Attention is currently required from: Jakub Czapiga, Julius Werner, Maximilian Brune, Philipp Hug, ron minnich.
Nico Huber has posted comments on this change by Nico Huber. ( https://review.coreboot.org/c/coreboot/+/79907?usp=email )
Change subject: [RFC] region: Hide struct region members ......................................................................
Patch Set 8:
(2 comments)
Patchset:
PS8:
One thing though: Can't region-test.c just use the getter/setter functions instead of directly accessing the properties?
I thought that wouldn't work because of the explicit overflows (see comment on top). But looking closer, it might just fit in the limits of region_create(). I'll try it out...
File src/arch/arm/fit_payload.c:
https://review.coreboot.org/c/coreboot/+/79907/comment/f68a0909_00644b5e?usp... : PS8, Line 3: /* FIXME: should use the high-level region api */
I don't think we have an actual maintainer for the FIT payload support. I have only tried to use a FIT payload once (~1 year ago) and at the time it didn't even work (I think). I wouldn't be surprised if it still doesn't work and no one noticed, since no one uses it. If it were up to me I would just remove the FIT payload entirely and just use SELF (maybe generate a SELF from a FIT in build time like we do with everything else). But that is something for another time/patch.
Ok, now I'm surprised. I somehow expected it'd be used with RISCV... what do you use as payload there? I also had the impression that it would be needed for the UPL thing. But even then, adding it as SELF might be reasonable *shrug*
IIRC, it was originally introduced with the canceled ARM server project (Cavium). When I saw it adopted by other architectures, I thought it's used.
I would suggest we ignore the FIT stuff and just move on with this patch (aka marking this as resolved).
It just wouldn't be right to leave the FIXME, I guess. I'll think about it.