On Fri, 27 Jul 2001, David Burgess wrote:

> Bill Studenmund wrote:
> >
> > On Thu, 26 Jul 2001, Michael Wolfson wrote:
> >
> > > I was thinking along the lines of:
> > > com0 at [cardbus hardware info]
> > > com0 attached to /dev/tty10
> > > com1 at [usb hardware info]
> > > com1 attached to /dev/ttyU01
> >
> > Because the names of the device nodes are the result of a policy decision
> > made by the super user and which is denoted in /dev. i.e. device name
> > generation is not the business of the kernel. :-)
>
> I'm still doing the research, but the major/minor numbers of each of
> devices would determine what it was called.

It'd be fine for the kernel to print out major & minor numbers - those are
its domain.

> > > That'd make life a lot simpler.  Maybe as a compromise, a simple program
> > > (called something like ttyinfo) which just spits out:
> > > com0 is /dev/tty10
> > > com1 is /dev/ttyU01
> > > zstty0 is /dev/tty0
> > > zstty1 is /dev/tty1
> >
> > Sounds fine.
>
> I missed something important here.  Normally, you'd do this with
> either hard links or duplicate allocations against the device numbers,
> once you know what the numbers are going to be.

The printout would be fine too.

Note: links (hard or soft) would be safer than duplicate allocations. That
way there'd be one set of permissions on a given device.

> What I'd expect to see would be something like 'com0 is 30/10, com1
> is 31/0' at which point you would establish the 'symbolic' name
> from mknod.
>
> Or are we working at a whole different level?

In the past we've just assumed the sys admin is familiar enough with the
OS to know what is happening.

Take care,

Bill