On Mon, Jul 26, 2010 at 03:20:42PM +0300, Paul Irofti wrote: > On Sun, Jul 25, 2010 at 10:45:04PM +0200, Mark Kettenis wrote: > > > Date: Sat, 24 Jul 2010 22:50:39 +0200 > > > From: Mike Belopuhov <[email protected]> > > > > > > Hi, > > > > > > the following diff fixes a hang on boot when acpivideo(4) tries to > > > evaluate _DOD method for nonexistent video device. > > > > > > Long story short, there are crappy laptops out there (like Samsung Q45) > > > that have modifications with advanced video cards (nVidia) and regular > > > Intel ones. To save costs [or for whatever other reasons] manufacturer > > > uses the same ACPI config for both modifications. > > > > > > In this case there are two acpivideo devices: NVID and GFX0 with ADRs > > > Zero and 0x00020000 respectively, so it seems natural to knock out NVID > > > by checking its ADR. FWIW, Linux also checks for device presence in a > > > similar manner. > > > > I don't think this will work. A fair number of machines have a video > > card at dev 0 function 0, which would translate into _ADR being zero. > > > > Below is a diff that I have had in my tree for a while to fix a > > similar issue on my Dell laptop. It checks whether the bus on which > > the card supposedly sits really exists. > > Didn't I okay this diff during c2k10?
I think I did too but this ties into the pci->dev stuff jordan is working on.
