How does double-blind review handle author rebuttals?

When a conference is configured double-blind, how are reviewer identities masked during the rebuttal phase? Do authors ever see reviewer names?