How can I run a MPC simulation on command line using a Custom QP Solver?

I have designed a mpc object (mpc1) for a control scenario (scenario1) using mpcDesigner.

Then, I have exported mpc1 to Workspace and I have generated a MATLAB script to reproduce current controller design (see Figure 1).

Now, I want to use a Custom QP Solver, so I run on the command line the following:

src = which('mpcCustomSolver.txt');
dest = fullfile(pwd,'mpcCustomSolver.m');
copyfile(src,dest,'f');

In this way, the Custom QP Solver to be used is 'quadprog'.

      function [x, status] = mpcCustomSolver(H, f, A, b, x0)
      % mpcCustomSolver allows user to specify a custom quadratic programming
      % (QP) solver to solve the QP problem formulated by MPC controller.  When
      % the "mpcobj.Optimizer.CustomSolver" property is set true, instead of
      % using the built-in QP solver, MPC controller will now use the customer QP
      % solver defined in this function for simulations in MATLAB and Simulink.
      %
      % The MPC QP problem is defined as follows:
      %   Find an optimal solution, x, that minimizes the quadratic objective
      %   function, J = 0.5*x'*H*x + f'*x, subject to linear inequality
      %   constraints, A*x >= b.
      %
      % Inputs (provided by MPC controller at run-time):
      %       H: a n-by-n Hessian matrix, which is symmetric and positive definite.
      %       f: a n-by-1 column vector.
      %       A: a m-by-n matrix of inequality constraint coefficients.
      %       b: a m-by-1 vector of the right-hand side of inequality constraints.
      %      x0: a n-by-1 vector of the initial guess of the optimal solution.
      %
      % Outputs (fed back to MPC controller at run-time):
      %       x: must be a n-by-1 vector of optimal solution. 
      %  status: must be an finite integer of:
      %           positive value: number of iterations used in computation
      %                        0: maximum number of iterations reached
      %                       -1: QP is infeasible
      %                       -2: Failed to find a solution due to other reasons
      % Note that even if solver failed to find an optimal solution, "x" must be
      % returned as a n-by-1 vector (i.e. set it to the initial guess x0)
      %
      % DO NOT CHANGE LINES ABOVE

      % The following code is an example of how to implement the custom QP solver
      % in this function.  It requires Optimization Toolbox to run.

      % Define QUADPROG options and turn off display of optimization results in
      % Command window.
      options = optimoptions('quadprog');
      options.Display = 'none';  
      % By definition, constraints required by "quadprog" solver is defined as
      % A*x <= b.  However, in our MPC QP problem, the constraints are defined as
      % A*x >= b.  Therefore, we need to implement some conversion here:
      A_custom = -A;
      b_custom = -b;
      % Compute the QP's optimal solution.  Note that the default algorithm used
      % by "quadprog" ('interior-point-convex') ignores x0.  "x0" is used here as
      % an input argument for illustration only.
      H = (H+H')/2; % ensure Hessian is symmetric
      [x, ~, Flag, Output] = quadprog(H, f, A_custom, b_custom, [], [], [], [], x0, options);
      % Converts the "flag" output to "status" required by the MPC controller.
      switch Flag
          case 1
              status = Output.iterations;
          case 0
              status = 0;
          case -2
              status = -1;
          otherwise
              status = -2;
      end
      % Always return a non-empty x of the correct size.  When the solver fails,
      % one convenient solution is to set x to the initial guess.
      if status <= 0
          x = x0;
      end

As a custom QP solver is going to be used, I set

mpc1.Optimizer.CustomSolver=true

Finally, to reproduce the simulation with the Custom QP Solver I run the following in the command line:

%%specify simulation options
options = mpcsimopt();
options.MVSignal = mpc1_MVSignal;
options.RefLookAhead = 'on';
options.MDLookAhead = 'off';
options.Constraints = 'on';
options.OpenLoop = 'off';
%%run simulation
[y,t,u]=sim(mpc1, 151, mpc1_RefSignal, mpc1_MDSignal, options);

But I always get this error:

The number of rows and columns in H must equal the number of elements of f.

Error in quadprog (line 254)
            error(message('optim:quadprog:MismatchObjCoefSize'));

Error in mpcCustomSolver (line 48)
[x, ~, Flag, Output] = quadprog(H, f, A_custom, b_custom, [], [], [], [], x0, options);

Error in mpc_solveQP (line 39)

Error in mpc_simPostprocess (line 12)

Error in sim (line 94)

As it is said in function [x, status] = mpcCustomSolver(H, f, A, b, x0) H,f are provided by MPC controller at run-time, so I can't guess what am I doing wrong.

% Inputs (provided by MPC controller at run-time):
      %       H: a n-by-n Hessian matrix, which is symmetric and positive definite.
      %       f: a n-by-1 column vector.
      %       A: a m-by-n matrix of inequality constraint coefficients.
      %       b: a m-by-1 vector of the right-hand side of inequality constraints.
      %      x0: a n-by-1 vector of the initial guess of the optimal solution.

Taking all these into account, every help is wellcome in order to solve the following questions :

1) Is the approach explained above correct? I mean, Can I follow this strategy in order to run the simulation using a Custom QP Solver?

2) In case that the approach explained above is not correct, How can I run the simulation on command line using a Custom QP Solver? Is there an alternative?

2) In case that the approach explained above is correct, How can I solve the error in quadprog ?

The number of rows and columns in H must equal the number of elements of f.

    Error in quadprog (line 254)
                error(message('optim:quadprog:MismatchObjCoefSize'));