[ILUG] SCSI CD and DA problems

Niall O Broin niall at magicgoeshere.com
Mon May 29 15:16:12 IST 2000


On Mon, May 29, 2000 at 03:44:31AM +0100, Andy Ferguson(ilug) wrote:

> I don't know what the sr* devices are but they seem to be redirected
> somehow to equivalent generic devices. My CDROM is on sg2 so
> paranoia relocates sr0 as sg2

s?? devices are as follows:

st?	tape	sd?	disk	sr?	CDROM	sg?	generic

The kernel has special support for SCSI disks, tapes and CDROMs hence the
specific major device numbers allocated for that kind of device. sg devices
are generic i.e. no special support provided by the kernel (how could it  ? -
it doesn't know what they are) and to read a DA track you need to access the
CD drive as a generic device.  /dev/scd? are just links (not really, but
effectively) to the corresponding sr devices.

> The command will therefore work by specifying the generic device
> directly as in
> 
> cdparanoia -vsQ -d /dev/sg2
> or even
> cdparanoia -vsQ -d /dev/scd0

This did not work for me, but now it does. I'm buggered if I know what was
wrong. I know when I tried to access /dev/sg0 I was wrong (because sg? maps
directly to SCSI ID whereas sr? maps to CDROM drives as found. My only SCSI
CDR is at ID 3 so it's sr0 but the generic device is sg3. Under the current
device naming conventions that's reasonably logical, even if I didn't
realise it, but it's yet another argument for devfs [I was really pissed
with the transition from SunOS to Solaris device naming conventions - I
found /dev/sd0a to be a lot handier than /dev/dsk/c0t3d0s0 but I came around
eventually, especially when you're dealing with a big box you don't know,
when under the old system, the disks could be nearly anywhere]. 

However, when I was having problems, cdparanoia would work with neither. It
told me that if couldn't find a generic device but I did have sg support - I
burned a CD. 

So, problem solved. Thanks to those who made suggestions - I still have no
idea what was wrong, and that really annoys me.


Regards,


Niall  O Broin




More information about the ILUG mailing list