https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65813
--- Comment #1 from Ian Lance Taylor <ian at airs dot com> --- I think you know this, but to be clear, the test is supposed to dereference a null pointer, and then it's supposed to recover from the run time panic. The program should unwind the stack for the signal and call __go_check_defer which should invoke the deferred function. Can you find out where that process is going awry? If the recover* tests pass, then this test ought to pass, or at least it ought not to fail in this way.