MkIsofs vs. dd

Hi,

I need to manage some master ISO9660 images created with mkisofs and in addition guarantee the integrity of copies that are made from those images. The way I've been doing this was by reading the data back using dd and computing a checksum, like so:

dd if=-dev-cdrom md5sum

That checksum could then be compared with the checksum of the master ISO9660 image.

Well, it soon turned out that this was a bummer since the amount of 0-filled sectors at the end is rather unpredictable. It seems to depend on the burner, and to lesser extent on the reader. This led to an enhanced way of retrieving the checksum from an image:

dd if=-dev-cdrom bs=2048 count=`isosizedev-cdrom -d 2048`

I'm still having problems with some ISO9660 images containing 0-filled sectors at the end, however. It appears that some drives simply truncate those, which renders them useless for the purpose of computing checksum.

I've two questions:

1) Is this dd count=`isosize ...` business a sane idea or is there some fundamental problem that makes computation of checksums an, err, more capricious venture?

2) How can 0-filled sectors at the end of ISO9660 images happen in the first place? Well, one way is by re-reading a CD and including some run-out sectors, of course, but are there others?

TIA -richy. -- Richard B. Kreckel