We investigated implications of recently released 'Joint Light-curve Analysis' (JLA) supernova Ia (SNe Ia) data for dark energy models with time varying equation of state of dark energy, usually expressed as w(z) in terms of variation with corresponding redshift z. From a comprehensive analysis of the JLA data, we obtain the observational constraints on the different functional forms of w(z), corresponding to different varying dark energy models often considered in literature, viz. CPL, JBP, BA and Logarithmic models. The constraints are expressed in terms of parameters (w a , w b ) appearing in the chosen functional form for w(z), corresponding to each of the above mentioned models. Realising dark energy with varying equation of state in terms of a homogeneous scalar field φ, with its dynamics driven by a k−essence Lagrangian L = V F (X) with a constant potential V and a dynamical term F (X) with X = (1/2)∇ µ φ∇ µ φ we reconstructed form of the function F (X). This reconstruction has been performed for different varying dark energy models at best-fit values of parameters (w a , w b ) obtained from analysis of JLA data. In the context of k−essence model, we also investigate the variation of adiabatic sound speed squared, c 2 s (z), and obtained the domains in (w a , w b ) parameter space corresponding to the physical bound c 2 s > 0 implying stability of density perturbations.1