Using matrix inputs with level 2 S-functions

I'm attempting to pass a matrix into a Matlab level 2 S-function block in simulink. My goal is to allow an inherited input port size, so I have in my setup block:

    block.SetPreCompInpPortInfoToDynamic;
    for i=1:3
      block.InputPort(i).DatatypeID  = 0;  % double
      block.InputPort(i).Complexity  = 'Real';
      block.InputPort(i).SamplingMode = 'Sample';
      block.InputPort(i).SampleTime = [-1,0.0];
      block.InputPort(i).Dimensions=-1;
      block.InputPort(i).DirectFeedthrough = true;
  end  
  block.RegBlockMethod('SetInputPortDimensions',@SetInpPortDims);

which simply does

 function SetInpPortDims(block, idx, di)
     block.InputPort(idx).Dimensions = di;

The toy model I build to illustrate the problem simply takes a constant (1), pads it to be a 100 element vector, and then uses a vector concatenate block to create a 100x3 matrix. When I pass this into my S-function block I get the error:

Error in port widths or dimensions. Output port 1 of 'untitled/Vector Concatenate' is a [100x3] matrix. Error in port widths or dimensions. Invalid dimension has been specified for input port 1 of 'untitled/Convert ECEF to NED'.

If I set the input size of the S function to [100 3] on the correct input ports this works fine. The problem seems to be that the function SetInpPortDims (registered to do 'SetInputPortDimensions') is only called once for my third input (which is not a matrix). It is never called for the two matrix inputs.

The documentation says that when Simulink cannot fully determine the size of the input port, this function isn't called. How can I help if fully define the input signal size? I've tried adding a signal specification block with fixed size of 100x3 which only changes my error to

Error in port widths or dimensions. Output port 1 of 'untitled/Signal Specification' is a [100x3] matrix. Error in port widths or dimensions. Invalid dimension has been specified for input port 1 of 'untitled/Convert ECEF to NED'.

Thanks for your support