In 3.5, audiocd:/ is a big mess. Most of the time it does nothing, and in the remaining cases it either crashes or uses 100% CPU. And it's not really KDE's fault, either. It's due to cdparanoia. See, the interface to cdda in cdp contains a CAM struct -- but this CAM struct only is valid for SCSI CDs. In the case that you have a regular ATAPI device, that CAM struct is not initialized to anything useful, just a 0 pointer. Dereffing the device_path is sortof safe, that's still NULL.
In any case, the code in KDE must check if the resulting cdrom_drive struct is
SCSI or not; if not, the CAM struct isn't to be trusted.
If it is not a SCSI device, there is _no_ way to find out what the actual
device name is - the cdda interface doesn't contain a buffer for the device
name in this case! Well, I suppose we could iterate over /dev/ to remap the
ioctl fd to a device name again, but that seems icky.
So, summary: ATAPI is broken because of cdparanoia, but the KDE code is broken
because it doesn't check enough.
--
These are your friends - Adem
GPG: FEA2 A3FE Adriaan de Groot
pgp4YNvhot5dX.pgp
Description: PGP signature
_______________________________________________ kde-freebsd mailing list [email protected] http://freebsd.kde.org/mailman/listinfo/kde-freebsd
