On Wed, 23 Mar 2022, Simon Ser contact@emersion.fr wrote:
On Wednesday, March 23rd, 2022 at 13:02, Jani Nikula jani.nikula@intel.com wrote:
Simon and Daniel also tell me on IRC we can't just modify the base block extension count to match HF-EEODB to dodge the problem, because the EDID gets exposed to userspace.
I'm not familiar how the EDID blob gets exposed to user-space. If the EDID data gets copied when creating the blob, and the blob is created before the kernel mutates the EDID to accomodate for HF-EEODB, then this proposal might still be workable.
You'd still end up with tracking separate copies of the EDID, which is not necessarily easier. There are almost 70 calls to update the connector EDID property that gets exposed to userspace, and the call sites would need to know which copy to pass.
BR, Jani.