Skip to content

[flang][OpenMP] Implement copyin for pointers and allocatables. #107425

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Sep 10, 2024
Merged

[flang][OpenMP] Implement copyin for pointers and allocatables. #107425

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ee837f6
[flang][OpenMP] Implement copyin for pointers and allocatables.
DavidTruby Sep 5, 2024
9bd09f5
update copyVarHLFIR with correct behaviour for copyin
DavidTruby Sep 9, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 4 additions & 3 deletions flang/lib/Lower/Bridge.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1261,9 +1261,10 @@ class FirConverter : public Fortran::lower::AbstractConverter {
auto loadVal = builder->create<fir::LoadOp>(loc, rhs);
builder->create<fir::StoreOp>(loc, loadVal, lhs);
} else if (isAllocatable &&
flags.test(Fortran::semantics::Symbol::Flag::OmpFirstPrivate)) {
// For firstprivate allocatable variables, RHS must be copied only when
// LHS is allocated.
(flags.test(Fortran::semantics::Symbol::Flag::OmpFirstPrivate) ||
flags.test(Fortran::semantics::Symbol::Flag::OmpCopyIn))) {
// For firstprivate and copyin allocatable variables, RHS must be copied
// only when LHS is allocated.
hlfir::Entity temp =
hlfir::derefPointersAndAllocatables(loc, *builder, lhs);
mlir::Value addr = hlfir::genVariableRawAddress(loc, *builder, temp);
Expand Down
4 changes: 1 addition & 3 deletions flang/lib/Lower/OpenMP/ClauseProcessor.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -619,9 +619,7 @@ bool ClauseProcessor::processCopyin() const {
checkAndCopyHostAssociateVar(&*mem, &insPt);
break;
}
if (semantics::IsAllocatableOrObjectPointer(&sym->GetUltimate()))
TODO(converter.getCurrentLocation(),
"pointer or allocatable variables in Copyin clause");

assert(sym->has<semantics::HostAssocDetails>() &&
"No host-association found");
checkAndCopyHostAssociateVar(sym);
Expand Down
94 changes: 94 additions & 0 deletions flang/test/Lower/OpenMP/copyin.f90
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -356,3 +356,97 @@ subroutine common_2()
end do
!$omp end parallel do
end subroutine

! CHECK-LABEL: func.func @_QPpointer() {
! CHECK: %[[VAL_0:.*]] = fir.address_of(@_QFpointerEp) : !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>
! CHECK: %[[VAL_1:.*]]:2 = hlfir.declare %[[VAL_0]] {fortran_attrs = #fir.var_attrs<pointer>, uniq_name = "_QFpointerEp"} : (!fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>) -> (!fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>, !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>)
! CHECK: %[[VAL_2:.*]] = omp.threadprivate %[[VAL_1]]#1 : !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>> -> !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>
! CHECK: %[[VAL_3:.*]]:2 = hlfir.declare %[[VAL_2]] {fortran_attrs = #fir.var_attrs<pointer>, uniq_name = "_QFpointerEp"} : (!fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>) -> (!fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>, !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>)
! CHECK: omp.parallel {
! CHECK: %[[VAL_4:.*]] = fir.alloca !fir.box<!fir.heap<!fir.array<?xi32>>> {pinned}
! CHECK: %[[VAL_5:.*]] = omp.threadprivate %[[VAL_1]]#1 : !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>> -> !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>
! CHECK: %[[VAL_6:.*]]:2 = hlfir.declare %[[VAL_5]] {fortran_attrs = #fir.var_attrs<pointer>, uniq_name = "_QFpointerEp"} : (!fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>) -> (!fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>, !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>)
! CHECK: %[[VAL_7:.*]] = fir.load %[[VAL_3]]#0 : !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>
! CHECK: fir.store %[[VAL_7]] to %[[VAL_6]]#0 : !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>
! CHECK: omp.barrier
! CHECK: %[[VAL_8:.*]] = fir.load %[[VAL_6]]#0 : !fir.ref<!fir.box<!fir.ptr<!fir.array<?xi32>>>>
! CHECK: %[[VAL_9:.*]]:2 = hlfir.copy_in %[[VAL_8]] to %[[VAL_4]] : (!fir.box<!fir.ptr<!fir.array<?xi32>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> (!fir.box<!fir.ptr<!fir.array<?xi32>>>, i1)
! CHECK: %[[VAL_10:.*]] = fir.box_addr %[[VAL_9]]#0 : (!fir.box<!fir.ptr<!fir.array<?xi32>>>) -> !fir.ptr<!fir.array<?xi32>>
! CHECK: %[[VAL_11:.*]] = fir.convert %[[VAL_10]] : (!fir.ptr<!fir.array<?xi32>>) -> !fir.ref<!fir.array<?xi32>>
! CHECK: fir.call @_QPsub7(%[[VAL_11]]) fastmath<contract> : (!fir.ref<!fir.array<?xi32>>) -> ()
! CHECK: hlfir.copy_out %[[VAL_4]], %[[VAL_9]]#1 to %[[VAL_8]] : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, i1, !fir.box<!fir.ptr<!fir.array<?xi32>>>) -> ()
! CHECK: omp.terminator
! CHECK: }
! CHECK: return
! CHECK: }
subroutine pointer()
integer, pointer, save :: p(:)
!$omp threadprivate(p)

!$omp parallel copyin(p)
call sub7(p)
!$omp end parallel
end subroutine

! CHECK-LABEL: func.func @_QPallocatable() {
! CHECK: %[[VAL_0:.*]] = fir.address_of(@_QFallocatableEp) : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
! CHECK: %[[VAL_1:.*]]:2 = hlfir.declare %[[VAL_0]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFallocatableEp"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>)
! CHECK: %[[VAL_2:.*]] = omp.threadprivate %[[VAL_1]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
! CHECK: %[[VAL_3:.*]]:2 = hlfir.declare %[[VAL_2]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFallocatableEp"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>)
! CHECK: omp.parallel {
! CHECK: %[[VAL_4:.*]] = omp.threadprivate %[[VAL_1]]#1 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>> -> !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
! CHECK: %[[VAL_5:.*]]:2 = hlfir.declare %[[VAL_4]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFallocatableEp"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>)
! CHECK: %[[VAL_6:.*]] = fir.load %[[VAL_3]]#0 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
! CHECK: hlfir.assign %[[VAL_6]] to %[[VAL_5]]#0 realloc : !fir.box<!fir.heap<!fir.array<?xi32>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
! CHECK: omp.barrier
! CHECK: %[[VAL_7:.*]] = fir.load %[[VAL_5]]#0 : !fir.ref<!fir.box<!fir.heap<!fir.array<?xi32>>>>
! CHECK: %[[VAL_8:.*]] = fir.box_addr %[[VAL_7]] : (!fir.box<!fir.heap<!fir.array<?xi32>>>) -> !fir.heap<!fir.array<?xi32>>
! CHECK: %[[VAL_9:.*]] = fir.convert %[[VAL_8]] : (!fir.heap<!fir.array<?xi32>>) -> !fir.ref<!fir.array<?xi32>>
! CHECK: fir.call @_QPsub8(%[[VAL_9]]) fastmath<contract> : (!fir.ref<!fir.array<?xi32>>) -> ()
! CHECK: omp.terminator
! CHECK: }
! CHECK: return
! CHECK: }
subroutine allocatable()
integer, allocatable, save :: p(:)
!$omp threadprivate(p)

!$omp parallel copyin(p)
call sub8(p)
!$omp end parallel
end subroutine

! CHECK-LABEL: func.func @_QPallocatable2() {
! CHECK: %[[VAL_0:.*]] = fir.address_of(@_QFallocatable2Ea) : !fir.ref<!fir.box<!fir.heap<i32>>>
! CHECK: %[[VAL_1:.*]]:2 = hlfir.declare %[[VAL_0]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFallocatable2Ea"} : (!fir.ref<!fir.box<!fir.heap<i32>>>) -> (!fir.ref<!fir.box<!fir.heap<i32>>>, !fir.ref<!fir.box<!fir.heap<i32>>>)
! CHECK: %[[VAL_2:.*]] = omp.threadprivate %[[VAL_1]]#1 : !fir.ref<!fir.box<!fir.heap<i32>>> -> !fir.ref<!fir.box<!fir.heap<i32>>>
! CHECK: %[[VAL_3:.*]]:2 = hlfir.declare %[[VAL_2]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFallocatable2Ea"} : (!fir.ref<!fir.box<!fir.heap<i32>>>) -> (!fir.ref<!fir.box<!fir.heap<i32>>>, !fir.ref<!fir.box<!fir.heap<i32>>>)
! CHECK: omp.parallel {
! CHECK: %[[VAL_4:.*]] = omp.threadprivate %[[VAL_1]]#1 : !fir.ref<!fir.box<!fir.heap<i32>>> -> !fir.ref<!fir.box<!fir.heap<i32>>>
! CHECK: %[[VAL_5:.*]]:2 = hlfir.declare %[[VAL_4]] {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFallocatable2Ea"} : (!fir.ref<!fir.box<!fir.heap<i32>>>) -> (!fir.ref<!fir.box<!fir.heap<i32>>>, !fir.ref<!fir.box<!fir.heap<i32>>>)
! CHECK: %[[VAL_6:.*]] = fir.load %[[VAL_5]]#0 : !fir.ref<!fir.box<!fir.heap<i32>>>
! CHECK: %[[VAL_7:.*]] = fir.box_addr %[[VAL_6]] : (!fir.box<!fir.heap<i32>>) -> !fir.heap<i32>
! CHECK: %[[VAL_8:.*]] = fir.convert %[[VAL_7]] : (!fir.heap<i32>) -> i64
! CHECK: %[[VAL_9:.*]] = arith.constant 0 : i64
! CHECK: %[[VAL_10:.*]] = arith.cmpi ne, %[[VAL_8]], %[[VAL_9]] : i64
! CHECK: fir.if %[[VAL_10]] {
! CHECK: %[[VAL_11:.*]] = fir.load %[[VAL_3]]#0 : !fir.ref<!fir.box<!fir.heap<i32>>>
! CHECK: %[[VAL_12:.*]] = fir.box_addr %[[VAL_11]] : (!fir.box<!fir.heap<i32>>) -> !fir.heap<i32>
! CHECK: %[[VAL_13:.*]] = fir.load %[[VAL_12]] : !fir.heap<i32>
! CHECK: hlfir.assign %[[VAL_13]] to %[[VAL_5]]#0 realloc : i32, !fir.ref<!fir.box<!fir.heap<i32>>>
! CHECK: }
! CHECK: omp.barrier
! CHECK: %[[VAL_14:.*]] = arith.constant 1 : i32
! CHECK: hlfir.assign %[[VAL_14]] to %[[VAL_5]]#0 realloc : i32, !fir.ref<!fir.box<!fir.heap<i32>>>
! CHECK: omp.terminator
! CHECK: }
! CHECK: return
! CHECK: }
subroutine allocatable2()
integer, allocatable, save :: a
!$omp threadprivate(a)

!$omp parallel copyin(a)
a = 1
!$omp end parallel
end subroutine
Loading