http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49214
--- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-05-29 11:00:42 UTC --- AFAICT this PR is fixed by the patch in http://gcc.gnu.org/ml/fortran/2011-05/msg00220.html , except may be for the following test program allocate_assign implicit none type label integer, allocatable :: parts(:) end type label type table type(label), allocatable :: headers(:) end type table integer, parameter :: ik8 = selected_int_kind(18) type(table) x1(1) type(table) t1 type(table), allocatable :: x(:) integer k integer(ik8) s t1 = table([label([1])]) x1 = [t1] s = 0 do k = 1, 2 write(*,*) "1> ", k,s call foo write(*,*) "2> ",k,s, size (x) call bar end do write(*,*) s contains subroutine foo x = x1 end subroutine subroutine bar s = s+x(2)%headers(2)%parts(2) end subroutine end program allocate_assign which gives a segmentation fault [macbook] f90/bug% gfc pr47517_1.f90 [macbook] f90/bug% a.out 1> 1 0 2> 1 0 1 Backtrace for this error: 0 libgfortran.3.dylib 0x0000000100005dd7 _gfortrani_show_backtrace + 55 1 ??? 0x00007fff5fc0f131 0x0 + 140734799868209 2 a.out 0x0000000100002280 MAIN__ + 2848 3 a.out 0x0000000100002dd7 main + 54 4 a.out 0x000000010000166c start + 52 5 ??? 0x0000000000000001 0x0 + 1 Segmentation fault Further testing pending. Thanks for the patch.
