On 1/22/08, Jon Loeliger <[EMAIL PROTECTED]> wrote:
> Grant Likely wrote:
> > On 1/22/08, Jon Loeliger <[EMAIL PROTECTED]> wrote:
> >> Based on the Kurobox DTS files.
> >>
> >> Signed-off-by: Andy Wilcox <[EMAIL PROTECTED]>
> >> Signed-off-by: Jon Loeliger <[EMAIL PROTECTED]>
> >
> > Comments below
> >
> >> +
> >> +/ {
> >> + model = "StorCenter";
> >> + compatible = "storcenter";
> >
> > Be specific! "iomega,storcenter". Even better if you put in the model
> > number.
>
> As I mentioned vefore, there is no further model number.
> That _is_ the model name.
/me has a short memory.
> >> + compatible = "fsl,mpc8241", "mpc10x";
> >
> > fsl,mpc8241-immr would be better; this node describes the internally
> > memory mapped registers; not the entire soc.
>
> Uh, whatever? :-) 'Cuz how many other DTS files say that?
True, we haven't been doing this until recently; but we've also been
stumbling about over the last 2 years figuring out how best to use
this fancy device tree thing in embedded systems effectively. Best
practices are starting to emerge (wadda you know, the open firmware
recommended practices actually have good thought behind them) and
being specific with the compatible property is one aspect. I'm moving
the 5200 over to specifying -immr and some of the other soc ports are
doing so also. Chat with Scott Wood.
> >> + linux,stdout-path = "/soc/[EMAIL PROTECTED]";
> >
> > /[EMAIL PROTECTED]/ perhaps?
>
> Not really necessary to specify the unit number.
Okay, I wasn't sure if we could get away with that with the fdt.
Cheers,
g.
--
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-dev
