https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82250
Tom de Vries <vries at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |vries at gcc dot gnu.org
--- Comment #1 from Tom de Vries <vries at gcc dot gnu.org> ---
In libgomp/openacc.f90, we find:
....
module openacc_internal
use openacc_kinds
implicit none
interface
function acc_on_device_h (d)
import
integer (acc_device_kind) d
logical acc_on_device_h
end function
end interface
interface
function acc_on_device_l (d) &
bind (C, name = "acc_on_device")
use iso_c_binding, only: c_int
integer (c_int) :: acc_on_device_l
integer (c_int), value :: d
end function
end interface
end module
module openacc
use openacc_kinds
use openacc_internal
implicit none
public :: acc_on_device
interface acc_on_device
procedure :: acc_on_device_h
end interface
end module
function acc_on_device_h (d)
use openacc_internal, only: acc_on_device_l
use openacc_kinds
integer (acc_device_kind) d
logical acc_on_device_h
if (acc_on_device_l (d) .eq. 1) then
acc_on_device_h = .TRUE.
else
acc_on_device_h = .FALSE.
end if
end function
...
AFAICT, a fortran test-case with an acc_on_device function call is translated
to an acc_on_device_h function call at gimple level. This is treated by the
rest of the compiler as a normal function call, there's no connection to a
builtin.
