heteroskedasticity and Wu-Hausman endogeneity?

[econmcg]

Hey this might be a simple question, but does anyone know what the effects are of
heteroskedasticity and/or auto-correlation in the errors in a 2sls fixed effect panel model
on the wu-hausman endogeneity test and it's assumptions for the instruments?

Am I right in assuming that the exogeneity tests of the instruments are no longer correct
because instruments are regressed on residuals of the 2sls model? And those residuals
are no longer correct under heteroskedasticity.

And the same for the wu-Hausman test itself, where the structual model is regressed on the
reduced form residuals. Which would no longer be correct due to heterodkedasticity?

Thanks a million in advance.