// to unlink an object from the heap, this will shade it.
//
// 2. shade(ptr) prevents a mutator from hiding an object by moving
// the sole pointer to it from its stack into a black object in the
// heap. If it attempts to install the pointer into a black object,
// this will shade it.
//
// 3. Once a goroutine's stack is black, the shade(ptr) becomes
// unnecessary. shade(ptr) prevents hiding an object by moving it from

		}
	} else {
		// Note: can't use rawmem (which avoids zeroing of memory), because then GC can scan uninitialized memory.
		to = mallocgc(tomem, et, true)
		if copymem > 0 && writeBarrier.enabled {
			// Only shade the pointers in old.array since we know the destination slice to
			// only contains nil pointers because it has been cleared during alloc.
			//
			// It's safe to pass a type to this function as an optimization because

// RUN: tf-opt -split-input-file -verify-diagnostics --tf-shape-inference %s | FileCheck %s

module attributes {tf.versions = {producer = 888 : i32}} {
// CHECK-LABEL: testConv2dShapeValidPadding
func.func @testConv2dShapeValidPadding(%arg0: tensor<1x112x80x128xf32>, %arg1: tensor<128x3x3x128xf32>, %arg2: tensor<128xf32>) -> tensor<1x?x?x128xf32> {

  if (pack_axis < 0) {
    pack_axis += rank;
  }

  // Concat out shape.
  for (int i = 0; i < rank; ++i) {
    int64_t dim_size = input_type.getDimSize(i);
    if (i == pack_axis) {
      dim_size *= count;
    }
    concat_out_shape.push_back(dim_size);
  }

  // Pack out shape.
  int j = 0;
  for (int i = 0; i < rank + 1; ++i) {
    if (i == pack_axis) {

  return transposed_type;
}

// Returns shape of a ranked tensor.
// Precondition: output_val's is ranked tensor.
// Returns a truncated shape when `truncate` is set to true.
inline DenseElementsAttr GetShape(Value output_val, bool truncate = false) {
  auto output_shape = output_val.getType().dyn_cast<ShapedType>().getShape();

  SmallVector<int32_t> shape;
  shape.reserve(output_shape.size());

// Extracts shape from XlaArgument as TensorShape. If shape is a xla::Shape,
// that is converted to a TensorShape.
absl::StatusOr<TensorShape> GetTensorShapeFromXlaArgument(
    const XlaArgument& arg) {
  if (absl::holds_alternative<xla::Shape>(arg.shape)) {
    TensorShape arg_shape;
    TF_RETURN_IF_ERROR(
        XLAShapeToTensorShape(std::get<xla::Shape>(arg.shape), &arg_shape));
    return arg_shape;

    // Merge node causes a loop so we remove NextIteration->Merge edge before
    // performing shape inference. But removing those edges also prevents us
    // from inferring output shape for Merge node (we need shapes for all its
    // inputs).
    // For loop invariant resource input's Merge node, we set output resource
    // shape as Enter node's resource shape.
    // TODO(b/129367850): clean this up.

  // CHECK: %[[CUSTOM_AGGREGATOR_2:.*]], {{.*}}, {{.*}}, {{.*}} = "tf.CustomAggregator"(%[[XLA_CALL_MODULE:.*]])

// Given a tensor of `shape` and `type`, as what shape should it be stored on
// the TPU device? This function tranposes or flattens the excessively-padded
// tensors to rank 1, but leaves other tensor shapes alone.
absl::StatusOr<xla::Shape> TpuShapeRepresentation(
    const TensorShape& shape, DataType type, bool use_fast_memory,
    XlaLayoutPreference layout_preference) {
  xla::Shape xla_shape;
  TF_RETURN_IF_ERROR(

  auto c = ops::Add(root.WithOpName("C"), a, b);
  auto d = ops::Neg(root.WithOpName("D"), c);

  a.node()->AddAttr("_index", 0);
  b.node()->AddAttr("_index", 1);

  std::unique_ptr<Graph> graph(new Graph(OpRegistry::Global()));
  TF_CHECK_OK(root.ToGraph(graph.get()));

  std::map<int, InferredShape> arg_shapes;
  arg_shapes[0].shape = TensorShape({2, 3});
  arg_shapes[1].shape = TensorShape({2, 3});