================ ---------------- NewSigma wrote:
Hi @ChuanqiXu9, I noticed fixing the failing test breaks the behavior introduced in #49264. Specifically, there are two inconsistencies: 1. The O0 pipeline schedules AlwaysInliner before CoroEarly, while others schedule CoroEarly before AlwaysInliner. 2. Switch ABI requires the frontend to set the `presplitcoroutine` attribute, while other ABIs infer the attribute during CoroEarly. (I think they would suffer from bug #49264, though) Is there a specific reason we do not promote CoroEarly ahead of AlwaysInliner? I'd appreciate any context on the current ordering. https://github.com/llvm/llvm-project/pull/169866 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
