On Sat, 14 Apr 2012 18:54:46 +0200, Daniel Vetter <daniel at ffwll.ch> wrote: > To contrast with the state before these two patches: > - No longer splattered all over (some of it was deep down in the encoder > code, e.g. sdvo). > - No more stupid confusion about interlaced timings. Good things for sure. My biggest concern is that we have conflicting requirements upon the mode as setup in crtcinfo that are not being detected. The task of mode-valid is to find a compromise mode that works for all attached connectors, and more importantly make sure we reject combinations of connectors that would result in no compatible modes (above and beyond the combination of encoders that are illegal). And most of that concern is due to the fact that the modesetting code seems to have evolved ad-hoc with very few sanity checks and no overarching design. -Chris -- Chris Wilson, Intel Open Source Technology Centre
