I think your patch makes sense, and you can add my ack.
I do have the feeling that the reason we need the quirk is because we
aren't handling those PNP device resources quite correctly in the first place. But I don't have any good ideas about what's wrong yet.
My first thought was that those regions might be marked as ACPI_PRODUCER resources (i.e., the PNP device forwards transactions in that region to a downstream device). In that case, the PNP device
is really a bridge, and it shouldn't reserve the region (or at least
should not mark it as busy). But I think we currently ignore ACPI_PRODUCER resources completely, so I don't see how that could cause this problem.