Re: Compatible string best practices

[Mark Rutland <mark.rutland-5wv7dgnIgG8@public.gmane.org>]

On Fri, May 23, 2014 at 04:03:06PM +0100, Sören Brinkmann wrote:
> Hi,

Hi,

> I have a question regarding how a compatible string should be created.
> The simple case is obvious:
>  <IP-vendor>,<IP-type/name>-<IP-revision>
> 
> But what is the recommended string for a SOC specific implementation of
> that IP? Let's say for
>  <SOC-vendor>,<SOC-name>-<SOC-revision>
> how would I assemble a compatible string to identify the SOC-specific
> implementation of IP?
> I think my biggest confusion is which vendor string to use in that case.
> But in general I'm curious whether some best practice exists for these
> cases.
> 
> I think what I've seen most is:
>  <SOC-vendor>,<SOC-name>-<IP-type>
> 
> But one could probably argue whether to rather use IP-vendor, appending
> all the revision strings, ...

I would expect that the vendor prefix would be the last entity along the
chain which altered the IP in some way. Say foo sells device to bar, who
do nothing other than sell it to baz who do some hacks to integrate it.
For that I'd expect something like:

compatible = "baz,soc-device", "foo,device";

I'm not sure with revisions (they're not always that useful), but I'd
expect that to be kept next to the element it represent the revision
for:

"baz,soc-vX-device-vY", "bar,soc-device-vY", "foo,device-vY",
"foo,device-vY";

Cheers,
Mark.
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html