adams381 wrote: @andykaylor I've addressed your feedback about the noinline divergence from classic codegen in commit 2ce0ac32fd20.
**Changes made:** - Removed the `hasEmittedBuiltinCall` tracking mechanism - Now ALL non-always_inline functions get `noinline` at `-O0`, matching classic codegen exactly - Added 154 OGCG checks to `builtin-floating-point.c` to verify CIR-produced LLVM matches classic codegen - Updated test expectations in `builtin-fcmp-sse.c` and `avx512f16c-builtins.c` The simplification removes ~17 lines of code and eliminates the special-case logic that was causing the divergence. https://github.com/llvm/llvm-project/pull/175233 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
