[Mlir-commits] [mlir] [mlir][gpu] GPUToROCDL: Add C++ argument to populate allowedDialects (PR #157402)

Pablo Antonio Martinez llvmlistbot at llvm.org
Tue Sep 9 00:34:10 PDT 2025


================
@@ -288,9 +287,12 @@ struct GPUShuffleOpLowering : public ConvertOpToLLVMPattern<gpu::ShuffleOp> {
 struct LowerGpuOpsToROCDLOpsPass final
     : public impl::ConvertGpuOpsToROCDLOpsBase<LowerGpuOpsToROCDLOpsPass> {
   LowerGpuOpsToROCDLOpsPass() = default;
-  LowerGpuOpsToROCDLOpsPass(const std::string &chipset, unsigned indexBitwidth,
-                            bool useBarePtrCallConv,
-                            gpu::amd::Runtime runtime) {
+  LowerGpuOpsToROCDLOpsPass(ConvertGpuOpsToROCDLOpsOptions options)
+      : ConvertGpuOpsToROCDLOpsBase(options) {}
+  LowerGpuOpsToROCDLOpsPass(
+      const std::string &chipset, unsigned indexBitwidth,
+      bool useBarePtrCallConv, gpu::amd::Runtime runtime,
+      std::optional<llvm::SmallDenseSet<StringRef>> allowedDialects) {
     if (this->chipset.getNumOccurrences() == 0)
----------------
pabloantoniom wrote:

Absolutely, thanks for the suggestion. By removing this we also match what `LowerGpuOpsToNVVMOpsPass` does so it should be easier for you if you want to do some cleanup later.

https://github.com/llvm/llvm-project/pull/157402


More information about the Mlir-commits mailing list