Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754507Ab2HTQYv (ORCPT ); Mon, 20 Aug 2012 12:24:51 -0400 Received: from youngberry.canonical.com ([91.189.89.112]:54288 "EHLO youngberry.canonical.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753789Ab2HTQYt (ORCPT ); Mon, 20 Aug 2012 12:24:49 -0400 Date: Mon, 20 Aug 2012 11:24:44 -0500 From: Seth Forshee To: Matthew Garrett Cc: Dave Airlie , Daniel Vetter , David Airlie , dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, Andreas Heider Subject: Re: [RFC PATCH 7/7] drm/pci: Defer initialization of secondary graphics devices until switcheroo is ready Message-ID: <20120820162444.GG11013@thinkpad-t410> Mail-Followup-To: Matthew Garrett , Dave Airlie , Daniel Vetter , David Airlie , dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, Andreas Heider References: <20120806045150.GA23652@thinkpad-t410> <1345476664-22066-1-git-send-email-seth.forshee@canonical.com> <1345476664-22066-8-git-send-email-seth.forshee@canonical.com> <20120820153640.GA29614@srcf.ucam.org> <20120820155633.GF11013@thinkpad-t410> <20120820155740.GA30611@srcf.ucam.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20120820155740.GA30611@srcf.ucam.org> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1478 Lines: 31 On Mon, Aug 20, 2012 at 04:57:41PM +0100, Matthew Garrett wrote: > On Mon, Aug 20, 2012 at 10:56:33AM -0500, Seth Forshee wrote: > > On Mon, Aug 20, 2012 at 04:36:40PM +0100, Matthew Garrett wrote: > > > Won't this break the multiple cards with independent outputs case? > > > > Yes, if they don't have a switcheroo handler. I only have experience > > with one such machine, which had optimus graphics. My recollection is > > that it did have a switcheroo handler, which was only capable of > > controlling power to the discrete card. > > So if I have a desktop machine and install two graphics cards? Yeah, that would likely be broken. I'm not sure how we support both of these cases without doing something more like what I originally proposed, i.e. registering the LVDS connector even if it doesn't look like a panel is attached. I still honestly favor that approach, although it does come with its own set of challenges. The only other option I can come up with is to reprobe LVDS after switcheroo and add the connector at that time. I haven't investigated this option in detail, but at first glance it looks like there are at least some places where DRM isn't prepared to cope with adding connectors after initialization. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/