When mounting multi-session CD, CD label should be read from last (actual) session, not first one. How to repeat: Take a CD-RW, create a multi-session CD with any contents, use -V parameter for mkisofs. Create another session using different -V. Current behavior: CD label is read from first session Expected behavior: CD label is read from last session
Tested on hal-0.4.7, SuSE Linux 9.3 x86_64.
(I've sent a may to Kay Sievers so he can add him as Cc since Bugzilla wont let me do that for some reason) Yeah, that should probably be fixed in volume_id; Kay, any idea how difficult this is to fix? Long term, we should probably expose all the sessions (and partitions within) of the CD as device objects in hal and make fstab-sync (or other tools, such as pmount) do the right thing for adding fstab entries / mounting the filesystem. Unfortunately, the kernel doesn't expose these, but we can still do it all from user space.
Related bug: http://bugzilla.gnome.org/show_bug.cgi?id=99621 util-linux also has a heuristics to find continuation session (to be able to mount CD-Extra data track). There is not probable, that there will be partitions in sessions (because TOC of session often refers files with previous sessions). Normal behavior is mounting last session. Imagine multi-session CD as block media with unreadable blocks between sessions and a special block - TOC, which says, where you should search sessions. It is possible to create multiple mount points, if you insert multi-session CD, all with the same device node, but for old sessions with additional session=number in mount options. But it must be done in a user-friendly way (but mount point structure must be usable, if user wants mount points named as volume label and label is different in old sessions). Special case is be CD-Extra - audio tracks can be mounted as CDDA or played by audio CD player, but data track can be mounted as data. It is not sure, what label should be used for such media - data track ISO9660 volume id or Album title from CDDB.
> When mounting multi-session CD, CD label should be read from last > (actual) session, not first one. Good point. We sould do something here. > Yeah, that should probably be fixed in volume_id; Kay, any idea how difficult > this is to fix? That's a mess with all the CD formats. But if we decide what exactly we want to be supported, I will try to do that, sure. Btw: UDF can have partitons along with sessions too. Apple uses HFS, and partitions too. We need to set a limit here. :)
We probe the last session from an optical disk now. This should work much better and is probably the expected behavior in most cases. http://cvs.freedesktop.org/hal/hal/hald/linux2/probing/probe-volume.c?r1=1.6&r2=1.7
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.