The "CRC" algorithm reveals the low byte of the previously-used CRC table element. The low byte can be used to determine the index in the table (because each element has a unique low byte), and the index can be used to determine the previous character of the password (up to 7 characters). The final character of the password can be easily brute-forced.

https://gist.github.com/SeanPesce/fdc9f5eaac66b11e1077250773fd12a5

Original writeup (https://gist.github.com/SeanPesce/fdc9f5eaac66b11e1077250773fd12a5).