanks to Arcady Genkin
286 <antipode@users.sourceforge.net> for reporting the bug.
287
288 3. Made -e/-E options of dump accept an unlimited number of inodes
289 to be excluded, rather than a hardcoded maximum. Thanks to
290 Dietrich Rothe <d-rothe@users.sourceforge.net> for the patch.
291
292 4. Updated the autoconf system to 2.50. Forced the -D_BSD_SOURCE
293 and -D_USE_BSD_SIGNAL defines in configure in order to solve
294 64bit build problems because quad_t is redifined with a
295 different signature. Thanks to Mike Harris <mharris@redhat.com>
296 for reporting this bug.
297
298 5. Made restore build on Solaris, making possible to
299 restore Linux's "enhanced" tapes. Thanks to Uwe Gohlke
300 <uwe@ugsoft.de> for the patch.
301
302 6. Made an extension in the dump tape format capable of saving
303 MacOSX specific inode extensions. Uwe Gohlke <uwe@ugsoft.de>
304 wrote the extension and contributed the restore code back
305 into this codebase. The same extension mechanism will be
306 used in the future to save ACLs...
307
308 7. Made rmt work correctly with regard to QFA and local/remote
309 files and tapes. The remote access will however work only
310 when the dump provided rmt version is used. If you want to
311 use another rmt server, please do not use the QFA feature.
312 Thanks to Uwe Gohlke <uwe@ugsoft.de> for the patch.
313
314 Changes between versions 0.4b33 and 0.4b34 (released April 18, 2003)
315 ====================================================================
316
317 1. Fixed the INSTALL file to reflect the actual install paths.
318 Thanks to David Ronis <ronis@ronispc.chem.mcgill.ca> for
319 reporting the bug.
320
321 2. Fixed the configure script to only check for headers presence
322 instead of trying to compile them. This should fix issues
323 with old build environments. Thanks to Kari Lempiainen
324 <kari@funky.pp.fi> for reporting the bug.
325
326 3. Fixed restore to correctly ignore sockets when comparing
327 a dump (as socket cannot be properly restored anyway). Thanks
328 to Gunther Reiszig <gunther@mit.edu> for reporting the bug.
329
330 4. Fixed restore to correctly access the archive file (-A argument)
331 even when using a remote tape. Thanks to Fabrice Bellet
332 <fabrice@bellet.info> for reporting the bug.
333
334 5. Fixed (again) handling of long (largefile) seeks in rmt.
335 Thanks to Fabrice Bellet <fabrice@bellet.info> for reporting
336 the bug.
337
338 6. Fixed restore corner case when dealing with large block sizes
339 dump is able to write now (-b 1024). Thanks to Fabrice Bellet
340 <fabrice@bellet.info> for reporting the bug.
341
342 7. Fixed a bug preventing dump to access a filesystem having
343 a label exactly 16 bytes in length. Thanks to <minduni@ti-edu.ch>
344 for reporting the bug.
345
346 8. Made dump store dump dates using explicit timezones, fixing a
347 problem with incremental dumps if the timezone is modified
348 between the dumps. Thanks to Philippe Troin <phil@fifi.org> for
349 the bug report and the patch.
350
351 9. Fixed a bug encountered when dumping individual files (not full
352 filesystems or directories) and dangling symbolic links happen
353 to be in the list of files. For as far as dump is concerned,
354 dangling symbolic links are allowed, and are dumped as is.
355 Thanks to Jin-su Ahn <jsahn@ee.snu.ac.kr> for reporting the
356 bug and providing the fix.
357
358 10. Fixed open and creation modes and permissions for QFA and
359 table-of-contents files in dump and restore. Thanks to
360 Philippe Troin <phil@fifi.org> for the patch.
361
362 11. Fixed the archive file descriptor handling enabling it to be 0.
363 This can happen in some cases when shell redirections are used.
364 Thanks to Philippe Troin <phil@fifi.org> for the patch.
365
366 12. Delayed the opening of archive file until after suid had been
367 dropped (fixing a possible security issue if dump is suid).
368 Thanks to Philippe Troin <phil@fifi.org> for the patch.
369
370 13. Fixed the 'S' command handling in the rmt client part.
371 Thanks to Philippe Troin <phil@fifi.org> for the patch.
372
373 14. Modified the end-of-tape script handling to print out statistics
374 (and stop the timer) before launching the eot script. Also, the eot
375 script does not get run anymore when using -M (which makes sense) or
376 when multiple tapes are listed on the command line
377 (-f tape0,tape1,tapen) (which also makes sense).
378 Thanks to Philippe Troin <phil@fifi.org> for the patch.
379
380 15. Relicensed dump/restore under the 'revised' BSD license, as per
381 ftp://ftp.cs.berkeley.edu/ucb/4bsd/README.Impt.License.Change.
382
383 16. Added LZO compression to dump. This new compression method has
384 the advantage of being super fast, thus not killing tape streaming
385 on slow machines. Thanks to Helmut Jarausch
386 <jarausch@igpm.rwth-aachen.de> for the patch and to
387 Markus Oberhumer <markus@oberhumer.com> for giving special permission
388 to include his miniLZO project (GPL licensed) in dump/restore.
389
390 17. Some small buffer overruns fixes in rmt. Thanks to Antonomasia
391 <ant@notatla.demon.co.uk> for reporting the bugs.
392
393 18. Added a special rmt version which can do encryption when writing
394 to tape. Read examples/encrypted_rmt/README for details on
395 how to enable and configure it. Thanks to Ken Lalonde
396 <ken@globalremit.com> for the patch.
397
398 19. Made dump work with 2.5 kernel end of tape early warning semantics.
399 Thanks to Kai Makisara <Kai.Makisara@kolumbus.fi> for the patch.
400
401 20. Fixed a bug which caused dump -w|-W not to work anymore, because
402 the fs_freq and fs_passno fields in /etc/mtab are always set
403 to 0 0. Thanks to Trent Piepho <xyzzy@speakeasy.org> for
404 reporting the bug.
405
406 Changes between versions 0.4b32 and 0.4b33 (released February 10, 2003)
407 =======================================================================
408
409 1. Added a note in the restore man page clarifying the question
410 'set the permissions on the current directory ?' asked by
411 restore at the end of treatment in -i and -x modes.
412
413 2. Fixed the endianess issues when reading compressed tapes.
414 Thanks to Dark Force <daq4th@users.sourceforge.net> for
415 reporting this bug and providing test cases.
416
417 3. Fixed the "ACL won't be dumped" warning message (which showed
418 an extra, unrelated error message). Thanks to Dragan Krnic
419 <dkrnic@lycos.com> for reporting this bug.
420
421 4. Made dump look first into /etc/mtab, then into /etc/fstab
422 when searching for filesystem paths. Also fixed some problems
423 caused by binding mounts or multiple block device mounts.
424 Thanks to Matus Uhlar <uhlar@fantomas.sk>, Elliott Mitchell
425 <ehem@m5p.com>, Greg Edwards <gedwards@users.sourceforge.net>,
426 Brian Hoy <brian.hoy@opus.co.nz>. (fixes Debian bugs #147086
427 and #160305, Sourceforge bugs #618699 and #679832).
428
429 5. Made dump's -I option accept the value '0' meaning all the
430 read errors will be ignored. This can be useful when running
431 dump from unattended sessions (like cron jobs). Thanks to
432 John I Wang <jiwang@users.sourceforge.net> for the suggestion.
433
434 6. Fixed the output of dump to indicate 'blocks' instead of
435 'tape blocks' in the various messages (blocks are always
436 1 Kilobyte, tape blocks are 1 BK * '-b' argument), and
437 made it clearly print the current blocksize at the start of
438 a dump. Thanks to Michal Szymanski <msz@astrouw.edu.pl> for
439 the suggestions.
440
441 7. Made rmt understand long (largefiles) seeks.
442
443 8. Fixed build with very old versions of libext2fs, where
444 EXT2_FT_* constants were undefined.
445
446 9. Made dump accept the dumpdates path on the command line
447 (-D file option) instead of using only the hardcoded one.
448 Thanks to Piete Brooks <pb22@users.sourceforge.net> for the
449 suggestion.
450
451 10. Enabled rmt, LFS, readline, QFA options by default in
452 ./configure. Updated the configure process (new versions
453 of config.guess, config.sub etc).
454
455 Changes between versions 0.4b31 and 0.4b32 (released November 15, 2002)
456 =======================================================================
457
458 1. Changed dump to use fcntl(F_SETLK) style locking instead
459 of flock() when locking the dumpdates file. With the old
460 locking scheme, a local user having read rights on the
461 dumpdates file could be able to do a Denial of Service attack
462 on dump. In order to lock the dumpdates file with the new
463 scheme, the user would need to have write access on the file.
464 Thanks to Richard Johnson <Richard.Johnson3@ey.com> for
465 reporting the bug (originally a bugtraq post).
466
467 2. Fixed interactive 'ls' which caused spurious errors warnings
468 about 'undefined filetypes' detected. Thanks to Jorgen Ostling
469 <jorgen_ostling@users.sourceforge.net> for reporting this
470 bug.
471
472 3. Fixed dump's estimate when dealing with sparse inodes.
473
474 4. Modified dump to allow setting a blocksize bigger than 32kB
475 (raised the limit to 1024kB), because newer hardware needs
476 this for proper operation. Thanks to Dirk Traenapp
477 <dtraenapp@users.sourceforge.net> for reporting this.
478
479 5. Fixed a bug causing Dump to stop and report an error if an
480 inode number in the exclude file was followed by some amount
481 of whitespace. Thanks to Jeffrey Sofferin
482 <sofferin@users.sourceforge.net> for reporting this bug.
483
484 6. Fixed a bug which caused restore, in some particular cases,
485 to ask some 'scary' questions and leave a bunch of RSTTMP
486 directories behind when restoring incremental tapes. Thanks
487 to Philippe Troin <phil@fifi.org> for reporting this bug and
488 providing the test cases.
489
490 7. Changed the wording when inodes are excluded from dump:
491 replaced 'Added inode 7 to exclude list' with
492 'Excluding inode 7 (resize inode) from dump', as suggested
493 by Elliott Mitchell <ehem@m5p.com> in a Debian bug report.
494
495 Changes between versions 0.4b30 and 0.4b31 (released July 30, 2002)
496 ===================================================================
497
498 1. Fixed rmt open flags transmission (GNU's symbolic syntax over
499 rmt) which I broke in 0.4b29. Thanks to Eros Albertazzi
500 <eros@lamel.bo.cnr.it> for reporting the bug.
501
502 Changes between versions 0.4b29 and 0.4b30 (released July 25, 2002)
503 ===================================================================
504
505 1. Made dump print out the ext2 logical block number in case of a read
506 error, which can be given as an argument to debugfs' ncheck command,
507 in order to find out the affected inode. Added note about this
508 usage in dump's man page.
509
510 2. Fixed a problem in restore when reading tapes written on big
511 endian machines with very old versions of dump. The patch was
512 contributed by George Helffrich <george@geology.bristol.ac.uk>.
513
514 3. Fixed the tape length calculation when using large tapes
515 and compression. Thanks to Georg Lippold
516 <g_lippold@sourceforge.net> for reporting the bug.
517
518 4. Added a new set of examples in dump_on_cd_2 directory, based
519 on dump_on_cd examples but somewhat enhanced, supporting DVD
520 media, and localized in english and german. Thanks to
521 Georg Lippold <g_lippold@sourceforge.net> for the new scripts.
522
523 5. Made dump save 32 bit UID/GID. Internally, this was achieved
524 by switching from the old BSD inode format to the new BSD
525 inode format, which means that the tape format was changed.
526 However, since all restore versions out there should
527 transparently support both inode formats, the change should
528 have no side effects. Thanks to John Yu <jky@cs.bu.edu> for
529 reporting the bug.
530
531 6. Fixed a lot of warnings in the code shown when compiling
532 with 'gcc -W'. Thanks to Matthias Andree
533 <matthias.andree@stud.uni-dortmund.de> for reporting this.
534
535 7. Fixed a small markup bug in the dump man page. Thanks to
536 Eric S. Raymond <esr@minx.thyrsus.com> for submitting the
537 patch.
538
539 8. Rewrote entirely the man pages using the tmac.an macro
540 package (Linux man page format) instead of the original BSD
541 format. They should be now cleaner and easier to modify.
542
543 Changes between versions 0.4b28 and 0.4b29 (released June 8, 2002)
544 ==================================================================
545
546 1. Fixed a problem in the rmt ioctl command, where ioctl's issued from
547 non Linux clients were misinterpreted. The description of the problem
548 (incompatible numbering in Linux mtio opcodes) is documented at
549 ftp://ftp.fokus.gmd.de/pub/unix/star/README.mtio . Thanks to
550 Jörg Schilling <schilling@fokus.gmd.de> for reporting this bug and
551 providing an excellent, cross-platform replacement for rmt in his
552 star package.
553
554 2. Fixed a bug reported by Andy Mentges <amentges@jumpline.com>
555 which caused restore to fail when the inode maps were not entirely
556 contained on the first volume (when using really small volumes or
557 when dumping a huge number of inodes).
558
559 3. Fixed a problem in dump, where files in subdirectories of directories
560 that have the nodump flag set (or being excluded with -e/-E) were
561 sometimes incorrectly dumped (instead of being excluded). The original
562 fix is from David C Lawrence <tale@dd.org> in the FreeBSD version
563 of dump: http://www.FreeBSD.org/cgi/query-pr.cgi?pr=32414
564 Thanks to Ted Grzesik <tedgyz@roostme.com> for reporting the bug and
565 help testing the patch.
566
567 4. Added some example scripts from Gerd Bavendiek <bav@epost.de>
568 which makes one able to pipe the output of dump, by the net, to
569 a remote CD-burner server.
570
571 5. Made dump use O_CREAT|O_TRUNC both locally and remotely (over rmt),
572 and use GNU's symbolic syntax over rmt instead of numerical values
573 to assure multiple platform compatibility.
574
575 6. Documented the -d option in restore.
576
577 7. Added a -v (verbose) mode to dump. For now it just prints the number
578 of the inode being dumped, but this could evolve in future versions
579 to include interesting debugging output.
580
581 8. Added a -o flag to restore, which automatically restores the current
582 directory permissions in -i or -x mode, without asking the operator
583 whether to do so. Patch submitted by Tony Clayton <tonyc@e-smith.com>
584 and Peter Samuel <peters@e-smith.com>.
585
586 Changes between versions 0.4b27 and 0.4b28 (released April 12, 2002)
587 ====================================================================
588
589 1. Fixed a bug in the QFA feature which could have caused
590 a bad tape positionning on restore (causing extra delays in
591 restore). Credits go to Uwe Gohlke <uwe@ugsoft.de>.
592
593 2. Added a small note in the dump man page specifying that
594 there should be no white space between the option letter and
595 the -j or -z parameter, thanks to Kenneth Porter <shiva@well.com>
596
597 3. Made restore work with older versions of the readline library,
598 by checking in configure for several extended symbols. Restore
599 can now be compiled with a readline as old as the 2.0 release
600 (though it may be a good idea to upgrade it to a more recent
601 version...). Thanks to Andrew Donkin <ard@waikato.ac.nz> for
602 reporting the build failures.
603
604 4. Fixed a performance problem with the QFA file creation in
605 dump, which made unnecessary seeks on the tape slowing down
606 the dump. Thanks to Andrew Donkin <ard@waikato.ac.nz> for
607 reporting this issue.
608
609 5. Removed the inclusion of some kernel headers in the dump
610 source, which prevented the compile in some kernel/glibc
611 headers/architecture combination. Thanks to Bdale Garbee
612 <bdale@gag.com> for reporting the bug.
613
614 6. Added the appropriate error message when dump fails to
615 open the output file for writing. Thanks to Amith Varghese
616 <amithv@yahoo.com> for reporting this bug.
617
618 7. Made restore able to understand large Solaris ufsdump tapes
619 (containing inodes bigger than 4194304). Sun have introduced
620 an "extension" to the dump tape format when dealing with
621 those inodes, which was uncorrectly handled by Linux restore.
622 Thanks to Uwe Gohlke <uwe@ugsoft.de> for reporting the bug and
623 providing a test case.
624
625 8. Added the -m parameter to dump which optimises the output for
626 inodes having been changed but not modified since the last dump
627 ('changed' and 'modified' have the meaning defined in stat(2)).
628 For those inodes, dump will save only the metadata, instead of
629 saving the entire inode contents. Inodes which are either
630 directories or have been modified since the last dump are saved
631 in a regular way. Uses of this flag must be consistent, meaning
632 that either every dump in an incremental dump set have the flag,
633 or no one has it.
634
635 Changes between versions 0.4b26 and 0.4b27 (released February 15, 2002)
636 =======================================================================
637
638 1. Fixed behaviour of dump when exceeding resource limits
639 (SIGXFSZ treatment).
640
641 2. Added the -L flag to restore to allow the user to specify a
642 maximal allowed number of miscompares when using restore
643 with the -C option to check the backup.
644
645 3. Detailed the manual entry for the -N option of restore.
646
647 4. Added the -a flag to restore to make able doing unattended
648 restores in -i or -x mode (automatically walks through the
649 multiple dump volumes).
650
651 5. Extended the QFA mode to work with local files and/or
652 remote tapes and files. This way, restore can know in advance
653 the tape number and the offset for the inodes to extract and
654 can minimize the extraction time by seeking directly to the
655 good tape/offset.
656
657 6. Added the -A <archive> option to both dump and restore,
658 which makes dump to archive a dump table-of-contents in
659 the specified file to be used by restore to determine
660 whether a file is in the dump file that is being restored.
661 (the archive file syntax is also compatible with the
662 Solaris ufsdump generated one).
663
664 7. Small fix in restore making it able to read some (broken ?)
665 Solaris ufsdump tapes.
666
667 8. Fixed dump to correctly recognise the root filesystem when using
668 ext2 disk labels (LABEL=/). Thanks to John Yu <jky@it.bu.edu>
669 for reporting this bug.
670
671 9. Added the -P <file> option to restore to create a
672 Quick File Access file from an already made dump. Patch
673 contributed by Uwe Gohlke <uwe@ugsoft.de>.
674
675 10. Made restore compile and run on Solaris, making it a
676 possible replacement for the standard ufsrestore. Port was
677 contributed by Uwe Gohlke <uwe@ugsoft.de>.
678
679 Changes between versions 0.4b25 and 0.4b26 (released January 7, 2002)
680 =====================================================================
681
682 1. Added a set of backup scripts from Eugenio Diaz
683 <getnito@yahoo.com> in the examples section. It features
684 automatic (cron based) full and incremental dumping of
685 several filesystems on a separate filesystem.
686
687 2. Fixed a off-by-one miscalculation which disabled dumping
688 a one letter subdirectory. Thanks to Chris Siebenmann
689 <cks@utcc.utoronto.ca> for reporting the bug.
690
691 3. Fixed several restore bugs occuring when trying to
692 restore some missing files on the tape. Thanks to Chris
693 Siebenmann <cks@utcc.utoronto.ca> for reporting the bug.
694
695 4. Fixed --with-ldopts configure argument passing, installing from
696 a separate object directory, makefile cleanups contributed
697 by <splite@purdue.edu>.
698
699 5. Fix a bug which could caused, in some conditions, the highest
700 number inode of a filesystem, to not be dumped. Many thanks
701 to Chris Siebenmann <cks@utcc.utoronto.ca> for helping me
702 chase this bug.
703
704 Changes between versions 0.4b24 and 0.4b25 (released November 17, 2001)
705 =======================================================================
706
707 1. Added a mini howto from Patrick Walsh in the examples
708 directory.
709
710 2. Minor man pages syntax corrections. Thanks to
711 Chris Danis <screechco@home.com> for reporting the bugs.
712
713 3. Added a script from David B. Peterson <dave@toppledwagon.com>
714 to the examples section. It features dumping several
715 filesystems to a remote tape drive upon ssh.
716
717 4. Added a patch provided by Richard Jones <rich@annexia.org>
718 which allows BRADEMAX (number of read errors tolerated by
719 dump) to be adjusted using the -I option.
720
721 5. Fixed a bug which disabled doing "restore -C -f -". Thanks
722 to Clemens Stiglechner <a9401816@unet.univie.ac.at> for the
723 patch.
724
725 6. Add the -l option to restore to specify if, when doing a
726 remote restore, the file used to access the backup is a
727 regular file (the defaults being a tape drive). Restore needs
728 to know this information when reading a remote compressed
729 dump. Previously, this information was autodetected, but
730 the autodetection code fails (with ioctl: Inappropriate ioctl
731 for device) when using a non Linux remote box. Thanks to
732 many users and especially to Eros Albertazzi
733 <eros@lamel.bo.cnr.it> for reporting this.
734
735 7. Found a workaroung for the dump deadlock problem (3 childs
736 stuck in pause(), father in read()). The workaround seems
737 to work for me and several beta-testers. If it doesn't work
738 for you, please report back.
739
740 8. Updated the RPM spec file (BuildPrereq, URL etc).
741
742 Changes between versions 0.4b23 and 0.4b24 (released September 12, 2001)
743 ========================================================================
744
745 1. Fixed the permissions of a newly created QFA file by dump.
746
747 2. Cleaned up the source of dump (the external variables
748 definition was a complete mess, making possible to have
749 objects overlap).
750
751 3. Fixed restore to use the full tape volume path when doing
752 a compare (since it changes the working directory to the
753 filesystem being compared in the process).
754
755 4. Added the -q option to dump which makes dump abort
756 whenever it needs operator attention. It should be
757 easier to use dump in scripts with this option.
758
759 5. Detect the use of incompatible options to dump and
760 refuse them (like -a and -B options together).
761
762 6. Added bzip2 compression to dump/restore (use option -j level
763 to select it). Note that you will need a restore version
764 >= 0.4b24 in order to restore a bzip2 compressed dump.
765 The same warning as for the zlib compression applies:
766 the tape format of a bzip2 dump is not compatible with the
767 original BSD tape format.
768
769 7. Fixed a overflow problem in dump corrupting the dump when
770 very large files were encountered. Thanks to Vladimir Nadvornik
771 for the bug report.
772
773 8. Added a ioctl(BLKFLSBUF) in dump which should flush the
774 kernel buffer/page cache before starting the dump, helping
775 a bit those who use dump on mounted filesystems. Thanks to
776 John Yu <jky@it.bu.edu> and to Theodore T'so <tytso@mit.edu>
777 for this suggestion.
778
779 9. Updated the RPM spec file following the RedHat changes
780 (dynamically linked binaries now in /usr/sbin etc).
781
782 10. Added a patch from Helmut Jarausch <jarausch@igpm.rwth-aachen.de>
783 which enables restore to recognise multi volume compressed dumps
784 done on CD. Included his scripts for dump (which pipe the dump
785 output directly into cdrecord) and restore. There is now
786 possible to to backups to CD on the fly!
787
788 Changes between versions 0.4b22 and 0.4b23 (released July 20, 2001)
789 ===================================================================
790
791 1. Fixed a buffer overflow in restore/tape.c. Patch provided by
792 Marcus Meissner (Caldera International Security Dept.).
793
794 2. Implement the Sun rmt extended protocol. Patch submitted
795 by Ian Gordon <iangordon@users.sourceforge.net>.
796
797 3. Check for the e2fsprogs header <ext2fs/ext2_fs.h> instead of
798 the linux kernel header. This ensures that dump always has the
799 latest version of this file. Patch submitted by
800 Andreas Dilger <adilger@turbolinux.com>.
801
802 4. Report any filesystem present in either /etc/fstab with a
803 positive passno or /etc/dumpdates in dump -w output.
804 Patch submitted by Andreas Dilger <adilger@turbolinux.com>.
805
806 5. Fixed the looping problem in dump introduced in the
807 previous version.
808
809 6. Changed the -B option of dump to limit the size of
810 _compressed_ output per volume if compression is on.
811 Patch contributed by Helmut Jarausch
812 <jarausch@igpm.rwth-aachen.de>. Note however that, since
813 it is impossible to predict the size of the compressed
814 data before doing the compression, the -B limit is a bit
815 conservative.
816
817 7. Fixed a bug in reading the operator typed file/tape path for
818 the next volume in restore.
819
820 8. Implemented a "-F script" option to restore which permits the
821 user to specify a script which will be launched at the
822 beginning of each tape, useful for automatic programming of
823 tape changers for example. See the restore man page for the
824 script parameters and return codes.
825
826 9. Small fix for the QFA routines provided by Uwe Gohlke
827 <uwe@ugsoft.de>, and some recommendations for QFA uses in
828 the man pages.
829
830 10. Fixed the multivolume restoring where making a mistake
831 to the 'Mount next tape' prompt caused several blocks to
832 be lost.
833
834 11. Enhanced the -e option of dump to take as a parameter a
835 comma separated list of inode numbers.
836
837 12. Added the -E option to dump which specify a file containing
838 inode numbers to exclude from the dump.
839
840 13. Fixed the compressed multi-volume dump + restore.
841
842 Changes between versions 0.4b21 and 0.4b22 (released May 12, 2001)
843 ==================================================================
844
845 1. Made dump report the number of blocks dumped per volume.
846 Thanks to Kenneth Porter <shiva@well.com> for the suggestion.
847
848 2. Fix a bug in dump where waiting too much at the 'change volumes'
849 question caused the volume to be rewritten. Thanks to
850 Graham Reed <greed@users.sourceforge.net> for reporting the
851 bug and providing a patch.
852
853 3. Added a compression option to dump, contributed by Jerry
854 Peters <gapeters@worldnet.att.net>.
855
856 WARNING: the compression works only on files/pipes and
857 drives supporting variable block size.
858
859 WARNING2: tapes created using the compression feature are
860 incompatible with the standard dump tape format, and a
861 version of restore >= 0.4b22 will be needed for extraction.
862
863 4. Fixed some compilation problems for glibc 2.2.2 and 64 bit
864 architectures. Thanks to Paul Slootman <paul@debian.org> for
865 the patch and to Bdale Garbee <bdale@gag.com> for forwarding
866 it upstream.
867
868 5. Many cleanups (CPP defines, const char warnings, check of
869 ext2fs COMPAT flags, time_t cleanups, added libext2 version
870 in dump usage text) by Andreas Dilger <adilger@turbolinux.com>.
871
872 6. Made --prefix option work in configure. All the install path
873 are now based on the configure parameters.
874
875 7. Added the Quick File Access mode in dump/restore, contributed
876 by Uwe Gohlke <uwe@ugsoft.de>. In this mode, dump stores in
877 a file tape position for each inode, and this file is used by
878 restore (if called with parameter Q and the filename)
879 to directly position the tape at the file restore is currently
880 working on. This saves hours when restoring single files from
881 large backups, saves the tapes and the drive's head. Use
882 --enable-qfa option of configure to compile in the QFA support.
883
884 8. Added the possibility to dump several files and directories
885 in a single invocation of dump. Thanks to Uwe Gohlke
886 <uwe@ugsoft.de> for implementing this option.
887
888 9. Fixed the dumping and restoring of files with holes
889 (files ending with holes didn't get dumped or restored
890 correctly).
891
892 10. Fixed a socket descriptor leak in dump, which leaved opened
893 3 file descriptors per dump process (and there is one dump
894 process per tape).
895
896 11. Fixed dump large file system support, by explicit use of
897 open64/lseek64/etc functions (explicit use needed because
898 e2fsprogs libraries don't behave well when compiled with
899 FILE_OFFSET_BITS=64).
900
901 Changes between versions 0.4b20 and 0.4b21 (released January 13, 2001)
902 ======================================================================
903
904 1. Fixed some bugs in the dump -w|-W logic introduced by
905 the previous version. Thanks to Andreas Dilger
906 <adilger@turbolinux.com> for his help on this one.
907
908 2. Fixed again a compilation problem when using old e2fs
909 headers (filesystem label related). Thanks to many users
910 who reported this stupid error.
911
912 3. Fixed a build problem on old lib5 based systems dealing with
913 _PATH_MNTTAB being undefined. Thanks to John Adams
914 <johna@onevista.com> for reporting the bug.
915
916 4. Improved the error detection in restore when creating
917 the temporary files in TMPDIR. Now restore will corectly
918 report a 'no space left on device' error instead of
919 strange errors which could imply an archive corruption.
920 Thanks to Gabriel Somlo <somlo@cs.colostate.edu> and
921 bgjenero <bgjenero@sympatico.ca> for reporting the bug.
922
923 5. Added the throughput information in dump's progression
924 messages. Thanks to Andreas Dilger <adilger@turbolinux.com>
925 for the patch.
926
927 6. Use libext2fs's inode scanning routines, which are particularly
928 robust when dealing with errors (especially when having some
929 bad blocks on the filesystem). Many thanks to Colin
930 <colin@horizon.com> for reporting the bug, investigating
931 the issues, finding a workaround, writing the patch and
932 fully testing it... (of course, if this patch will break
933 anything, he is to blame for :-)).
934
935 7. Made dump and restore LFS aware. Dump can dump a filesystem
936 containing large files, generate a large file on output and
937 restore can restore them. This feature is not enabled by
938 default, use --enable-largefile option of configure to enable
939 it (you need to have a LFS aware glibc though). Thanks to
940 Andrea Arcangeli <andrea@suse.de> for submitting the patch,
941 and to Theodore T'so <tytso@valinux.com> for his always
942 useful thoughts.
943
944 8. Made dump ask upon a tape write error if it should rewrite
945 the current volume (assume this is a bad tape) or if it should
946 assume an end-of-tape condition (useful for tape drives which
947 behaves badly at the end of the tape). Thanks to Andreas
948 Dilger <adilger@turbolinux.com> for the suggestion.
949
950 Changes between versions 0.4b19 and 0.4b20 (released November 10, 2000)
951 =======================================================================
952
953 1. Fixed a small compilation problem due to a change
954 in the definintion of the struct sigaction in
955 glibc 2.0 / libc5. Thanks to Gunther Schlegel
956 <schlegel@riege.de> for reporting the bug and to
957 Dave Platt <dplatt@snulbug.mtview.ca.us> for suggesting
958 a fix.
959
960 2. Modified the label and uuid analysis in order to be
961 self-contained (independant of kernel/libc headers). This
962 should fix the compile with older kernel/libc headers and
963 will preserve the functionality. Thanks to Bernhard Erdmann
964 <bernhard.erdmann@gmx.de> for reporting the bug.
965
966 3. The 'exclude inode' option, if applied to a directory
967 inode, excludes now correctly all the leaves of this
968 directory. Thanks to John R. Dennison
969 <gerdesas@users.sourceforge.net> for reporting the bug.
970
971 4. Fixed the '-e' option to disable the possibility
972 to exclude the root inode (which causes the dumps to
973 be unreadable by restore). Prevented array overflow
974 when multiple -e options are used.
975
976 5. Fixed dump to correctly interpret a filesystem argument
977 which is a mountpoint and it is not an absolute path
978 (as specified in the fstab). Thanks to Bernhard R. Erdmann
979 <be@berdmann.de> for reporting the bug.
980
981 6. Made dump able to backup files larger than 2 GB. Note that
982 dump still doesn't cope with files larger than 4 GB.
983
984 7. Restore the real uid and gid before invoking an external
985 RSH program (big hole when dump or restore is suid root!).
986
987 8. Get the values for the default tape device and the location
988 of fstab file from the system headers. Thanks to
989 Andreas Dilger <adilger@turbolinux.com> for the patch.
990
991 9. Made dump -w|-W report all recognized filesystems
992 present in either /etc/fstab or /etc/dumpdates, and present
993 the list in the same order as in fstab file. Thanks
994 to Andreas Dilger <adilger@turbolinux.com> for the patch.
995
996 10. Made dump's -a (automatic end of tape detection) the
997 default. Specifying one of -B, -c, -d or -s options will
998 override the EOM detection. Thanks to Andreas Dilger
999 <adilger@turbolinux.com> for the patch.
1000
1001 11. Save the ext2 filesystem label into the dump volume label.
1002 Specifying a volume label on the command line (-L option)
1003 will override this feature. Thanks to Andreas Dilger
1004 <adilger@turbolinux.com> for the patch.
1005
1006 Changes between versions 0.4b18 and 0.4b19 (released August 20, 2000)
1007 =====================================================================
1008
1009 1. Fixed the signal handling in dump (which I broke in 0.4b17)
1010 which was causing several strange dump failures (dump
1011 hanged or segmentation faults etc).
1012
1013 2. Specified the default blocksize in the dump man page.
1014
1015 3. Changed two info messages of restore to be written on stdout
1016 instead of stderr in order to leave stderr only for errors
1017 or warnings. Thanks to Stephen Atwell
1018 <satwell@urbana.css.mot.com> for the suggestion.
1019
1020 4. Corrected an off by one calculation which prevented
1021 dumping a directory having a 1 character name length.
1022 Thanks to Bernhard Erdmann <bernhard.erdmann@gmx.de>
1023 for reporting the bug.
1024
1025 5. Reinforce dump in order to allow only level 0 and no
1026 -u option when dumping a subdirectory, and document
1027 this in the man page. Thanks to Bernhard Erdmann
1028 <bernhard.erdmann@gmx.de> for reporting the bug.
1029
1030 6. Fixed a small harmless bug in restore which caused
1031 in some conditions hard links to be restored several
1032 times (therefore generation some warning messages).
1033 Thanks to Kenneth Porter <shiva@well.com> for
1034 reporting the bug.
1035
1036 7. Updated the RPM spec file to the latest RedHat version,
1037 providing FHS packaging and other cosmetic changes.
1038 You will need RPM version >= 3.0.5 in order to build the RPM.
1039
1040 8. Updated the configure script to check for libtermcap
1041 before checking for libreadline, because we need this
1042 library in order to compile the readline support.
1043
1044 9. Made dump understand the LABEL= and UUID= notation
1045 both in /etc/fstab and on the command line. Note that
1046 you will need the /proc filesystem in order to use
1047 these notations. Thanks to Erik Troan <ewt@redhat.com>
1048 for providing the patch.
1049
1050 Changes between versions 0.4b17 and 0.4b18 (released June 30, 2000)
1051 ===================================================================
1052
1053 1. Fixed a potential buffer overflow in restore. Thanks
1054 to Stan Bubrouski <satan@fastdial.net> for reporting
1055 the bug.
1056
1057 2. Fixed a readline-related bug which prevented
1058 'cat DUMPFILE | restore -i -f -' from working. Thanks
1059 to Charles Karney <karney@users.sourceforge.net>
1060 for the bug report.
1061
1062 3. Changed a few "panic" into "exit", causing restore to
1063 be more stable against some attacks (like the last one
1064 reported on Bugtraq, although the last version of restore
1065 was not vulnerable - just dumped core). Thanks to
1066 Andreas Hasenack <andreas@conectiva.com.br> for reporting
1067 the bugs to me.
1068
1069 4. Removed the suid-root bit on dump and restore in the
1070 default build (and generated RPMs). It should be safer
1071 now. Users who need the suid-root capabilities in order
1072 to make network backups should read first the man page
1073 of dump and enable the suid bit manually.
1074
1075 5. Added -ltermcap to the compile parameters for restore
1076 when using readline support, in order to make the compile
1077 process work on some systems (SuSE ?). Thanks to
1078 Patrik Schindler <poc@pocnet.net> for reporting the bug.
1079
1080 Changes between versions 0.4b16 and 0.4b17 (released June 1st, 2000)
1081 ====================================================================
1082
1083 1. The -F script is called now *only* at the end of a tape,
1084 not at the end of the dump. Thanks to Christian Haul
1085 <haul@informatik.tu-darmstadt.de> for the bug report.
1086
1087 Normally, the device name and the current volume number
1088 are passed on the command line of the script. However,
1089 if you want the old style script (with no arguments
1090 passed on the command line), you can enable it in
1091 configure with the --enable-oldstylefscript.
1092
1093 2. Use posix signal handling to preserve dump functionality
1094 with libc5. Thanks to Jeff Johnson <jbj@redhat.com> for
1095 the patch.
1096
1097 3. Made the exit code of restore in case of a 'C'ompare
1098 command reflect the result of the compare. An exit status
1099 of 0 means the dump archive is correct, 1 means tape errors,
1100 2 means that some files were modified. Thanks to Kenneth Porter
1101 <shiva@well.com> for the suggestion.
1102
1103 4. Made (finally) quotes work in interactive restore.
1104
1105 5. Small fixes in order to allow dump to compile with a
1106 really old e2fsprogs version. Thanks to Ian Zimmerman
1107 <itz@speakeasy.org> for the bug report.
1108
1109 6. Add GNU readline capabilities to interactive restore.
1110 Use configure's --enable-readline flag to enable this feature.
1111 Thanks to Patrik Schindler <poc@pocnet.net> for the
1112 suggestion.
1113
1114 7. Do the compare on the fly in restore 'C' mode (this will
1115 allow not to exhaust the available /tmp space when
1116 ccmparing large files). Thanks to Kenneth Porter
1117 <shiva@well.com> for the suggestion.
1118
1119 Changes between versions 0.4b15 and 0.4b16 (released March 11, 2000)
1120 ====================================================================
1121
1122 1. Fixed some several duplicate 'DUMP: DUMP:' in the
1123 output of dump.
1124
1125 2. Corrected the estimation of blocks to dump. Note that
1126 this is still wrong for large directory inodes, due
1127 to the size of a BSD directory entry on the tape
1128 (estimation is based on the size of the ext2 directory,
1129 which is smaller because it doesn't contain the null
1130 character at the end).
1131
1132 3. Made dump report the total number of MB written to
1133 tape at the end of run. Thanks to W. Reilly Cooley
1134 <wcooley@nakedape.cc> for the patch.
1135
1136 4. Added the -X option to restore giving the possibility
1137 to read the names of the files to be extracted/listed
1138 from a text file (in addition of the command line).
1139 Thanks to Dejan Muhamedagic <dejan@quant-x.com> for the
1140 patch.
1141
1142 5. Added the device name and the current volume number
1143 as arguments to the end of tape script (dump -F option).
1144
1145 6. Made the multi-volume dump work again (after having
1146 broken it in 0.4b15).
1147
1148 Changes between versions 0.4b14 and 0.4b15 (released March 2, 2000)
1149 ===================================================================
1150
1151 1. Added a prompt command in interactive restore mode. Thanks
1152 to Andreas Dilger <adilger@home.com> for the patch.
1153
1154 2. Fixed a buffer overflow problem in dump (caused by
1155 not checking the size of the filesystem parameter).
1156 Thanks to Kim Yong-jun <loveyou@hackerslab.org> for
1157 reporting this on Bugtraq (and to several dump users
1158 who forwarded me his mail).
1159
1160 3. Added the '-F script' option to dump in order to
1161 launch a script at the end of each tape (to be used
1162 with a tape changer, or to notify the sysadmin by
1163 pager etc.).
1164
1165 4. Fixed a bug in restore compare code caused by the changes
1166 I made in 0.4b14.
1167
1168 5. Fixed the treatment of options using the old BSD syntax
1169 in both dump and restore.
1170
1171 Changes between versions 0.4b13 and 0.4b14 (released February 10, 2000)
1172 =======================================================================
1173
1174 1. Fixed a bug in dump which may have caused invalid deleted
1175 directories to be dumped out if they were deleted after the
1176 mapping pass. This could occure on active filesystem and lead
1177 to heap corruption (causing dump malfunction in many possible ways).
1178 Thanks to Jan Sanislo <oystr@cs.washington.edu> for finding this
1179 bug and submitting the patch.
1180
1181 2. Fixed the handling of the filesystem parameter in dump. A
1182 '/mnt/disk' parameter caused the disk contents to be dumped,
1183 but a '/mnt/disk/' parameter caused the mountpoint directory
1184 to be dumped (generally an empty directory).
1185
1186 3. Improved the output of dump in order to tell which directory
1187 it is currently dumping (when dumping a subtree).
1188
1189 4. Added the '-e' exclude inode option to dump. Thanks to
1190 Isaac Chuang <ike@isl.stanford.edu> for contributing with the patch.
1191
1192 5. Added a REPORTING-BUGS file in order to provide a guide
1193 on how to correctly report a bug in dump/restore.
1194
1195 6. Fixed a restore problem when restoring a hard link to an inode
1196 having the immutable or append-only attribute set. Thanks to
1197 Ambrose Li <acli@mingpaoxpress.com> for submitting the patch.
1198
1199 7. Fixed a compatibility problem between dumps made on little
1200 endian machines (the format was correct) and big endian
1201 machines (the format was incorrect). This fix break the
1202 compatibility with the older dumps made on big endian
1203 machines (sparc, m86k, ppc etc). For the first time in
1204 linux dump's history (I believe), the dumps made by *BSD,
1205 Linux/alpha, Linux/sparc and Linux/x86 are compatible,
1206 so interchangeable. Thanks to Rob Cermak
1207 <cermak@ahab.rutgers.edu> for submitting the bug and
1208 helping me test the fix.
1209
1210 8. Fixed the way dump reports the remaining percent/time, if
1211 the number of blocks actually dumped exceeds the estimated
1212 number of blocks. Thanks to Jean-Paul van der Jagt
1213 <jeanpaul@dutepp0.et.tudelft.nl> for reporting the bug.
1214
1215 Changes between versions 0.4b12 and 0.4b13 (released January 21, 2000)
1216 ======================================================================
1217
1218 1. Small Y2K fix in the man pages :). Thanks to Bernhard Sadlowski
1219 <sadlowsk@Mathematik.Uni-Bielefeld.DE> for reporting the bug.
1220
1221 2. Removed the requirement to build the RPM as root from the
1222 spec file. Thanks to Christian Weisgerber
1223 <naddy@mips.rhein-neckar.de> for submitting this.
1224
1225 3. Fixed a bug in dump related to the 'filetype' feature of ext2fs,
1226 causing dump to block when dumping really huge filesystems.
1227 Many thanks to Patrik Schindler <poc@pocnet.net> for
1228 helping me find this bug.
1229
1230 4. Fixed the treatment for an interrupt signal when dump access
1231 the remote tape through RSH. Thanks to Christian Weisgerber
1232 <naddy@mips.rhein-neckar.de> for providing the patch.
1233
1234 5. Fixed a bug which was causing dump/restore to display
1235 garbage characters instead of the remote host name.
1236
1237 Changes between versions 0.4b11 and 0.4b12 (released January 8, 2000)
1238 =====================================================================
1239
1240 1. Small fix in the dump man page. Thanks to Thorsten Kukuk
1241 <kukuk@suse.de> for submitting the patch.
1242
1243 2. Fix for the exit code when using the size estimate option of
1244 dump. Thanks to Matti Taskinen <mkt@rni.helsinki.fi> for
1245 submitting the patch.
1246
1247 3. Handle EINTR in atomical reads/writes in dump, which was causing
1248 dump to fail on some systems. Thanks to Eric Jergensen
1249 <eric@dvns.com> for reporting the bug and submitting the patch.
1250
1251 4. Handle more than 16 characters for the device names in dumpdates.
1252 (up to 255 now). Thanks to Rainer Clasen <bj@ncc.cicely.de> for
1253 tracking down the problem and proposing the solution.
1254
1255 5. Fixed a bug in dump which prevented the creation of the
1256 dumpdates file when doing a 0-level dump without already
1257 having a dumpdates file. Thanks to Patrik Schindler
1258 <poc@pocnet.net> for reporting the bug.
1259
1260 6. Changed the way dump 'S' flag reports the size estimate
1261 from number of blocks into bytes (making it compatible
1262 with the Solaris version, and simplifying things for
1263 amanda users). Thanks to Jason L Tibbitts III
1264 <tibbs@math.uh.edu> for reporting the bug.
1265
1266 7. Fixed a compatibility problem in linux/alpha dump tape format.
1267 Now the linux/alpha dump are (again) compatible with the
1268 other dump formats. But this breaks compatibility with
1269 older dumps made on alpha. Thanks to Mike Tibor
1270 <tibor@lib.uaa.alaska.edu> for helping me in finding this bug.
1271
1272 Changes between versions 0.4b10 and 0.4b11 (released December 5, 1999)
1273 ======================================================================
1274
1275 1. Added a '--enable-kerberos' to configure.
1276
1277 2. Added a 'S' option to dump which determines the amount of space
1278 that is needed to perform the dump without actually doing it, similar
1279 to the Sun's ufsdump 'S' option. Patch contributed by Rob Cermak
1280 <cermak@ahab.rutgers.edu>.
1281
1282 3. Added a 'M' multi-volume option to dump and restore which enables
1283 dumping to multiple files (useful when dumping to an ext2
1284 partition to make several dump files in order to bypass the 2GB
1285 file size limitation). The 'f' argument is treated as a prefix and
1286 the output files will be named <prefix>001, <prefix>002 etc. With
1287 the 'M' flag, restore automatically selects the right file without
1288 asking to enter a new tape each time.
1289
1290 4. Fixed a memory leak which was causing dump to grow very big
1291 (270MB when dumping a 10GB filesystem...). Thanks to Jason
1292 Fearon <jasonf@netrider.org.au> for reporting the bug.
1293
1294 Changes between versions 0.4b9 and 0.4b10 (released November 21, 1999)
1295 ======================================================================
1296
1297 1. Make configure test if the system glob routines support
1298 extended syntax (ALTDIRFUNC). If not, use the internal glob
1299 routines instead of system ones. Thanks to Bernhard Sadlowski
1300 <sadlowsk@Mathematik.Uni-Bielefeld.DE> for reporting the bug
1301 and helping me resolve this and other minor libc5 compiling
1302 glitches.
1303
1304 2. Fix a problem when dumping a ext2fs with the 'filetype'
1305 feature enabled. Thanks to Patrick J. LoPresti
1306 <patl@cag.lcs.mit.edu> for reporting the bug and to
1307 Theodore Y. Ts'o <tytso@mit.edu> for providing the patch.
1308
1309 3. Made the nodump flag work on directories. A directory which
1310 has the nodump flag gets never dumped, regardless of its
1311 contents.
1312
1313 4. Integrate a patch from Jeremy Fitzhardinge <jeremy@goop.org>
1314 which allows dump on an active ext3 filesystem. However, this
1315 is a "quick and dirty" patch which enables backup of an ext3
1316 filesystem through the ext2 compatibility (by ignoring the
1317 NEEDS_RECOVERY bit). The journal file is not recognized and
1318 it is dumped (it should not).
1319
1320 5. Test the superblock compatibility flags when dumping, in order
1321 to be sure that we know how to deal with specific features.
1322
1323 Changes between versions 0.4b8 and 0.4b9 (released November 5, 1999)
1324 ====================================================================
1325
1326 1. Use lchown instead of chown, fixing a possible security problem
1327 when restoring symlinks (a malicious user could use this
1328 to deliberately corrupt the ownership of important system files).
1329 Thanks to Chris Siebenmann <cks@utcc.utoronto.ca> for detecting
1330 this and providing the patch.
1331
1332 Changes between versions 0.4b7 and 0.4b8 (released November 3, 1999)
1333 ====================================================================
1334
1335 1. Put dump sources under CVS, added Id tags in all files so
1336 one can use 'ident' on binary files.
1337
1338 2. Added the dump/restore version in the usage text so one can
1339 easily verify the version he is using.
1340
1341 3. Small patch from Nuno Oliveira <nuno@eq.uc.pt> which fixes
1342 a va_start/va_end problem on linux-ppc (always call va_start
1343 va_end in pairs each time we use a vararg function).
1344
1345 4. Added again the DT_* constants because old libc does not
1346 contain them :(. Thanks to Eric Maisonobe <virnet@nat.fr>
1347 for submitting the bug report.
1348
1349 5. Use ext2fs_llseek instead of llseek. With recent e2fsprogs
1350 this should enable dumping big (huge) filesystems.
1351
1352 6. Added the RSH environment variable in order to be able to
1353 use a rsh replacement like ssh when doing remote backups (and
1354 bypass the security limitations of rcmd). Now you can do remote
1355 backups without being root (or making dump setuid root).
1356
1357 7. Modified again the way dumpdates works. For incremental dumps,
1358 we need to read dumpdates even if we are not using 'u' option.
1359 Thanks to Bdale Garbee <bdale@gag.com> for his ideas on how
1360 this should work.
1361
1362 Changes between versions 0.4b6 and 0.4b7 (released October 8, 1999)
1363 ===================================================================
1364
1365 1. Removed the 'k' flag from the restore 'about' text if kerberos
1366 was not compiled in.
1367
1368 2. Prototyped (f)setflags from e2fsprogs and corrected the calls
1369 to them (fsetflags takes a char*, setflags an open fd!).
1370
1371 3. (f)setflags is called only if the flags aren't empty. If the
1372 file is a special file, a warning is printed, because changing
1373 flags implies opening the device. Normally, a special file
1374 should not have any flag... (Debian bug #29775, patch provided
1375 by Abhijit Dasgupta <abhijit@ans.net>).
1376
1377 4. Made possible to dump a file system not mentioned in /etc/fstab.
1378 (Debian bug #11904, patch provided by Eirik Fuller <eirik@netcom.com>).
1379
1380 5. Changed the default behaviour to not create dumpdates
1381 unless 'u' option is specified. Removed the old "debian-patch"
1382 which provided the same thing. (Debian bug #38136, #33818).
1383
1384 6. Removed all those dump*announce, since they were getting old...
1385
1386 7. Added warning messages if dumpdates does not exist and
1387 when an inferior level dump does not exist (except for a level 0
1388 dump).
1389
1390 8. Debugged the glob calls in interactive mode: restore used a
1391 dirent struct which was different from the /usr/include/dirent.h
1392 one (this used to work, is it a glibc2 change?), so none of the
1393 compat glob (which used /usr/include/dirent.h) or the system glob
1394 worked. Restore use now the system dirent (and the system
1395 DT_* constants), which are compatible with BSD ones.
1396
1397 9. Added a configure flag (--with-dumpdatespath) to specify
1398 the location of dumpdates file. By default, it is
1399 /etc/dumpdates.
1400
1401 10. Added the "AUTHOR" and "AVAILABILITY" sections and
1402 included the current date/version in man pages.
1403
1404 11. Corrected the estimation of remaining time when
1405 the operator doesn't change the tapes quickly enough. This
1406 was an old bug, I thought I corrected it, and discovered
1407 that in fact it was corrected in two different places, so
1408 the results canceled each other...
1409
1410 Changes between versions 0.4b5 and 0.4b6 (released October 1, 1999)
1411 ===================================================================
1412
1413 1. Integrated multiple patches from RedHat, Debian and SuSE:
1414
1415 - tweak dump/itime.c to not try to read dumpdates if the 'u' option
1416 isn't specified.
1417 - several fixes in the man pages.
1418 - update the default tape device to /dev/st0.
1419 - many updates for Linux Alpha (byte ordering, size_t etc).
1420 - buffer overruns.
1421 - use environment variable for TMPDIR (instead of /tmp).
1422 - use sigjmp_buf instead of jmp_buf (RedHat bug #3260).
1423 - workaround egcs bug (RedHat bugs #4281 and #2989).
1424 - wire $(OPT) throughout Makefile's.
1425
1426 2. Upgrade the dump revision to 1, making possible to dump filesystems
1427 made with e2fsprogs-1.15 or newer. Nothing seems to break...
1428
1429 3. Fix some compile warnings, prototype all functions.
1430
1431 4. Use glibc err/glob instead of internal compatibility
1432 routines (only if available).
1433
1434 5. Fix a compile error on Linux 2.2.7 / libc5 (5.4.44) (patch provided
1435 by Bernhard Sadlowski <sadlowsk@mathematik.uni-bielefeld.de>).
1436
1437 Changes between versions 0.4b4 and 0.4b5 (released September 22, 1999)
1438 ======================================================================
1439
1440 1. Integrated the changes from FreeBSD-3.1-RELEASE
1441 (mostly bug fixes, buffer overruns, dump has now an "automatic
1442 tape length calculation" flag, dump/restore can use kerberos now
1443 (this is NOT tested), use environment variables for TAPE and
1444 RMT etc.).
1445
1446 2. Integrated three RedHat patches ("glibc", "kernel" and "bread" patches)
1447
1448 3. Corrected a bug in restore when using 'C' option with multi-volumes
1449 tapes (files splited accros two tapes give "size changed" errors
1450 when comparing).
1451
1452 4. Corrected the long standing bug when dumping multiple tapes.
1453 This works for me, needs further testing.
1454
1455 Changes between versions 0.4b3 and 0.4b4 (released January 17, 1997)
1456 ====================================================================
1457
1458 1. Dump now runs correctly on kernels 2.1.x
1459 Fix made by Gerald Peters <gapeters@worldnet.att.net>
1460
1461 Changes between versions 0.4b2 and 0.4b3
1462 ========================================
1463
1464 1. Use realpath() if available
1465
1466 2. Report statistics
1467
1468 Changes between versions 0.4b1 and 0.4b2
1469 ========================================
1470
1471 1. Fixed the bug fix from Greg Lutz (I had made a mistake when integrating
1472 the patch)
1473
1474 2. Fixed restore to make it able to read FreeBSD 2.x dumps again
1475
1476 3. Fixed configure.in to correctly handle --enable-rmt
1477
1478 Changes between versions 0.3 and 0.4b1
1479 ======================================
1480
1481 1. Integrated the changes from 4.4BSD-Lite2
1482
1483 2. Integrated the patches from Debian and Red Hat
1484
1485 3. Portability changes: use the __u32, __u16, __s32, and __s16 types
1486
1487 4. Changed dump to use the Ext2fs library to get block addresses. This
1488 should solve the endianness problem on SparcLinux.
1489
1490 5. Created a configure.in file (shamelessly stolen from the e2fsprogs
1491 distribution's one) to use autoconf
1492
1493 6. Fixed a few minor bugs
1494
1495 Changes between versions 0.2e and 0.2f
1496 ======================================
1497
1498 1. Added the creation of named pipes (fifos) in restore.
1499
1500 2. Added the -N flag in the restore manual page.
1501
1502 3. Added the file kernel.patch which contains the llseek() optimization
1503 patch for 1.2.x kernels.
1504
1505 4. Fixed a bug in the restoration of symbolic links: owner and group were
1506 not restored.
1507
1508 5. Integrated some changes from FreeBSD 2.2.
1509
1510 6. Added a call to ftruncate() after restoring each file to restore
1511 correctly files ending by a hole.
1512
1513 Changes between versions 0.2d and 0.2e
1514 ======================================
1515
1516 1. Fixed a bug in the "set owner/mode" process. Every file was restored
1517 with owner = root (0) and group = root/wheel/whatever (0).
1518
1519 Changes between versions 0.2c and 0.2d
1520 ======================================
1521
1522 1. Dump is now able to backup 2GB+ filesystems.
1523
1524 2. Dump and restore can now be linked as static binaries.
1525
1526 Changes between versions 0.2b and 0.2c
1527 ======================================
1528
1529 1. Fixed a bug when dumping ``slow'' (i.e. normal) symbolic links.
1530
1531 Changes between versions 0.2a and 0.2b
1532 ======================================
1533
1534 1. Really fixed the bug that I should have corrected in 0.2a.
1535
1536 2. Enabled optimization again.
1537
1538 Changes between versions 0.2 and 0.2a
1539 =====================================
1540
1541 1. Disabled the optimization during compilation.
1542
1543 Changes between versions 0.1 and 0.2
1544 ====================================
1545
1546 1. Fixed a bug in fstab.c which caused a null pointer to be stored in
1547 the fs_type field (actually, I modified the file fstab.c to make it
1548 use the mntent functions).
1549
1550 2. Dump and restore now use a 4.3 BSD compatible dump format. Backups
1551 made by dump should be readable by the BSD restore and backups made
1552 by the BSD dump should be readable by restore. Unfortunately, this
1553 means that the dump format has changed between version 0.1 and version
1554 0.2 :-(
1555
1556 3. Dump is now able to backup a subtree, it is no longer limited to whole
1557 filesystems like the BSD version.
1558
1559 4. Dump now uses ext2_llseek() so it is able to backup filesystems bigger
1560 than 2 GB.
1561
1562 Changes between versions 0.0 and 0.1
1563 ====================================
1564
1565 1. Now create links rdump and rrestore during the `make install' step.
1566
1567 2. Linux port specific bugs added to the manual pages
1568
1569 3. Incorrect estimation of the number of tapes blocks fixed when doing
1570 incremental backups.
1571
1572 4. Better ls-like format in restore in interactive mode.