From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ville =?iso-8859-1?Q?Syrj=E4l=E4?= Subject: Re: [PATCH 0/9] drm/i915: Plane rotation support Date: Mon, 30 Sep 2013 21:36:59 +0300 Message-ID: <20130930183659.GH9395@intel.com> References: <1380552272-32585-1-git-send-email-ville.syrjala@linux.intel.com> <20130930170949.GE9395@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Return-path: Content-Disposition: inline In-Reply-To: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dri-devel-bounces+sf-dri-devel=m.gmane.org@lists.freedesktop.org Errors-To: dri-devel-bounces+sf-dri-devel=m.gmane.org@lists.freedesktop.org To: Daniel Vetter Cc: Intel Graphics Development , "dri-devel@lists.freedesktop.org" List-Id: intel-gfx@lists.freedesktop.org On Mon, Sep 30, 2013 at 08:21:33PM +0200, Daniel Vetter wrote: > On Mon, Sep 30, 2013 at 7:46 PM, Rob Clark wrote: > > I guess the main thing I care about is that we don't advertise things > > to userspace that we can't actually do. I'm not sure what other hw > > out there supports rotation in hw in some form or another, but it > > might be a good time to hear from 'em about whether these property > > values work for them or not. > = > Hm, I've thought the plan was to let userspace figure that out with a > dry-run flag, and if a certain configuration doesn't work it needs to > fall back to rendering-based compositioning for the given surface. I > don't think there's really much more we can do for fully generic > compositors. Yeah, if no one comes up with anything better, the trial and error approach is the plan in my mind. But in case the hardware never ever supports certain property/value I think we shouldn't expose it. Like 90/270 degree rotation in case of intel hardware. > Tha might leave strange hw in the dust where planes aren't symmetric > in capabilities and hence a simple linear walk over surfaces/planes, > ordered by bw-savings or so, yields extermely bad surface->plane > assignements. But my impression is that hw is moving to unified > stacks of planes so I hope we can punt on solving this in a generic > way (and resort to quick platform hacks in userspace where it's really > needed to hit e.g. video playback power targets). Right. If someone has really specific needs I think they can just go and write their own compositor (or plugin if the compositor in question has such things). -- = Ville Syrj=E4l=E4 Intel OTC