Comment 8 for bug 1177206

>OK, the memory corruption seems that it was gdb/optimizations. After building XB and mini-InnoDB with optimizations off, the recv_addr and others look much more rational. Now...

I do not think so because when I reproduced the bug with disabled optimization I saw log records list is corrupted. So I think we should not exclude memory corruption. We need test case that reproduce the bug consistently to explore it more deeply.

> Something is out of sync between the bpage read in the buf_page_io_complete and the actual page...

*.delta file for space 24 is null-filled and data file that corresponds to space 24 is absent in base backup. That means space 24 was created between base and first incremental backups.

Maybe the steps that create space in server runtime differ from the steps that are reproduced during log recovery.