Hi,
Following the presentation byt Joakim: https://docs.google.com/presentation/d/1CvKBBZ33ggzyhP2ub8iZ410I_KGrFjHftZLm...
The following does **NOT** represent consensus. I'd like to use those statements to trigger discussion:
- a DTB (say osDTB) is passed from firmware to "downstream" OS; firmware can't use that DTB (read/write) after that event (Grant - separate active discussion thread in boot-architecture mailing list) - that osDTB is distinct from any DTBes that are used by firmware components. All may still be derived from a single repository (Joakim) - osDTB can be the result of applying programmatic fixups by diverse firmware components or by providing separate overlays to be merged later (last firmware component in the boot chain or by the OS itself), or by directly merging overlays (FF) - there are provisions in U-Boot to "sign" pieces of osDTB (Simon) - we need policies on what can be updated on osDTB, who can do it and how to verify them (Grant) - We don't want private keys in device so sign parts of osDTBs - there are no tools in Linux to deal with overlays (Heinrich), so for the moment we need firmware to aggregate any overlays into osDTB (FF)
Cordially,
Francois-Frederic