http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60255
--- Comment #5 from Antony Lewis <antony at cosmologist dot info> --- The patch generated a SIGSEGV in test code (which works with ifort), but could be another unrelated issue. Here's another simple test case for the original issue: program test character(LEN=:), allocatable :: S call subP(S) contains subroutine subP(P) class(*) :: P end subroutine end program