Re: Bug in devicetree_for_each_node() in xen/arch/arm/bootfdt.c ?

[Ian Campbell <ian.campbell@citrix.com>]

On Tue, 2015-06-23 at 11:32 +0100, Julien Grall wrote:
> 
> On 23/06/2015 11:27, Ian Campbell wrote:
> > On Tue, 2015-06-23 at 11:08 +0100, Julien Grall wrote:
> >
> >>>              reg = <0x0 0x0 0x0 0x40000000 0x0 0x40000000 0x0 0x40000000>;
> >>
> >> Although, what the rest of the node used for? Do we expect to parse it?
> >> I wasn't able to find a suitable bindings in the docs...
> >
> > A reg can encode multiple regions, by just listing them one after the
> > other. It'll be in the generic docs about reg properties I expect,
> > there's nothing special about a reg property in a memory node in this
> > regard.
> >
> > Xen parses these just fine, via the loop and use of device_tree_get_reg
> > in process_memory_node
> 
> I know that a "reg" can encode multiple regions... My question was 
> related to the "rest of the node" and not the rest of the property... i.e
> 
>     region@10000000 {
> 
>                  contiguous-region;
> 
>                  reg = <0x10000000 0x1f800000>;
> 
>                  linux,phandle = <0x2>;
> 
>                  phandle = <0x2>;
> 
>             };
> 
> AFAICT we don't parse it.

Indeed, but I'm pretty sure that whatever they are it doesn't relate to
the "only seeing 2GB out of 3GB issue", that's solely down to the main
bit of the memory node only encoding 2GB AFAICT.

Ian.