On 07/22/2015 01:56 AM, Lorenzo Pieralisi wrote:

On Tue, Jul 21, 2015 at 11:07:57AM +0100, Marc Zyngier wrote:

quoted

struct device_node is very much DT specific, and the original authors
of the irqdomain subsystem recognized that tie, and went as far as
mentionning that this could be replaced by some "void *token",
should another firmware infrastructure be using it.

Yes, that's similar to the problem Rafael solved with fwnode_handle,
I do not know if we can extend the fwnode_handle to manage these
generic tokens too, but the approach you have taken seem the right
one to me (and you are doing this for components that are not really
attached to struct device so I am not sure the fwnode_handle approach
can be shoehorned to solve it).

If I understand correctly, fwnode_handle is not better , since
we need to update every caller of irqdomain functions as it's a new
type of pointer, that's will be the big change, void * is just fine.

Thanks
Hanjun