On Tue, Feb 23, 2016 at 10:36:36AM +0000, Lionel Landwerlin wrote: > On 23/02/16 00:38, Matt Roper wrote: > >On Mon, Feb 22, 2016 at 02:18:08PM +0000, Lionel Landwerlin wrote: > >>Implement Daniel Stone's recommendation to not read registers to infer > >>the hardware's state. > >> > >>Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@xxxxxxxxx> > >Do we need to ensure that software and hardware state are synchronized > >at startup? A boot firmware might have set it to something different > >before our driver starts up; if we use 'fastboot' then we might not do > >any modesets and might wind up with 0 (8BIT) in our state variable, but > >something else actually programmed into the hardware. > > > > > >Matt > > Thanks Matt, > > It makes sense know, I couldn't understand why this would ever be at > something different that 8bit mode... > I guess the value should be read from the intel_color_init() > function upon startup. We have a hardware state readout where we reconstruct the hardware state for various things like plane state and such; you could add color management readout to that. Alternatively, we could just force known-good values into the hardware at startup as we do (or will very soon) for stuff we don't trust the BIOS to get right like watermarks. Matt > > - > Lionel -- Matt Roper Graphics Software Engineer IoTG Platform Enabling & Development Intel Corporation (916) 356-2795 _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel
