From mboxrd@z Thu Jan 1 00:00:00 1970 From: Greg KH Subject: Re: [PATCH] Add intel drm blacklist to intel_opregion_present detect Date: Tue, 24 Aug 2010 14:35:19 -0700 Message-ID: <20100824213519.GD4979@suse.de> References: <4C7426DB020000230001F3A8@novprvlin0050.provo.novell.com> <20100824142219.GB16200@srcf.ucam.org> <20100824145130.GA26232@suse.de> <20100824145803.GB17629@srcf.ucam.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20100824145803.GB17629@srcf.ucam.org> Sender: linux-kernel-owner@vger.kernel.org To: Matthew Garrett Cc: Joey Lee , trenn@suse.de, linux-acpi@vger.kernel.org, linux-kernel@vger.kernel.org, Dennis.Jansen@web.de List-Id: linux-acpi@vger.kernel.org On Tue, Aug 24, 2010 at 03:58:03PM +0100, Matthew Garrett wrote: > On Tue, Aug 24, 2010 at 07:51:30AM -0700, Greg KH wrote: > > On Tue, Aug 24, 2010 at 03:22:19PM +0100, Matthew Garrett wrote: > > > Erm. Sorry about that - I just realised that there's actually a fairly > > > significant problem with it :) Right now you'll perform the register > > > call unconditionally on module load. You need to make sure that there's > > > actually Poulsbo hardware in the system, either by registering as a > > > proper PCI driver and calling acpi_video_register in your probe function > > > (and unregister in your remove function) or walking the PCI device list > > > looking for a psb device. > > > > Ick, don't walk the pci device list, just register a "real" pci device > > and do it that way, it's much simpler and plays nicer with the driver > > model and the rest of the kernel. > > My only concern with that (and it's not a strong one) is that it'll > block the psb driver from binding. Of course, the psb driver should be > doing the acpi handling itself anyway, so... If a system really does have this type of driver on it, it can unbind this driver and then bind itself, from userspace, so I don't see a problem with this. thanks, greg k-h