[Mlir-commits] [mlir] [NFC][mlir][bufferization] Remove getMemRefType() helper (PR #199034)

[Matthias Springer]

================
@@ -1006,7 +983,7 @@ FailureOr<BufferLikeType> bufferization::detail::defaultGetBufferType(
     return op->emitError("could not infer memory space");
 
   return cast<BufferLikeType>(
-      getMemRefType(tensorType, options, /*layout=*/{}, *memSpace));
+      options.unknownTypeConverterFn(tensorType, *memSpace, options));
----------------
matthias-springer wrote:

Let's update this function:
```
/// Default unknown type converter: Use a fully dynamic layout map.
BaseMemRefType
defaultUnknownTypeConverter(TensorType tensorType, Attribute memorySpace,
                            const BufferizationOptions &options) {
  return getMemRefTypeWithFullyDynamicLayout(tensorType, memorySpace);
}
```

It should return an unranked memref if the tensor is unranked.

Same for this in `Bufferize.cpp`:
```
      opt.unknownTypeConverterFn = [=](TensorType tensorType,
                                       Attribute memorySpace,
                                       const BufferizationOptions &options) {
```


https://github.com/llvm/llvm-project/pull/199034