MySQL 5.6.45 contains a change that refers to Oracle Bug #25289359 DML/DDL ON LARGE FULLTEXT TABLES CAUSE SEMAPHORE TIMEOUTS AND ASSERTION/SUICIDE.
There is no test case, but there is debug instrumentation. The change continues the bogus assumption that time(NULL)) is a monotonically increasing sequence. That assumption is demonstrably broken in
MDEV-14154. The change might also assume fair scheduling among the threads, which might not hold on a heavily loaded system.
I think that we should study what the problem is and whether it affects MariaDB, and then come up with a better fix.