Re: "memory" binding issues

From: Olof Johansson
Date: Tue Sep 17 2013 - 17:33:57 EST


On Tue, Sep 17, 2013 at 2:19 PM, Tomasz Figa <tomasz.figa@xxxxxxxxx> wrote:
> On Tuesday 17 of September 2013 14:15:52 Olof Johansson wrote:
>> On Tue, Sep 17, 2013 at 2:08 PM, Frank Rowand <frowand.list@xxxxxxxxx>
> wrote:
>> > On 9/17/2013 9:43 AM, Olof Johansson wrote:
>> >> On Tue, Sep 17, 2013 at 09:56:39AM +0200, Tomasz Figa wrote:
>> >>> I'm afraid that I must disagree. For consistency I'd rather go with
>> >>> what Ben said. Please see ePAPR chapter 2.2.1.1, which clearly
>> >>> defines how nodes should be named.
>> >>
>> >> 2.2.1.1 is there to point out that unit address _has_ to reflect reg.
>> >>
>> >> 2.2.3 says that unit addresses can be omitted.
>> >
>> > 2.2.3 is talking about path names.
>> >
>> > 2.2.1.1 is talking about node names.
>> >
>> > 2.2.1.1 _does_ require the unit address in the node name, 2.2.3 does
>> > not remove that requirement.
>>
>> Sigh, that's horrible. OF clearly doesn't require it.
>>
>> I guess people prefer to follow ePAPR even though it's broken? That
>> means someone needs to cleanup the current dts files. Any takers?
>
> I don't think it's broken, why do you think so? It's at least consistent.
> Probably not perfect and not complete, but IMHO a reasonable base for
> further work. (Also at least something written down that people can learn
> from and/or refer to.)

So, I stand corrected. It seems that at least one legacy system I'm
looking at always specifies unit address as well. I don't like it but
I'll stop arguing.

Ben: The interesting part is that it does _not_ specify it on /memory
though. Nor, of course, on /cpus or /openprom. So assuming /memory@0
exists will break even on some powerpc platforms.


-Olof
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/