[Mlir-commits] [mlir] [mlir][amdgpu] Add lowering for make_dma_descriptor (PR #169955)

[Krzysztof Drewniak]

================
@@ -1341,14 +1345,58 @@ def AMDGPU_MakeDmaDescriptorOp :
     `globalSize` custom<DynamicIndexList>($global_dynamic_sizes, $global_static_sizes)
     `globalStride` custom<DynamicIndexList>($global_dynamic_strides, $global_static_strides)
     `sharedSize` custom<DynamicIndexList>($shared_dynamic_sizes, $shared_static_sizes)
-    ( `padShared` `(` $pad^ `every` $pad_every `)` )?
+    ( `padShared` `(` $pad_amount^ `every` $pad_interval `)` )?
     ( `atomicBarrier` `(` $atomic_barrier_address^ `[` $atomic_barrier_indices `]`
                       `:` type($atomic_barrier_address) `)`)?
     ( `iterate` $global_increment^ `,` $lds_increment `,` $iteration_count )?
     attr-dict `:` qualified(type($base)) `->` type(results)
   }];
 
+  let extraClassDeclaration = [{
+    int getRank() {
+      return getGlobalStaticSizes().size();
+    }
+
+    int getElementTypeWidth() {
----------------
krzysz00 wrote:

1. Match the return type of `getIntOrFloatBitWidth()`
2. getIntOrFloatBitWidth() already has the assert you're implementing here, so this is a one-liner (`getBase().getType().getELementType().getIntOrFloatBitWidth()`

https://github.com/llvm/llvm-project/pull/169955