In this work, a diffusive eco-epidemiological model where the predator's population consume their species to survive. The proposed model is studied from both points of view, theoretical and numerical. Firstly, we deal with the behaviour of the constant positive steady state then nonconstant positive steady state. Sufficient conditions for local asymptotic stability and global asymptotic stability for a constant positive steady state are derived by linearization and Lyapunov function technique. Prior estimates of the positive steady state given, conditions are obtained for the non-existence of non-constant positive solution, by Cauchy and Poincaré inequality. The existence of non-constant positive steady states is studied by Leray-Schauder degree theory. These results indicate the importance of the large diffusivity which is responsible for the appearance and non-appearance of stationary patterns. We have discussed Turing instability, which ensures the existence of Turing patterns. Further, the effect of the cannibalistic attack rate and disease transmission rate observed on the dynamics of the proposed model system. Even it followed that in the absence of cannibalism, the spatial distribution is not possible and increment in the cannibalistic attack rate and disease transmission rate promote the Turing patterns. The similar effect observed for the disease transmission rate. Further, we have calculated Lyapunov exponents and saw the chaotic and stable fixed dynamics for various parameter settings. In the last, we have performed extensive numerical simulation and obtained Turing and spatiotemporal patterns. Keywords cannibalism • local stability • global stability • nonconstant positive steady state • priori estimates • Turing instability and Turing patterns • spatiotemporal patterns • Lyapunov exponents. Diffusion gives rise to a pattern formation phenomenon. Diffusion plays an important role in population biology. Starting from the Turing seminal work [27], Brusselator model [3], Gierer-Meinhardt model [10, 30], Sel'kov model [5, 29], Lotka-Volterra predator-prey model [17, 6] and the references therein, self diffusion and cross-diffusion are