On Thu, 2019-06-27 at 10:35 -0600, Logan Gunthorpe wrote: > My worry would be if the firmware depends on any of those PCI resources > for any of it's calls. For example, laptop firmware often has specific > code for screen blanking/dimming when the special buttons are pressed. > If it implements this by communicating with a PCI device then the kernel > will break things by reassigning all the addresses. > > However, having a kernel parameter to ignore the firmware choices might > be a good way for us to start testing whether this is a problem or not > on some systems As I consolidate that accross archs I can add such a parameter... I haven't quite folded x86 in yet, but I'm hoping I'll be able to do so soon. I plan to move some of those x86 specific kernel parameters into generic code while doing so. I can add this one. Cheers, Ben.
