## Known Caveats

*   TF Core:
    *   A longstanding bug in `tf.while_loop`, which caused it to execute
        sequentially, even when `parallel_iterations>1`, has now been fixed.
        However, the increased parallelism may result in increased memory use.
        Users who experience unwanted regressions should reset their
        `while_loop`'s `parallel_iterations` value to 1, which is consistent
        with prior behavior.

          while_op.getInput().getTypes(), while_op.getOutput().getTypes(),
          while_op.getBody().getArgumentTypes());
      return PropagateShapeToRegions(compatible_types,
                                     {&while_op.getCond(), &while_op.getBody()},
                                     max_iterations);
    }
    return PropagateShapeToRegions(while_op.getInput().getTypes(),

                set_op.getItem());
            element_shape_acquired = true;
          } else if (TF::WhileOp while_op =
                         llvm::dyn_cast<TF::WhileOp>(use.getOwner())) {
            // Tensorlist is passed into a while loop, check inside the body
            // function.
            auto inside_uses = while_op.body_function()
                                   .getArgument(use.getOperandNumber())

  // Create the while op.
  auto while_op = builder->create<mhlo::WhileOp>(loc, init_types_with_loop_iv,
                                                 init_values_with_loop_iv);
  auto ivs_count = init_types_with_loop_iv.size();

  {
    OpBuilder::InsertionGuard guard(*builder);

    // Build up the only block in the condition region.
    Region &condition = while_op.getCond();