Description of problem:
My driver disk, which was created in the same way as one that worked for Red Hat
Linux 8.0, has been broken since Red Hat Linux 9. It is for installing from
parallel port CD-ROM drives.
Version-Release number of selected component (if applicable):
Red Hat Linux 9, current Severn betas.
Steps to Reproduce:
1. Boot from bootdisk.img with 'linux dd hdc=none hdd=none'
2. Insert bpckdd.img when prompted
3. Try to get as far as a mount attempt from /dev/pcd0
Fails miserably because anaconda got dumber.
I'll attach the script I use to create bpckdd.img. Briefly though, it creates this:
"Parallel-port IDE CDROM"
paride: parport_pc parport_serial
pcd: aten bpck bpck6 comm dstr epat epia fit2 fit3 friq frpw kbic ktti on20
The problem boils down to: how does anaconda expect to know what device node to
attempt to mount if it is a parallel port device?
IMHO, if all else fails it should take the name of the driver (pcd) and attempt
to create /dev/pcd0 (which it knows the major/minor for), and try that.
Created attachment 94409 [details]
We don't at all support using the parallel port cdrom drive stuff anymore and a
lot of the random probing that would allow us to find these has now been pulled
while cleaning up cruft.
No, you don't understand: I have created my own, non-supported, driver disk for
providing the drivers.
Or is it explicit Red Hat policy to actively *prevent* such third-party driver
disks from working now?
We allow third-party driver disks for things that are, eg, SCSI, but that
doesn't generalize out to parallel port cd-rom drives which live in their own
universe in a lot of ways (especially wrt probing and device nodes).