Re: [PATCH 2/2] i2c: Add devtree-aware iic support for PPC4xx

["Grant Likely" <grant.likely@secretlab.ca>]

On 10/15/07, Scott Wood <scottwood@freescale.com> wrote:
> Grant Likely wrote:
> > On 10/15/07, Scott Wood <scottwood@freescale.com> wrote:
> >> For associating a device node with a human readable label, I'd
> >> prefer a "label" property in the device node, rather than doing it
> >> backwards with aliases.
> >
> > Here the corresponding problem; having to scan every device node to
> > make sure you don't assign a number already selected by another node
> > (in the case where one node is assigned a number and another is not).
> >
> Don't Do That(tm).  If you use this mechanism, and an adapter node
> doesn't have a bus number, then it doesn't get to pre-register devices,
> but instead must use i2c_new_device.

Even that doesn't work.  For example if a PCI device is probed which
registers an i2c bus; there needs to be a mechanism for the i2c layer
to know that an id is already spoken for, so once again there needs to
be a mechanism to map easily from id to device (or lack thereof).

> >>> As per your point below; if all the i2c devices are children of
> >>> the adapter, then yes you are right that the bus number doesn't
> >>> matter to the user.  But it *does* matter for things like serial
> >>> and ethernet ports.
> >> And a label property would be great for that. :-)
> >
> > Not really; if the user needs to renumber devices; you don't want him
> >  fiddling around in the hardware description.
>
> Why would the user renumber devices?

Where user == system integrator or firmware engineer.  ie. boards with
no-populate options which affect the numbering; changes to match the
silkscreening on the chassis when a common board is used by multiple
systems.  It's a conceivable scenario.  (Again; this is more relevant
to eth and serial devices than i2c).

>
> > Just like the chosen node; an aliases describes logical constructs,
> > not physical ones.  I don't think this is any different from the
> > linux,stdout-path property in chosen.
>
> Well, it's somewhat different in that stdout describes a usage of the
> device, not the identity.
>
> Still, I don't like linux,stdout-path. :-)
> At the very least it should be a phandle.

I'm cool with it being a phandle.  (insert obvious objection someone
will make about that not being OF compatible)  :-)

Perhaps aliases should look like (which can be generated from the OF
path form when the device tree if flattened):
aliases {
     linux,eth0 = <phandle1>;
     linux,eth1 = <phandle2>;
}

Cheers,
g.

-- 
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.
grant.likely@secretlab.ca
(403) 399-0195