On Wed, Nov 09, 2016 at 10:12:59AM +1100, Benjamin Herrenschmidt wrote:

On Tue, 2016-11-08 at 11:49 +0000, Mark Rutland wrote:

quoted

I believe that we could theoretically have multiple independent LPC/ISA
busses, as is possible with PCI on !x86 systems. If the current ISA code
assumes a singleton bus, I think that's something that needs to be fixed
up more generically.

I don't see why we should need any architecture-specific code here. Why
can we not fix up the ISA bus code in drivers/of/address.c such that it
handles multiple ISA bus instances, and translates all sub-device
addresses relative to the specific bus instance?

What in that code prevents that today ?

It appears I was mistaken w.r.t. the singleton comment. We can already
translate MMIO->MMIO addresses per-instance (in the presence of a ranges
property).

The big change would be to handle !MMIO translations, for which we'd
need a runtime registry of ISA bus instance to find the relevant
accessor ops and instance-specific data.

Thanks,
Mark.