Why doesn't Repeatable-read isolation level cause non-reaptable read problem?

Question

I know that both write and read are long-term locks. But even if it's all long-term, if transaction A reads the data with a read lock, and then transaction B changes and commits the data with a write lock, wouldn't fuzzy read occur when A reads the data again?

I really don't know how to solve the problem of fuzzy read at the read committed isolation level.