Jachym Holecek wrote:
>>I suppose worst case I could give foo a new function
>>foo_register_extended_if(...) and the parent drivers that support
> 
> Either this, or memset() foo_hw_if to zero in all drivers before
> filling it, or include an ioctl-like function through which you
> can "tunnel" new methods as they appear.
> 
> I think each of these approaches has precedents in NetBSD code.

Thanks! I wound up using the register_extended() approach--it still
feels a little kludgey but I could do it quickly and it works, and
didn't require my touching any of the drivers that don't use the new
interface, or verifying that they all follow some convention like
using memset. I guess it's ok with me if it's ok with TNF. :)

-Chap