]> git.wh0rd.org - dump.git/blob - CHANGES
Empty the sigaction structure to be portable.
[dump.git] / CHANGES
1 $Id: CHANGES,v 1.80 2000/08/30 08:55:15 stelian Exp $
2
3 Changes between versions 0.4b19 and 0.4b20 (released ???????????????)
4 =====================================================================
5
6 1. Fixed a small compilation problem due to a change
7 in the definintion of the struct sigaction in
8 glibc 2.0 / libc5. Thanks to Gunther Schlegel
9 <schlegel@riege.de> for reporting the bug and to
10 Dave Platt <dplatt@snulbug.mtview.ca.us> for suggesting
11 a fix.
12
13 2. Modified the label and uuid analysis in order to be
14 self-contained (independant of kernel/libc headers). This
15 should fix the compile with older kernel/libc headers and
16 will preserve the functionality. Thanks to Bernhard Erdmann
17 <bernhard.erdmann@gmx.de> for reporting the bug.
18
19
20 Changes between versions 0.4b18 and 0.4b19 (released August 20, 2000)
21 =====================================================================
22
23 1. Fixed the signal handling in dump (which I broke in 0.4b17)
24 which was causing several strange dump failures (dump
25 hanged or segmentation faults etc).
26
27 2. Specified the default blocksize in the dump man page.
28
29 3. Changed two info messages of restore to be written on stdout
30 instead of stderr in order to leave stderr only for errors
31 or warnings. Thanks to Stephen Atwell
32 <satwell@urbana.css.mot.com> for the suggestion.
33
34 4. Corrected an off by one calculation which prevented
35 dumping a directory having a 1 character name length.
36 Thanks to Bernhard Erdmann <bernhard.erdmann@gmx.de>
37 for reporting the bug.
38
39 5. Reinforce dump in order to allow only level 0 and no
40 -u option when dumping a subdirectory, and document
41 this in the man page. Thanks to Bernhard Erdmann
42 <bernhard.erdmann@gmx.de> for reporting the bug.
43
44 6. Fixed a small harmless bug in restore which caused
45 in some conditions hard links to be restored several
46 times (therefore generation some warning messages).
47 Thanks to Kenneth Porter <shiva@well.com> for
48 reporting the bug.
49
50 7. Updated the RPM spec file to the latest RedHat version,
51 providing FHS packaging and other cosmetic changes.
52 You will need RPM version >= 3.0.5 in order to build the RPM.
53
54 8. Updated the configure script to check for libtermcap
55 before checking for libreadline, because we need this
56 library in order to compile the readline support.
57
58 9. Made dump understand the LABEL= and UUID= notation
59 both in /etc/fstab and on the command line. Note that
60 you will need the /proc filesystem in order to use
61 these notations. Thanks to Erik Troan <ewt@redhat.com>
62 for providing the patch.
63
64 Changes between versions 0.4b17 and 0.4b18 (released June 30, 2000)
65 ===================================================================
66
67 1. Fixed a potential buffer overflow in restore. Thanks
68 to Stan Bubrouski <satan@fastdial.net> for reporting
69 the bug.
70
71 2. Fixed a readline-related bug which prevented
72 'cat DUMPFILE | restore -i -f -' from working. Thanks
73 to Charles Karney <karney@users.sourceforge.net>
74 for the bug report.
75
76 3. Changed a few "panic" into "exit", causing restore to
77 be more stable against some attacks (like the last one
78 reported on Bugtraq, although the last version of restore
79 was not vulnerable - just dumped core). Thanks to
80 Andreas Hasenack <andreas@conectiva.com.br> for reporting
81 the bugs to me.
82
83 4. Removed the suid-root bit on dump and restore in the
84 default build (and generated RPMs). It should be safer
85 now. Users who need the suid-root capabilities in order
86 to make network backups should read first the man page
87 of dump and enable the suid bit manually.
88
89 5. Added -ltermcap to the compile parameters for restore
90 when using readline support, in order to make the compile
91 process work on some systems (SuSE ?). Thanks to
92 Patrik Schindler <poc@pocnet.net> for reporting the bug.
93
94 Changes between versions 0.4b16 and 0.4b17 (released June 1st, 2000)
95 ====================================================================
96
97 1. The -F script is called now *only* at the end of a tape,
98 not at the end of the dump. Thanks to Christian Haul
99 <haul@informatik.tu-darmstadt.de> for the bug report.
100
101 Normally, the device name and the current volume number
102 are passed on the command line of the script. However,
103 if you want the old style script (with no arguments
104 passed on the command line), you can enable it in
105 configure with the --enable-oldstylefscript.
106
107 2. Use posix signal handling to preserve dump functionality
108 with libc5. Thanks to Jeff Johnson <jbj@redhat.com> for
109 the patch.
110
111 3. Made the exit code of restore in case of a 'C'ompare
112 command reflect the result of the compare. An exit status
113 of 0 means the dump archive is correct, 1 means tape errors,
114 2 means that some files were modified. Thanks to Kenneth Porter
115 <shiva@well.com> for the suggestion.
116
117 4. Made (finally) quotes work in interactive restore.
118
119 5. Small fixes in order to allow dump to compile with a
120 really old e2fsprogs version. Thanks to Ian Zimmerman
121 <itz@speakeasy.org> for the bug report.
122
123 6. Add GNU readline capabilities to interactive restore.
124 Use configure's --enable-readline flag to enable this feature.
125 Thanks to Patrik Schindler <poc@pocnet.net> for the
126 suggestion.
127
128 7. Do the compare on the fly in restore 'C' mode (this will
129 allow not to exhaust the available /tmp space when
130 ccmparing large files). Thanks to Kenneth Porter
131 <shiva@well.com> for the suggestion.
132
133 Changes between versions 0.4b15 and 0.4b16 (released March 11, 2000)
134 ====================================================================
135
136 1. Fixed some several duplicate 'DUMP: DUMP:' in the
137 output of dump.
138
139 2. Corrected the estimation of blocks to dump. Note that
140 this is still wrong for large directory inodes, due
141 to the size of a BSD directory entry on the tape
142 (estimation is based on the size of the ext2 directory,
143 which is smaller because it doesn't contain the null
144 character at the end).
145
146 3. Made dump report the total number of MB written to
147 tape at the end of run. Thanks to W. Reilly Cooley
148 <wcooley@nakedape.cc> for the patch.
149
150 4. Added the -X option to restore giving the possibility
151 to read the names of the files to be extracted/listed
152 from a text file (in addition of the command line).
153 Thanks to Dejan Muhamedagic <dejan@quant-x.com> for the
154 patch.
155
156 5. Added the device name and the current volume number
157 as arguments to the end of tape script (dump -F option).
158
159 6. Made the multi-volume dump work again (after having
160 broken it in 0.4b15).
161
162 Changes between versions 0.4b14 and 0.4b15 (released March 2, 2000)
163 ===================================================================
164
165 1. Added a prompt command in interactive restore mode. Thanks
166 to Andreas Dilger <adilger@home.com> for the patch.
167
168 2. Fixed a buffer overflow problem in dump (caused by
169 not checking the size of the filesystem parameter).
170 Thanks to Kim Yong-jun <loveyou@hackerslab.org> for
171 reporting this on Bugtraq (and to several dump users
172 who forwarded me his mail).
173
174 3. Added the '-F script' option to dump in order to
175 launch a script at the end of each tape (to be used
176 with a tape changer, or to notify the sysadmin by
177 pager etc.).
178
179 4. Fixed a bug in restore compare code caused by the changes
180 I made in 0.4b14.
181
182 5. Fixed the treatment of options using the old BSD syntax
183 in both dump and restore.
184
185 Changes between versions 0.4b13 and 0.4b14 (released February 10, 2000)
186 =======================================================================
187
188 1. Fixed a bug in dump which may have caused invalid deleted
189 directories to be dumped out if they were deleted after the
190 mapping pass. This could occure on active filesystem and lead
191 to heap corruption (causing dump malfunction in many possible ways).
192 Thanks to Jan Sanislo <oystr@cs.washington.edu> for finding this
193 bug and submitting the patch.
194
195 2. Fixed the handling of the filesystem parameter in dump. A
196 '/mnt/disk' parameter caused the disk contents to be dumped,
197 but a '/mnt/disk/' parameter caused the mountpoint directory
198 to be dumped (generally an empty directory).
199
200 3. Improved the output of dump in order to tell which directory
201 it is currently dumping (when dumping a subtree).
202
203 4. Added the '-e' exclude inode option to dump. Thanks to
204 Isaac Chuang <ike@isl.stanford.edu> for contributing with the patch.
205
206 5. Added a REPORTING-BUGS file in order to provide a guide
207 on how to correctly report a bug in dump/restore.
208
209 6. Fixed a restore problem when restoring a hard link to an inode
210 having the immutable or append-only attribute set. Thanks to
211 Ambrose Li <acli@mingpaoxpress.com> for submitting the patch.
212
213 7. Fixed a compatibility problem between dumps made on little
214 endian machines (the format was correct) and big endian
215 machines (the format was incorrect). This fix break the
216 compatibility with the older dumps made on big endian
217 machines (sparc, m86k, ppc etc). For the first time in
218 linux dump's history (I believe), the dumps made by *BSD,
219 Linux/alpha, Linux/sparc and Linux/x86 are compatible,
220 so interchangeable. Thanks to Rob Cermak
221 <cermak@ahab.rutgers.edu> for submitting the bug and
222 helping me test the fix.
223
224 8. Fixed the way dump reports the remaining percent/time, if
225 the number of blocks actually dumped exceeds the estimated
226 number of blocks. Thanks to Jean-Paul van der Jagt
227 <jeanpaul@dutepp0.et.tudelft.nl> for reporting the bug.
228
229 Changes between versions 0.4b12 and 0.4b13 (released January 21, 2000)
230 ======================================================================
231
232 1. Small Y2K fix in the man pages :). Thanks to Bernhard Sadlowski
233 <sadlowsk@Mathematik.Uni-Bielefeld.DE> for reporting the bug.
234
235 2. Removed the requirement to build the RPM as root from the
236 spec file. Thanks to Christian Weisgerber
237 <naddy@mips.rhein-neckar.de> for submitting this.
238
239 3. Fixed a bug in dump related to the 'filetype' feature of ext2fs,
240 causing dump to block when dumping really huge filesystems.
241 Many thanks to Patrik Schindler <poc@pocnet.net> for
242 helping me find this bug.
243
244 4. Fixed the treatment for an interrupt signal when dump access
245 the remote tape through RSH. Thanks to Christian Weisgerber
246 <naddy@mips.rhein-neckar.de> for providing the patch.
247
248 5. Fixed a bug which was causing dump/restore to display
249 garbage characters instead of the remote host name.
250
251 Changes between versions 0.4b11 and 0.4b12 (released January 8, 2000)
252 =====================================================================
253
254 1. Small fix in the dump man page. Thanks to Thorsten Kukuk
255 <kukuk@suse.de> for submitting the patch.
256
257 2. Fix for the exit code when using the size estimate option of
258 dump. Thanks to Matti Taskinen <mkt@rni.helsinki.fi> for
259 submitting the patch.
260
261 3. Handle EINTR in atomical reads/writes in dump, which was causing
262 dump to fail on some systems. Thanks to Eric Jergensen
263 <eric@dvns.com> for reporting the bug and submitting the patch.
264
265 4. Handle more than 16 characters for the device names in dumpdates.
266 (up to 255 now). Thanks to Rainer Clasen <bj@ncc.cicely.de> for
267 tracking down the problem and proposing the solution.
268
269 5. Fixed a bug in dump which prevented the creation of the
270 dumpdates file when doing a 0-level dump without already
271 having a dumpdates file. Thanks to Patrik Schindler
272 <poc@pocnet.net> for reporting the bug.
273
274 6. Changed the way dump 'S' flag reports the size estimate
275 from number of blocks into bytes (making it compatible
276 with the Solaris version, and simplifying things for
277 amanda users). Thanks to Jason L Tibbitts III
278 <tibbs@math.uh.edu> for reporting the bug.
279
280 7. Fixed a compatibility problem in linux/alpha dump tape format.
281 Now the linux/alpha dump are (again) compatible with the
282 other dump formats. But this breaks compatibility with
283 older dumps made on alpha. Thanks to Mike Tibor
284 <tibor@lib.uaa.alaska.edu> for helping me in finding this bug.
285
286 Changes between versions 0.4b10 and 0.4b11 (released December 5, 1999)
287 ======================================================================
288
289 1. Added a '--enable-kerberos' to configure.
290
291 2. Added a 'S' option to dump which determines the amount of space
292 that is needed to perform the dump without actually doing it, similar
293 to the Sun's ufsdump 'S' option. Patch contributed by Rob Cermak
294 <cermak@ahab.rutgers.edu>.
295
296 3. Added a 'M' multi-volume option to dump and restore which enables
297 dumping to multiple files (useful when dumping to an ext2
298 partition to make several dump files in order to bypass the 2GB
299 file size limitation). The 'f' argument is treated as a prefix and
300 the output files will be named <prefix>001, <prefix>002 etc. With
301 the 'M' flag, restore automatically selects the right file without
302 asking to enter a new tape each time.
303
304 4. Fixed a memory leak which was causing dump to grow very big
305 (270MB when dumping a 10GB filesystem...). Thanks to Jason
306 Fearon <jasonf@netrider.org.au> for reporting the bug.
307
308 Changes between versions 0.4b9 and 0.4b10 (released November 21, 1999)
309 ======================================================================
310
311 1. Make configure test if the system glob routines support
312 extended syntax (ALTDIRFUNC). If not, use the internal glob
313 routines instead of system ones. Thanks to Bernhard Sadlowski
314 <sadlowsk@Mathematik.Uni-Bielefeld.DE> for reporting the bug
315 and helping me resolve this and other minor libc5 compiling
316 glitches.
317
318 2. Fix a problem when dumping a ext2fs with the 'filetype'
319 feature enabled. Thanks to Patrick J. LoPresti
320 <patl@cag.lcs.mit.edu> for reporting the bug and to
321 Theodore Y. Ts'o <tytso@mit.edu> for providing the patch.
322
323 3. Made the nodump flag work on directories. A directory which
324 has the nodump flag gets never dumped, regardless of its
325 contents.
326
327 4. Integrate a patch from Jeremy Fitzhardinge <jeremy@goop.org>
328 which allows dump on an active ext3 filesystem. However, this
329 is a "quick and dirty" patch which enables backup of an ext3
330 filesystem through the ext2 compatibility (by ignoring the
331 NEEDS_RECOVERY bit). The journal file is not recognized and
332 it is dumped (it should not).
333
334 5. Test the superblock compatibility flags when dumping, in order
335 to be sure that we know how to deal with specific features.
336
337 Changes between versions 0.4b8 and 0.4b9 (released November 5, 1999)
338 ====================================================================
339
340 1. Use lchown instead of chown, fixing a possible security problem
341 when restoring symlinks (a malicious user could use this
342 to deliberately corrupt the ownership of important system files).
343 Thanks to Chris Siebenmann <cks@utcc.utoronto.ca> for detecting
344 this and providing the patch.
345
346 Changes between versions 0.4b7 and 0.4b8 (released November 3, 1999)
347 ====================================================================
348
349 1. Put dump sources under CVS, added Id tags in all files so
350 one can use 'ident' on binary files.
351
352 2. Added the dump/restore version in the usage text so one can
353 easily verify the version he is using.
354
355 3. Small patch from Nuno Oliveira <nuno@eq.uc.pt> which fixes
356 a va_start/va_end problem on linux-ppc (always call va_start
357 va_end in pairs each time we use a vararg function).
358
359 4. Added again the DT_* constants because old libc does not
360 contain them :(. Thanks to Eric Maisonobe <virnet@nat.fr>
361 for submitting the bug report.
362
363 5. Use ext2fs_llseek instead of llseek. With recent e2fsprogs
364 this should enable dumping big (huge) filesystems.
365
366 6. Added the RSH environment variable in order to be able to
367 use a rsh replacement like ssh when doing remote backups (and
368 bypass the security limitations of rcmd). Now you can do remote
369 backups without being root (or making dump setuid root).
370
371 7. Modified again the way dumpdates works. For incremental dumps,
372 we need to read dumpdates even if we are not using 'u' option.
373 Thanks to Bdale Garbee <bdale@gag.com> for his ideas on how
374 this should work.
375
376 Changes between versions 0.4b6 and 0.4b7 (released October 8, 1999)
377 ===================================================================
378
379 1. Removed the 'k' flag from the restore 'about' text if kerberos
380 was not compiled in.
381
382 2. Prototyped (f)setflags from e2fsprogs and corrected the calls
383 to them (fsetflags takes a char*, setflags an open fd!).
384
385 3. (f)setflags is called only if the flags aren't empty. If the
386 file is a special file, a warning is printed, because changing
387 flags implies opening the device. Normally, a special file
388 should not have any flag... (Debian bug #29775, patch provided
389 by Abhijit Dasgupta <abhijit@ans.net>).
390
391 4. Made possible to dump a file system not mentioned in /etc/fstab.
392 (Debian bug #11904, patch provided by Eirik Fuller <eirik@netcom.com>).
393
394 5. Changed the default behaviour to not create dumpdates
395 unless 'u' option is specified. Removed the old "debian-patch"
396 which provided the same thing. (Debian bug #38136, #33818).
397
398 6. Removed all those dump*announce, since they were getting old...
399
400 7. Added warning messages if dumpdates does not exist and
401 when an inferior level dump does not exist (except for a level 0
402 dump).
403
404 8. Debugged the glob calls in interactive mode: restore used a
405 dirent struct which was different from the /usr/include/dirent.h
406 one (this used to work, is it a glibc2 change?), so none of the
407 compat glob (which used /usr/include/dirent.h) or the system glob
408 worked. Restore use now the system dirent (and the system
409 DT_* constants), which are compatible with BSD ones.
410
411 9. Added a configure flag (--with-dumpdatespath) to specify
412 the location of dumpdates file. By default, it is
413 /etc/dumpdates.
414
415 10. Added the "AUTHOR" and "AVAILABILITY" sections and
416 included the current date/version in man pages.
417
418 11. Corrected the estimation of remaining time when
419 the operator doesn't change the tapes quickly enough. This
420 was an old bug, I thought I corrected it, and discovered
421 that in fact it was corrected in two different places, so
422 the results canceled each other...
423
424 Changes between versions 0.4b5 and 0.4b6 (released October 1, 1999)
425 ===================================================================
426
427 1. Integrated multiple patches from RedHat, Debian and SuSE:
428
429 - tweak dump/itime.c to not try to read dumpdates if the 'u' option
430 isn't specified.
431 - several fixes in the man pages.
432 - update the default tape device to /dev/st0.
433 - many updates for Linux Alpha (byte ordering, size_t etc).
434 - buffer overruns.
435 - use environment variable for TMPDIR (instead of /tmp).
436 - use sigjmp_buf instead of jmp_buf (RedHat bug #3260).
437 - workaround egcs bug (RedHat bugs #4281 and #2989).
438 - wire $(OPT) throughout Makefile's.
439
440 2. Upgrade the dump revision to 1, making possible to dump filesystems
441 made with e2fsprogs-1.15 or newer. Nothing seems to break...
442
443 3. Fix some compile warnings, prototype all functions.
444
445 4. Use glibc err/glob instead of internal compatibility
446 routines (only if available).
447
448 5. Fix a compile error on Linux 2.2.7 / libc5 (5.4.44) (patch provided
449 by Bernhard Sadlowski <sadlowsk@mathematik.uni-bielefeld.de>).
450
451 Changes between versions 0.4b4 and 0.4b5 (released September 22, 1999)
452 ======================================================================
453
454 1. Integrated the changes from FreeBSD-3.1-RELEASE
455 (mostly bug fixes, buffer overruns, dump has now an "automatic
456 tape length calculation" flag, dump/restore can use kerberos now
457 (this is NOT tested), use environment variables for TAPE and
458 RMT etc.).
459
460 2. Integrated three RedHat patches ("glibc", "kernel" and "bread" patches)
461
462 3. Corrected a bug in restore when using 'C' option with multi-volumes
463 tapes (files splited accros two tapes give "size changed" errors
464 when comparing).
465
466 4. Corrected the long standing bug when dumping multiple tapes.
467 This works for me, needs further testing.
468
469 Changes between versions 0.4b3 and 0.4b4 (released January 17, 1997)
470 ====================================================================
471
472 1. Dump now runs correctly on kernels 2.1.x
473 Fix made by Gerald Peters <gapeters@worldnet.att.net>
474
475 Changes between versions 0.4b2 and 0.4b3
476 ========================================
477
478 1. Use realpath() if available
479
480 2. Report statistics
481
482 Changes between versions 0.4b1 and 0.4b2
483 ========================================
484
485 1. Fixed the bug fix from Greg Lutz (I had made a mistake when integrating
486 the patch)
487
488 2. Fixed restore to make it able to read FreeBSD 2.x dumps again
489
490 3. Fixed configure.in to correctly handle --enable-rmt
491
492 Changes between versions 0.3 and 0.4b1
493 ======================================
494
495 1. Integrated the changes from 4.4BSD-Lite2
496
497 2. Integrated the patches from Debian and Red Hat
498
499 3. Portability changes: use the __u32, __u16, __s32, and __s16 types
500
501 4. Changed dump to use the Ext2fs library to get block addresses. This
502 should solve the endianness problem on SparcLinux.
503
504 5. Created a configure.in file (shamelessly stolen from the e2fsprogs
505 distribution's one) to use autoconf
506
507 6. Fixed a few minor bugs
508
509 Changes between versions 0.2e and 0.2f
510 ======================================
511
512 1. Added the creation of named pipes (fifos) in restore.
513
514 2. Added the -N flag in the restore manual page.
515
516 3. Added the file kernel.patch which contains the llseek() optimization
517 patch for 1.2.x kernels.
518
519 4. Fixed a bug in the restoration of symbolic links: owner and group were
520 not restored.
521
522 5. Integrated some changes from FreeBSD 2.2.
523
524 6. Added a call to ftruncate() after restoring each file to restore
525 correctly files ending by a hole.
526
527 Changes between versions 0.2d and 0.2e
528 ======================================
529
530 1. Fixed a bug in the "set owner/mode" process. Every file was restored
531 with owner = root (0) and group = root/wheel/whatever (0).
532
533 Changes between versions 0.2c and 0.2d
534 ======================================
535
536 1. Dump is now able to backup 2GB+ filesystems.
537
538 2. Dump and restore can now be linked as static binaries.
539
540 Changes between versions 0.2b and 0.2c
541 ======================================
542
543 1. Fixed a bug when dumping ``slow'' (i.e. normal) symbolic links.
544
545 Changes between versions 0.2a and 0.2b
546 ======================================
547
548 1. Really fixed the bug that I should have corrected in 0.2a.
549
550 2. Enabled optimization again.
551
552 Changes between versions 0.2 and 0.2a
553 =====================================
554
555 1. Disabled the optimization during compilation.
556
557 Changes between versions 0.1 and 0.2
558 ====================================
559
560 1. Fixed a bug in fstab.c which caused a null pointer to be stored in
561 the fs_type field (actually, I modified the file fstab.c to make it
562 use the mntent functions).
563
564 2. Dump and restore now use a 4.3 BSD compatible dump format. Backups
565 made by dump should be readable by the BSD restore and backups made
566 by the BSD dump should be readable by restore. Unfortunately, this
567 means that the dump format has changed between version 0.1 and version
568 0.2 :-(
569
570 3. Dump is now able to backup a subtree, it is no longer limited to whole
571 filesystems like the BSD version.
572
573 4. Dump now uses ext2_llseek() so it is able to backup filesystems bigger
574 than 2 GB.
575
576 Changes between versions 0.0 and 0.1
577 ====================================
578
579 1. Now create links rdump and rrestore during the `make install' step.
580
581 2. Linux port specific bugs added to the manual pages
582
583 3. Incorrect estimation of the number of tapes blocks fixed when doing
584 incremental backups.
585
586 4. Better ls-like format in restore in interactive mode.