[Mlir-commits] [mlir] [MLIR][GPU-LLVM] Add in-pass signature update option for opencl kernels (PR #105664)

[Petr Kurapov]

================
@@ -325,6 +369,20 @@ struct GPUToLLVMSPVConversionPass final
     LLVMTypeConverter converter(context, options);
     LLVMConversionTarget target(*context);
 
+    if (forceOpenclAddressSpaces) {
+      MemorySpaceToOpenCLMemorySpaceConverter converter;
+      AttrTypeReplacer replacer;
+      replacer.addReplacement([&converter](BaseMemRefType origType)
+                                  -> std::optional<BaseMemRefType> {
+        return converter.convertType<BaseMemRefType>(origType);
+      });
+
+      replacer.recursivelyReplaceElementsIn(getOperation(),
+                                            /*replaceAttrs=*/true,
+                                            /*replaceLocs=*/false,
+                                            /*replaceTypes=*/true);
+    }
+
----------------
kurapov-peter wrote:

I took a look into this once more. It seems like if I were to reuse `LLVMTypeConverter` I'd need to rewrite the `convertFunctionSignature` to specialize to the case. That drags out some implementation code as well as changes to the converter. This looks worse to me. If you have any better idea please let me know :).

https://github.com/llvm/llvm-project/pull/105664