On Thu, 01 Feb 2024, Maxime Ripard <mripard@xxxxxxxxxx> wrote: > We've discussed that on IRC today. I'm not sure there was a conclusion > other than "well this doesn't seem right". I think we should at least > provide different EDIDs depending on the connector type indeed, but > there was also a few discussions that arose: > > - Is it useful to have embedded EDIDs in the kernel at all, and could > we just get rid of it? > > - Should we expose those EDIDs to userspace, and what happens to the > compositor when we do? > > - The current way to generate those EDIDs isn't... optimal? Should we > get rid of that as well? > > Anyway, all of those issues have been here for a while so I don't really > expect this series to fix that. IMO the direction should be towards deprecating and removing the builtin firmware EDIDs from the kernel instead of adding more or expanding on them. They were only ever meant to be the immediate aid to get something on screen so the user could provide a proper EDID via userspace. BR, Jani. -- Jani Nikula, Intel
