feca17fa4b
When the Q35 machine type(s) of QEMU are used with libvirt, libvirt tends to place some devices behind PCI bridges. This is then reflected in the "bootorder" fw_cfg file. For example: /pci@i0cf8/pci-bridge@1e/pci-bridge@1/scsi@5/disk@0,0 /pci@i0cf8/pci-bridge@1e/pci-bridge@1/scsi@3/channel@0/disk@0,0 As yet QemuBootOrderLib doesn't support such OFW device paths. Add code that translates a sequence of pci-bridge nodes. In practice libvirt seems to insert two such nodes (*), hence increment EXAMINED_OFW_NODES with the same number. (* Background, paraphrasing Laine Stump's words: When the machine type is Q35, we create a dmi-to-pci bridge coming off of the pcie root controller, and a pci-to-pci bridge coming off of that, then attach most devices to the pci-to-pci bridge. This is done because you can't hotplug into pcie-root, can't (or at least shouldn't) plug a pci-to-pci bridge into pcie-root (so the next one has to be dmi-to-pci-bridge), and can't hotplug into dmi-to-pci-bridge (so you need to have a pci-to-pci bridge).) Contributed-under: TianoCore Contribution Agreement 1.0 Signed-off-by: Laszlo Ersek <lersek@redhat.com> Reviewed-by: Jordan Justen <jordan.l.justen@intel.com> git-svn-id: https://svn.code.sf.net/p/edk2/code/trunk/edk2@17385 6f19259b-4bc3-4df7-8a09-765794883524