> In the gnumach source, the ide-cd.c driver (which finds my ATAPI IDE > CD-Rom drive) supports all of the ioctl functionality available on linux > (in fact, it is in the linux sub-folder; hopefully I'm not > misunderstanding something here). This leads me to believe that there is > driver support in mach, but when I open /dev/hd1 (which mounds data CDs > correctly), any time I try to issue audio CD-specific ioctl requests, I > get "inappropriate ioctl for device". Is there something about how the > device files work with translators in /dev or does hurd > treat my CD drive as a regular ide device (using ide.c as > the driver) because it is located at /dev/hd1? If you are using the ``devices'' in dev, you are making a mistake. `/dev/hd1' is a store, i.e. it is managed by /hurd/storeio, which, for obvious reasons, does not understand those ioctls. If you want to send ioctls to the _device_, you will need to actually open the device. At the moment, devices do not live in the file system hierarchy. Rather, they are accessable via port send rights, which can be obtained using the device_open (found in <device/device.h>) function. You can see a good example of this functionality in hurd/utils/devprobe.c.
Description: PGP signature