[Mlir-commits] [mlir] [mlir] Fix GPU_Launch op getPrivateAttributions function (PR #178369)

[Durgadoss R]

================
@@ -1043,6 +1043,8 @@ def GPU_LaunchOp : GPU_Op<"launch", [
     ArrayRef<BlockArgument> getPrivateAttributions() {
       // Buffers on the private memory always come after buffers on the workgroup
       // memory.
+      if (getBody().empty())
+        return {};
----------------
durga4github wrote:

I wonder if it makes more sense to properly intialize `end` using `begin` + offset of `getNumPrivateAttributions`, similar to how this is handled in `getWorkgroupAttributions`


https://github.com/llvm/llvm-project/pull/178369