Still no hfsdump
Dump in Tiger version 10.4.3 is still broken. It has been broken since 10.0.0. It cannot read any filessystem created with Tiger - UFS or HFS+; because dump looks for at "magic number" that Tiger does not include in its partition table.
Somebody really needs to fix it, or write hfsdump and hfsrestore. Its beyond me.
here are the errors I get:
% dump -0 -f dumphfs /Volumes/hfs
DUMP: Date of this level 0 dump: Sat May 7 12:01:21 2005
DUMP: Date of last level 0 dump: the epoch
DUMP: Dumping /Volumes/hfs to dumphfs
DUMP: bad sblock magic number
DUMP: The ENTIRE dump is aborted.
don't even think about it, that does not work either:
% df -k
...
/dev/disk1s2 39065960 36268408 2797552 93% /Users/bob
/dev/disk2s2 2520 88 2432 3% /Volumes/hfs
% dump -0 -f dumphfs /dev/disk2s2
DUMP: Date of this level 0 dump: Sat May 7 12:09:49 2005
DUMP: Date of last level 0 dump: the epoch
DUMP: Dumping /dev/disk2s2 to dumphfs
DUMP: Cannot open /dev/disk2s2
For the details, read the following from way back in 2001
(http://www.omnigroup.com/mailman/archive/macosx-admin/2001-June/012204.html):
With considerable technical input from Dr. Marcel Bresink, I think we
now have a better handle on what is going on with dump/restore. Assuming
that the dump utility packaged with OSX 10.0.3, is compiled from the
corresponding Darwin 1.3 dump source, that source code contains the
following:
#define FS_MAGIC 0x011954 /* the fast filesystem magic number */
........
if (sblock->fs_magic != FS_MAGIC)
quit("bad sblock magic number\n");
...........
The magic number is that for the Berkeley FFS file system, and is not
for the HFS+ system. Thus the "bad sblock magic number" error when when
tries to use dump on HFS+ files. But one might think that dump could be
used for OS X UFS files, given that the UFS file system was supposedly
derived from the FFS system. Well, apparently that depends on when you
formatted your UFS file system. If you formatted with Mac OS X Server
1.x or a "DP" version of OS X, dump will work. (These used a BSD-Unix
partition table.) Mac OS X PB, 10.x, and OS X Server 10.x use a
"Macintosh" partition table, so the magic number is not where the dump
code can find it.
Since "restore" isn't shipped with OS X 10.x (even though it is part of
Darwin), all of this info on dump may be of little consequence. However,
IMHO this does point up some serious issues with Mac OS X:
1. Even though dump/restore may be a bit archaic, many Unix users
(myself included) would use dump/restore unless there is something
better in the offing.
2. Apple should not have allow a broken dump to appear in /sbin/dump, or
at least should have put a bug note on the man page.
3. While dump is certainly not restricted to FFS file systems (there is
Linux dump), I don't know what the implications of the retention of the
resource fork in HFS+ mean with respect to a dump for HFS+. BUT, if
Apple wants to see rapid adoption of OS X, they should have made sure
some sort of backup utility was shipped.
4. The fact that there are now two UFS systems is terrible. It opens the
whole question of just what Apple means when they say Darwin is the
underpinning of OS X. We need a consistent Unix base for OS X, and if
that is different from Darwin, Apple should say so.
--dick peskin
Richard L. Peskin, RLP Consulting, Londonderry, VT
http://www.rlpcon.com
http://www.caip.rutgers.edu/~peskin
Philip Wilk