Attention is currently required from: Felix Held, Felix Singer, Lean Sheng Tan, Martin L Roth, Nico Huber.
Maximilian Brune has posted comments on this change. ( https://review.coreboot.org/c/coreboot/+/80139?usp=email )
Change subject: util/crossgcc/buildgcc: Compile RISC-V GCC with medany ......................................................................
Patch Set 2:
(2 comments)
Patchset:
PS1:
could you resolve the build error?
I am honestly not sure why there even is a build error. Works fine on my local machine to build the mainboards that jenkins has a problem with. Maybe a rebase will solve it magically.
File util/crossgcc/buildgcc:
https://review.coreboot.org/c/coreboot/+/80139/comment/3b9ea32a_83362d04 : PS1, Line 779: CFLAGS_FOR_TARGET_EXTRA
Does this need to be restricted to GCC, or does something need to be done for CLANG?
Building using clang works fine. There are still a few uncertainties for me. - I am not sure how much we pull/need from libgcc.a anyway. I didn't even think we would link against libgcc in the first place before I hit this issue. - Looking at `arch/riscv/Makefile.mk` it looks like we are depending on GCC for RISC-V, but it still works when compiling for clang. - It seems like we always link against libgcc.a in RISC-V even if we use clang as compiler.
Bottom line there is nothing to do for clang regarding this patchset. But I will try to see if removing libgcc.a as dependency for RISC-V builds makes sense in another patchset. That should also remove the GCC specific bits from `arch/riscv/Makefil.mk`. If that works I will probably revert this patch, but for now I need it in order to build my mainboard.