Comment 3 for bug 1206309

The problem is a combination of 2 things:

1. InnoDB log block numbers wrap around at 0x40000000. So the next log block after 1073741823 has number 0, rather than 1073741824.
2. XtraBackup uses the following logic to detect "old" blocks (i.e. the current log tail after which it has to stop scanning): if the current log block number is less than the current scanned log block number, it is an old one, stop scanning.

#2 doesn't work in cases when log block numbers wrap around. I.e. we might be looking at an old block with a number before they had wrapped around, but the current "expected" number has already wrapped around.