> Hmmm, this is totally different that what I've been lead to believe, but
> I have no doubt that you're more knowledgable than me.  But I have seen
> cases where multiple sessions get mounted as different volumes on a
> Macintosh desktop, but that's what is NOT known as multi-volume, right? :-)

Yes, albeit, as I said in my previous post, some people keep calling those
things multi-volume. But that term is used for something different in ISO-9660
terminology.

> Hmmm, the more I think about it, the more I realize that using disklabel
> perhaps isn't the right thing to do.  The block numbers in the ISO filesystem
> are probably absolute, and shouldn't be offset-corrected by the disklabel
> routine.

That's correct. (At least it should. I can imagine people writing something
different on their CDs. In fact, thinking about it again, I'd not be surprised
if those "multi-volume" CDs do have relative addresses instead of absolute
ones).

> It almost sounds like it makes more sense to add functionality to
> mount_cd9660 so you could specifiy a session number.  You could then have
> it read the disklabel, figure out the start of the ISO filesystem, and pass
> that to the mount system call.  What do people think of that idea?

What disklabel are you referring to?

It should request the start of the requested session from the driver, read
the volume labels (block 16 and following) from there, and use the block
numbers given as absolute addresses on the CD. By default, it should use
the last session (requested by another ioctl).

If the assumption above is correct (that those "multi-volume" CDs are using
relative addresses), we proably need another flag to allow mounting of those,
too.
--
ws@TooLs.DE     (Wolfgang Solfrank, TooLs GmbH) 	+49-228-985800