Yes, normality is not a key word in SPC.

No, never eve. 

It depends on the root causes that first lead you to assert that the normaly distribution is not assumed. In fact you cannot directly go on using SPC methods but usually you can after some operations or some extra analysis and still continue and improve the process.