Formal methods could be used to prevent side channel attacks. However preventing each type of attack requires specifying properties which needs to be proven to

Formal methods could be used to prevent side channel attacks. However preventing each type of attack requires specifying properties which needs to be proven to assure it is is not possible. I am not very famililiar with TLA+, but I think in general it is not guaranteed to provide side channel attacks, such as based on timing, CPU state, etc.