Re: PATCH: cdrecord: avoiding scsi device numbering for ide devices

[Vojtech Pavlik <vojtech@suse.cz>]

On Fri, Aug 06, 2004 at 12:18:39PM +0200, Joerg Schilling wrote:

> >> -	ioctl(f, SCSI_IOCTL_GET_IDLUN, &sg_id)) and ioctl(f, SCSI_IOCTL_GET_BUS_NUMBER, &Bus)
> >> 	do not return instance numbers if applied to a fd from /dev/hd*
> >> 
> >> 	This bug has been reported 2 years ago.
> >> 
> >> 	Time to fix: 10 minutes
> 
> >It has been argued several times that we will not doctor bus/id/lun on
> >atapi devices, because it does not make sense. So I'm regarding this as
> >invalid.
> 
> Sorry, I have more than 20 years of experiences in Kernel and Application 
> hacking. I am able to decide myself what is a bug and what is something that
> needs to be called at least "non-orthogonal behavior". This is devinitely 
>  a bug or missing orthogonality. See above to understand (These are 
> primarily SCSI devices and therefore they need to have instance numbers within 
> the set of devices in the SCSI device tree).
> 
> If you do not fix this, you just verify that Linux does not like it's users.
> Linux users like to call cdrecord -scanbus and they like to see _all_ SCSI 
> devices from a single call to cdrecord. The fact that the Linux kernel does not
> return instance numbers for /dev/hd* SCSI devices makes it impossible to
> implement a unique address space :-(
 
I'm a long time Linux and cdrecord user, and I must say I always hated
the -scanbus thingy. It's so much easier to just pass the /dev/hdc
device node, where I _know_ my CD burner lives than to have to figure
out what fake SCSI address cdrecord has made up and requires me to pass
to it, even when I have just a single CD burner in my system.

-- 
Vojtech Pavlik
SuSE Labs, SuSE CR