There is a separate issue which I think only applies if the backup itself has over 2^32 blocks written to tape so that the next tape starting block has wrapped.
I haven't tried to reproduce this issue as I do not have any systems with anything like that amount of data to test with plus I suspect it will take a very long time to reproduce. This doesn't affect this bug which, as per testcase.sh, can be reproduced with only 10MB of data being dumped. I'm not reporting this other issue unless/until I can reproduce it somehow - particularly as I believe it will require a format change to the tape header and I backup via a pipe to a file rather than a tape device so I would never hit this even with 12T of data to backup. 12TB sparse files might make this reproduceable - I haven't tried, nor have I looked in detail yet how sparse files are represented on tape.
