addresses feedback on API use

Author: Jianhui-Li

mlir/lib/Dialect/XeGPU/Transforms/XeGPUBlocking.cpp

@@ -302,9 +302,9 @@ void XeGPUBlockingPass::runOnOperation() {
       Attribute encoding = tdescTy.getEncoding();
       // If the encoding is a ScatterTensorDescAttr, we need to
       // potentially adjust the chunk size based on the inst_data.
-      if (encoding && mlir::isa<xegpu::ScatterTensorDescAttr>(encoding)) {
-        auto scatterAttr =
-            mlir::dyn_cast<xegpu::ScatterTensorDescAttr>(encoding);
+      if (tdescTy.isScattered()) {
+        auto scatterAttr = tdescTy.getEncodingAsScatterTensorDescAttr();
+        //    mlir::dyn_cast<xegpu::ScatterTensorDescAttr>(encoding);
         int64_t chunkSize = scatterAttr.getChunkSize().getInt();
 
         if (chunkSize > 1) {

mlir/test/lib/Dialect/XeGPU/TestXeGPUTransforms.cpp

@@ -102,14 +102,12 @@ struct TestXeGPUUnrollingPatterns
           // attribute
           if (auto tdescTy = dyn_cast<xegpu::TensorDescType>(type)) {
             Attribute encoding = tdescTy.getEncoding();
-            auto layout = llvm::dyn_cast_if_present<xegpu::LayoutAttr>(
-                tdescTy.getLayout());
+            auto layout = tdescTy.getLayoutAttr();
 
             // If the encoding is a ScatterTensorDescAttr, we need to
             // potentially adjust the chunk size based on the inst_data.
-            if (encoding && mlir::isa<xegpu::ScatterTensorDescAttr>(encoding)) {
-              auto scatterAttr =
-                  mlir::dyn_cast<xegpu::ScatterTensorDescAttr>(encoding);
+            if (tdescTy.isScattered()) {
+              auto scatterAttr = tdescTy.getEncodingAsScatterTensorDescAttr();
               int64_t chunkSize = scatterAttr.getChunkSize().getInt();
 
               if (chunkSize > 1) {