We don’t even know what your problem is.
And how were you using CVX to solve a non-convex problem? Via some ,kind of SCP?
And you didn’t show the CVX output, so we don’t know whether Mosek was provided the dual, in which case Mosek’s dual infeasibility determination would correspond to primal infeasibility of the problem you provided to CVX.
Also, unless you can show the solver reached an erroneous conclusion, I would not characterize it as having failed. It may not be the result you were hoping for, but that doesn’t mean the solver failed.