https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121475
Bug ID: 121475
Summary: Missed finalization
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: abensonca at gcc dot gnu.org
Target Milestone: ---
The following appears to be a case where finalization of a function result is
missed by the latest gfortran:
module bugMod
type :: r
contains
procedure :: rma
generic :: assignment(=) => rma
end type r
type, extends(r) :: rm
integer :: i
contains
final :: rmd
end type rm
interface rm
module procedure rmc
end interface rm
contains
subroutine rmd(self)
implicit none
type(rm), intent(inout) :: self
write (0,*) "type(rm) destructor called"
return
end subroutine rmd
function rmc() result(self)
implicit none
type(rm) :: self
write (0,*) "type(rm) constructor called"
self%i=0
return
end function rmc
subroutine rma(to,from)
implicit none
class(r), intent( out) :: to
class(r), intent(in ) :: from
write (0,*) "class(r) assignment(=) called"
select type (to)
type is (rm)
select type (from)
type is (rm)
to%i=from%i
end select
end select
return
end subroutine rma
end module bugMod
program bugDestructFuncResult
use bugMod
implicit none
block
type(rm) :: i
write (0,*) "construct our instance"
i=rm()
write (0,*) "finished"
end block
end program bugDestructFuncResult
In this case I have a defined assignment for the parent class r of type rm.
When assignment is done (in the constructor call i=rm()) I expect two calls to
the finalizer for type rm (one to finalize the left hand side of the
assignment, and one to finalize the function result. But, with the current
gfortran this gives:
> gfortran -v
Using built-in specs.
COLLECT_GCC=gfortran
COLLECT_LTO_WRAPPER=/carnegie/nobackup/users/abenson/upstream/libexec/gcc/x86_64-pc-linux-gnu/16.0.0/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: ../configure
--prefix=/carnegie/nobackup/users/abenson/upstream --disable-multilib
--enable-checking=release --enable-host-shared --with-pic
--enable-languages=c,c++,fortran,jit,lto
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 16.0.0 20250804 (experimental) (GCC)
> gfortran -g bugDestructFuncResult.F90
> ./a.out
construct our instance
type(rm) constructor called
type(rm) destructor called
class(r) assignment(=) called
finished
type(rm) destructor called
so only one call to the finalizer between the construct of the instance and
finished outputs (there's a last finalizer call at the end block when i is
finalized).
With ifort, I get:
> ifort -g bugDestructFuncResult.F90
> ./a.out
construct our instance
type(rm) constructor called
type(rm) destructor called
class(r) assignment(=) called
type(rm) destructor called
finished
type(rm) destructor called
so two calls to the finalizer as expected.
