Matrix is close to singular or badly scaled. Results may be inaccurate solution ??

Hello, i try to solve a linear system, connecting c++ with matlab like this:

mxArray *Am = NULL;

Am = mxCreateDoubleMatrix (matrixSize, matrixSize, mxREAL);

memcpy( ( void* )mxGetPr(Am) , ( void* )Ac[0] , matrixSize * matrixSize * sizeof(double));

mxArray *Bm = NULL;

Bm = mxCreateDoubleMatrix (matrixSize, 1, mxREAL);

memcpy( ( void* )mxGetPr(Bm) , ( void* )Bc , matrixSize * sizeof(double) );

mxArray *Rm = NULL;

Rm = mxCreateDoubleMatrix (matrixSize, 1, mxREAL);

engPutVariable(ep, "Am",Am);

engPutVariable(ep, "Bm",Bm);

engPutVariable(ep, "Rm",Rm);

engEvalString (ep, "Rm = linsolve(Am,Bm);");

And i get the warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 1.665335e-017.

Now this is an issue of singular matrix. However if i give the two arrays by hand in matlab's workspace the linear system is solved with no warnings.

Any ideas? thank you all