Dear Riaz,

I have written a guide to EFA. Normality is usually not tested (according to Tabachnik and Fidel). Variables are often coarse, ordinal data. Multicollinearity is important. Field recommends calculating bivariate correlations between all items and looking for absolute values greater than 0.8, but this is not sufficient to identify all multicollinearity. Sample size adequacy can be left to the validation stage at the end of the analysis as items are often removed during the process and the analysis is repeated.

Thank you Dear @Peter Samuels sharing the facts.No doubt, sample size is important but i have studied some articles in which multivariate normality and collinearity of data have analysed before EFA. I also talking about multivariate normality not just normality as you suggested.