Forgive me if I’m repeating already-reported issues; I couldn’t find them.
I’d like to use Clang/LLVM (and eventually Swift) to write code for small Cortex M0 & M4 MCUs. There are a handful of efforts on this, one being LLVM-embedded-toolchain-for-Arm. Although the clang/llvm installed on my M1 Mac claim to support cortex-m0 and cortex-m4, I couldn’t figure out how to invoke them (and of course I need headers and a runtime), so I figured installing their toolchain would help me figure things out.
Unfortunately, they don’t support building on macOS, and it fails to build on macOS. I was able to patch llvm-project/runtimes/CMakeLists.txt by adding
set(APPLE False)
near the top (thanks to this PR).
They consider this a hack (as do I), and I wonder if we need to invoke the build differently, or if there needs to be an upstream change to improve this.