[Mlir-commits] [mlir] [MLIR][GPU] Add xevm-attach-target transform pass. (PR #147372)

[Adam Siemieniuk]

================
@@ -258,4 +258,38 @@ def GpuSPIRVAttachTarget: Pass<"spirv-attach-target", ""> {
   ];
 }
 
+def GpuXeVMAttachTarget : Pass<"xevm-attach-target", "mlir::gpu::GPUModuleOp"> {
----------------
adam-smnk wrote:

In this case I don't see any reason why either approach is better and I can't really construct an example that would behave differently.
I think this approach is equally fine as is.

However, there's value in consistency across the code base. Since this pass clearly mirrors the other two, there's no reason to diverge. Especially when the added complexity is like 2 extra lines.

https://github.com/llvm/llvm-project/pull/147372