arsenm wrote: I think the current TargetOptions has 3 different use cases:
1. CodeGen machinery options 2. CodeGen influencing options with no corresponding IR control 3. Cache for function attributes used locally to codegen It would help split this up based on those categorizations. The 3rd case is the fast math flag attribute case, which we're probably the closest to resolving https://github.com/llvm/llvm-project/pull/219894 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
