> Currently there is a small race window as pointed out by Jan [2] where, when > ext4 tries to allocate a written block for mapped files and if DIO read is in > progress, then this may result into stale data read exposure problem. Do we have a test case for the problem?
