On Thu, Mar 29, 2012 at 11:19:59PM +0100, Chris Wilson wrote: > On Thu, 29 Mar 2012 18:30:20 -0300, Paulo Zanoni <przanoni@xxxxxxxxx> wrote: > > From: Paulo Zanoni <paulo.r.zanoni@xxxxxxxxx> > > > > In the Kernel side, these are crtc properties (since in the hardware, > > underscan use the panel fitters, which are attached to the pipes). > > Ideally we should make these as crtc properties too, but since xrandr > > doesn't have support for them, we expose the atoms as output > > properties. This is not the best solution, but making the kernel > > underscan properties be output properties instead of crtc properties > > would be even more wrong. > > However, we have set the precedent with panel fitting already and this is > just a variation on that theme? My wish would be to be able describe the > dst box of the framebuffer on the crtc as part of the mode. I'm assuming you mean that you'd like to have the overscan borders be part of the mode structure itself. Agreed. However there's another problem with the current mode setting API. This is the full display pipeline as I see it: fb area [w0,h0] -> [x1,y1,w1,h1] plane transform [x2,y2,w2,h2] -> CRTC area [w3,h3] -> [0,0,w3,h3] crtc transform [x4,y4,w4,h4] -> display mode active area [hdisp,vdisp] Rectangle [x1,y1,w1,h1] is within the fb area [w0,h0] Rectangle [x2,y2,w2,h2] is within the CRTC area [w3,h3] Rectangle [x4,y4,w4,h4] is within the display mode active area [hdisp,vdisp] Currently we're missing the CRTC area and the overscan borders. This means you have to add some policy into the kernel on how to choose the display mode, and how to program panel fitters and whatnot. Sure you can add properties to influence those policy decisions, but what you end up with is still quite limited. If the API would allow you to specify both the borders, and the CRTC area size, you could do away with all mode selection policy, and panel fitter and border properties, and you would end up with a more powerful API. -- Ville Syrjälä Intel OTC _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/dri-devel
