https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117765
--- Comment #10 from kargls at comcast dot net --- On 11/25/24 09:50, anlauf at gcc dot gnu.org wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117765 > > --- Comment #7 from anlauf at gcc dot gnu.org --- > The patch catches functions but misses subroutine calls, as in: > > ! { dg-do compile } > ! > program foo > > implicit none > > integer i > integer :: j = 0 > real y(4) > > do concurrent(i=1:4) > y(i) = bar(i) ! { dg-error "Reference to impure function" } > call bla(i) ! <<< caught > end do > > do concurrent(i=1:4) > block > y(i) = bar(i) ! { dg-error "Reference to impure function" } > call bla(i) ! <<< MISSED! > end block > end do > > contains > > impure function bar(i) > real bar > integer, intent(in) :: i > j = j + i > bar = j > end function bar > > impure subroutine bla (j) > integer, intent(in) :: j > print *, j > end subroutine bla > > end program foo > > > If possible, can we fix that, too? > The attached patch now catches the impure subroutine calls. I have updated the testcase to include tests for calls. 2024-11-26 Steven G. Kargl <kar...@comcast.net> * resolve.cc (pure_subroutine): Check for an impure subroutine call in a BLOCK construct nested within a DO CONCURRENT block. 2024-11-25 Steven G. Kargl <kar...@comcast.net> PR fortran/117765 * gfortran.dg/impure_fcn_do_concurrent.f90: Update test to catch calls to an impure subroutine.
