================ @@ -48,6 +48,13 @@ bool hasValidBranchWeightMD(const Instruction &I); /// Nullptr otherwise. MDNode *getBranchWeightMDNode(const Instruction &I); +/// Get the branching metadata information +/// +/// \param I The Instruction to get the weights from. +/// \returns A pointer to I's branch weights metadata node, if it exists. +/// Nullptr otherwise. +MDNode *getDxBranchHint(const Instruction &I); ---------------- ilovepi wrote:
The description seems to conflict with what the implementation does. It doesn't return the branch weights at all, it returns a `MD_dxil_controlflow_hints` node. That aside, I don't think this logic belongs in ProfDataUtils. You're not changing or accessing profile data, you're interacting with a new metadata type. If there is going to be a generic IR construct for control flow hints, and they're going to be part of MD_prof, then I'd feel differently. ProfDataUtils was initially created to consolidate how we manipulate MD_prof metadata, and prevent (to a certain extent) people from accessing/manipulating that metadata in inconsistent ways. It also allows us to support optional metadata in the MD_prof node (e.g. https://llvm.org/docs/BranchWeightMetadata.html#branchinst), for things like MisExpect. That all could be documented better though. https://github.com/llvm/llvm-project/pull/116331 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
