The current documentation says:
When applications are using both checkpoint-level and commit-level durability, checkpoints taken while a backup cursor is open are not durable. That is, if a crash occurs when a backup cursor is open, then the system will be restored to the most recent checkpoint prior to the opening of the backup cursor, even if later database checkpoints were completed.
That could be read to imply that committed transactions in a logging system are discarded if we crash during a backup operation.