On Mon, Mar 13, 2017 at 10:26:12AM +0800, Shawn Lin wrote: > I just thought we should fix the domain number here by adding > "linux,pci-domain = <0>" for rk3399.dtsi, which seems more wise > to me now. Does it make sense to you? I think that's fine (as noted in response to your patch). That doesn't really prevent this from being a core PCI domain bug though... BTW, I've been using this patch set to do some repeated remove/probe testing, and aside from the domain renumbering question and a small memory leak noticed by kmemleak (an existing bug; sending a patch shortly), this has been working well for me. Brian
