If a prepared statement fails due eg to a disallowed duplicate column value then the handle is destroyed instead of reset. A seqfault follows if it is resused. See link above for test script and patch. If this is not fixed soon upstream should there be a gentoo patch? Reproducible: Always Steps to Reproduce: See perl script a link.
Created attachment 127820 [details] test case
Created attachment 127822 [details, diff] patch
Fixed quite some time ago according to usptream issue tracker.