Re: [kvm-unit-tests PATCH 1/2] devicetree: search up tree in dt_get_nr_cells

[Thomas Huth <thuth@redhat.com>]

On 11.05.2016 11:04, Andrew Jones wrote:
> On Wed, May 11, 2016 at 09:02:39AM +0200, Thomas Huth wrote:
>> On 10.05.2016 19:32, Andrew Jones wrote:
>>> Search up the tree until we find #address-cells/#size-cells.
>>> Also only assign outputs if both address and size are found.
>>>
>>> Signed-off-by: Andrew Jones <drjones@redhat.com>
>>> ---
>>>  lib/devicetree.c | 17 +++++++++++++----
>>>  lib/devicetree.h |  3 ++-
>>>  2 files changed, 15 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/lib/devicetree.c b/lib/devicetree.c
>>> index a5c7f7c69ddfd..d3751e2b7e7f9 100644
>>> --- a/lib/devicetree.c
>>> +++ b/lib/devicetree.c
>>> @@ -24,21 +24,30 @@ int dt_get_nr_cells(int fdtnode, u32 *nr_address_cells, u32 *nr_size_cells)
>>>  {
>>>  	const struct fdt_property *prop;
>>>  	u32 *nr_cells;
>>> -	int len;
>>> +	int len, nac, nsc;
>>> +
>>> +	while (1) {
>>> +		prop = fdt_get_property(fdt, fdtnode, "#address-cells", &len);
>>> +		if (prop != NULL || len != -FDT_ERR_NOTFOUND)
>>> +			break;
>>> +		fdtnode = fdt_parent_offset(fdt, fdtnode);
>>> +	}
>>
>> Why do you need this search? ePAPR clearly states:
>>
>> "The #address-cells and #size-cells properties are not inherited from
>> ancestors in the device tree. They shall be explicitly defined."
> 
> You are right.
> 
> My skimming of Documentation/devicetree/booting-without-of.txt was
> sloppy and I missed basically the same quote
> 
> "Note that the parent's parent definitions of #address-cells and
> #size-cells are not inherited so every node with children must
> specify them."
> 
>>
>> So as far as I can see, it should always be enough to look up the
>> properties in the parent of the node, no need for a recursion here?
> 
> So, yes, I should look in the current node, and then the parent, and
> then just assert.

AFAIK the #address-cells and #size-cells properties of the current node
only apply to the _children_ of the current node, but not to itself. I
think you always have to look at the parent to find out the correct
value for the current node.

 Thomas