On Mon, Jul 29, 2024 at 11:04:41AM +0300, Mika Westerberg wrote: > On Sun, Jul 28, 2024 at 05:41:09PM +0200, Lukas Wunner wrote: > > Do the DROMs on ICM root switches generally lack PCIe Upstream and > > Downstream Adapter Entries? > > My guess is that they are not populated for ICM host router DROM > entries. These are pretty much Apple stuff and USB4 dropped them > completely in favour of the router operations. I note that Microsoft specifies a "usb4-port-number" for PCIe Downstream Adapters as well as DP and USB ports: https://learn.microsoft.com/en-us/windows-hardware/drivers/pci/dsd-for-pcie-root-ports#mapping-native-protocols-pcie-displayport-tunneled-through-usb4-to-usb4-host-routers Presumably the "usb4-port-number" allows for associating a pci_dev with a tb_port. So that's a third way to do that, on top of DROM entries and the USB4 router operation. What I don't quite understand is whether the "usb4-port-number" is only present on PCIe Adapters of the Host Router or whether it can also exist on PCIe Adapters of Device Routers? I would also like to know whether "usb4-port-number" is set on the machines that Esther's quirk seeks to fix? I found this DSDT of an unknown Lenovo model which does not have "usb4-port-number" set on any PCIe Adapters, only on USB ports: https://gist.github.com/64kramsystem/ab2410f081a4f47d4a205699828ab2f9 I assumed that my solution to this problem would not be viable as we seem to be lacking port numbers for Host Router PCIe Adapters. Those are needed to associate a pci_dev with a tb_port on the Host Router and adjust its "untrusted" and "external_facing" properties. However if we do have the "usb4-port-number" on Host Router PCIe Adapters, then my solution would still be viable and I could look into adding support for the "usb4-port-number" property to it. Thanks, Lukas
