Re: DVD 1 of squeeze 6.0.1.a not OK?

["Thomas Schmitt" <scdbackup@gmx.net>]

Hi,

i still ponder about the supposed MBR allergy of the Macs.

Maybe one should make another test by transplanting the first 32 kB of
the i386 image into the first 32 kB of the sparc image.
If the allergy theory is right, then that should not mount any more:

  dd if=debian-6.0.1a-i386-DVD-1.iso bs=2K count=16 >test_mix.iso
  dd if=debian-6.0.1a-sparc-CD-1.iso bs=2K skip=16 >>test_mix.iso

If it mounts nevertheless, then one would have to explore the subtle
differences between genisoimage and xorriso in detail. I already
made a little list of suspects. (None convincing, though.)

------------------------------------------------------------------------

me:
> > A simple solution might be little de-MBR programs [...]
> >  Seek byte 0, write 32 kB of zeros, keep all other bytes unaltered.
Anne Bezemer:
> dd with conv=notrunc should do the trick.

I should read the dd man page more often.


> Of course, that way you change the file in-place ("make a backup first,
> if you want to keep the original").

Yep. Plus the risk to patch the wrong file.

The advantage of specialized programs would be that they could
insist in file name patterns like "debian-*.iso" and check
bytes 32768 to 32773 whether they are {0x01,'C','D','0','0','1'}.
That's a magic number for ISO 9660 and would thus tell with some
probability that patching the first 32 kB will not destroy directory
tree or file data.


> Also, publishing an official md5sum of the patched file would be a good
> idea.

Computable on the server side without actually patching an image:

  (dd bs=2K count=16 if=/dev/zero ; dd bs=2K skip=16 if=debian...iso) | md5sum

Nevertheless, one could argue that verifying the checksums before
patching should suffice if patching itself is trustworthy.


Have a nice day :)

Thomas