On Tue, Jul 17, 2018 at 04:00:53PM -0500, Bjorn Helgaas wrote: > On Tue, Jul 17, 2018 at 01:39:00PM -0700, Matthew Wilcox wrote: > > The Nehalem system makes an interesting testcase because it exposes some > > registers in fake PCIe devices that aren't behind the root ports. eg: > > > > ff:06.3 Host bridge: Intel Corporation Xeon 5500/Core i7 Integrated Memory Controller Channel 2 Thermal Control Registers (rev 04) > > I think these appear as conventional PCI devices; at least the ones > I've seen, e.g., [1], don't have a PCIe capability, so I think it > makes sense that they're not behind a root port. > > [1] https://bugzilla5.redhat.com/attachment.cgi?id=433169 Oh, I don't think we're doing anything wrong with how we're displaying them or what we're doing with what the system presents to us. My only point was that this is a good test-case for code which assumes that all PCI devices lie under a PCIe root port. At one point during development, my code reported that device up there as /06.3 Host bridge: Intel Corporation Xeon 5500/Core i7 Integrated Memory Controller Channel 2 Thermal Control Registers (rev 04) but since I had that system available to test with, I spotted that problem and made it present that device as ff:06.3 (both with and without -P).
