Il 29/04/2014 11:44, Michael S. Tsirkin ha scritto:
> I disagree. In fact, I think the opposite is true: the three
> functions you mention should take a PCIDevice*, and when this is
> done acpi_memory_plug_cb should be changed to take a DIMMDevice*
> too.
Well this just means we'll need another chipset-independent wrapper
to detect the correct hotplug type?
Yes, we could do that. Calling the wrapper is equivalent to the
"broadcast" concept you mentioned, though of course the implementations
differ.
Paolo
