On Mon, Nov 3, 2014 at 9:25 AM, Thierry Reding <treding@xxxxxxxxxx> wrote: >> The idea is that you'd grab the DPCD field anyway since it's needed all >> over the place. We have a pile of helpers already that take exactly this >> block and decode parts of it. So I think this makes sense - dp aux is fast >> but not entirely free, so caching seems useful. > > If we want to always cache part of the DPCD wouldn't it be better to add > the cache to struct drm_dp_aux instead of having to duplicate this in > every driver? Right now we don't (yet) have helper functions to probe SST DP sinks, so there's not really a good place to put it. But yeah longer-term I hope that some shared SST sink probe functions shows up (especially to handle all the various crazy corner-cases with VGA probing, e.g. the apple vga dongle patch Dave sent out many aeons ago), and then putting that cache into the dp aux struct makes sense. But before that happens I don't think it makes a lot of sense really, since then we'd also need to add code to invalidate that cache. Simpler if drivers just take care of that themselves. -Daniel -- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/dri-devel
