Attention is currently required from: Felix Held, Julius Werner, Lean Sheng Tan, Matt DeVillier, Maximilian Brune.
Benjamin Doron has posted comments on this change by Benjamin Doron. ( https://review.coreboot.org/c/coreboot/+/84796?usp=email )
Change subject: lib/{fit,fit_payload}.c: Enhance support for FIT images ......................................................................
Patch Set 6:
(1 comment)
File src/lib/fit_payload.c:
https://review.coreboot.org/c/coreboot/+/84796/comment/b3f31454_143c2f37?usp... : PS6, Line 131: continue;
Shouldn't you return a hard error that makes the parent function fail in these cases? That's what we […]
Secondary images aren't necessarily crucial for the boot, and there are other reasons why we might fail (images don't fit in memory, and theoretically a need to perform relocation, ours will never need that). For example, one of ours is network-fv, which contains a network stack. If it can't be unpacked, I'd still want the boot to continue. Even if if we're missing an image that's integrated more tightly into the codebase, such as sec-fv (contains secure boot and TPM), I'd still want to try booting.
This is because when I started with coreboot development, I didn't have the luxury of a serial port, so I developed an ideology of aiming to survive through the boot even if I was hitting errors. This was the only way I'd be able to extract the log from the memory and get some debug value from the boot.
I'll add a comment in the code about this if you're fine with the theory.