Verify tile and tensor's rank are the same

Author: amd-eochoalo

mlir/lib/Dialect/AMDGPU/IR/AMDGPUDialect.cpp

@@ -720,10 +720,16 @@ LogicalResult MakeDmaDescriptorOp::verify() {
   }
 
   ArrayRef<int64_t> globalStaticSizes = getGlobalStaticSizes();
-  if (globalStaticSizes.size() != globalStaticStrides.size()) {
+  size_t rank = globalStaticSizes.size();
+  if (rank != globalStaticStrides.size()) {
     return emitOpError("strides and sizes must have same rank.");
   }
 
+  ArrayRef<int64_t> sharedStaticSizes = getSharedStaticSizes();
+  if (rank != sharedStaticSizes.size()) {
+    return emitOpError("tensor must have same rank as tile.");
+  }
+
   return success();
 }
 

mlir/test/Dialect/AMDGPU/invalid.mlir

@@ -385,3 +385,12 @@ func.func @make_dma_descriptor_invalid_size_and_stride_sizes(%base: !amdgpu.tdm_
   func.return
 }
 
+// -----
+
+// CHECK-LABEL: func @make_dma_descriptor_invalid_shared_and_global_rank
+// CHECK-SAME: (%[[BASE:.+]]: !amdgpu.tdm_base<i32>)
+func.func @make_dma_descriptor_invalid_shared_and_global_rank(%base: !amdgpu.tdm_base<i32>) {
+  // expected-error@+1 {{'amdgpu.make_dma_descriptor' op tensor must have same rank as tile.}}
+  amdgpu.make_dma_descriptor %base globalSize [4, 4] globalStride [1, 1] sharedSize [2] : !amdgpu.tdm_base<i32> -> !amdgpu.tdm_descriptor
+  func.return
+}