[flang-commits] [flang] [lld] [libcxx] [clang] [lldb] [mlir] [llvm] [compiler-rt] [builtins] Refactor cpu_model support to reduce #if nesting. NFCI (PR #75635)

[Jon Roelofs via flang-commits]

jroelofs wrote:

@petrhosek I don't have a Fucshia sysroot to build this against, so I think it would help me a lot if you could grab the preprocessed version of the `cpu_model.c` before all these changes and stick it in a github gist... I feel a little bad for dragging this out in-tree.

> These #if __has_include(...) conditionals would evaluate to false on Fuchsia since we don't have those headers, but they have been moved in the refactored version.

In that case, if I'm reading it correctly, all of the `#if defined(__Fuchsia__)` support for both initializing the LSE atomics detection bool, and FMV in `cpu_model.c` are dead code, and we can drop those entries in the new `aarch64.c`.



https://github.com/llvm/llvm-project/pull/75635