maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/flang/test/Lower/array-temp.f90

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

414 lines
30 KiB
Fortran
Raw Normal View History

[flang][NFC] Add array lowering tests These tests were left behind during the upstreaming of parts lowering. This patch is part of the upstreaming effort from fir-dev branch. Reviewed By: jeanPerier Differential Revision: https://reviews.llvm.org/D128632 Co-authored-by: V Donaldson <vdonaldson@nvidia.com> Co-authored-by: Jean Perier <jperier@nvidia.com> Co-authored-by: Eric Schweitz <eschweitz@nvidia.com>
2022-06-27 20:19:23 +08:00
! RUN: bbc %s -o - | FileCheck %s
! CHECK-LABEL: func @_QPss1()
subroutine ss1
! CHECK: %[[aa:[0-9]+]] = fir.alloca !fir.array<2650000xf32> {bindc_name = "aa", uniq_name = "_QFss1Eaa"}
! CHECK: %[[shape:[0-9]+]] = fir.shape {{.*}} : (index) -> !fir.shape<1>
integer, parameter :: N = 2650000
real aa(N)
! CHECK: fir.array_coor %[[aa]](%[[shape]]) {{.*}} : (!fir.ref<!fir.array<2650000xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
aa = -2
! CHECK: %[[temp:[0-9]+]] = fir.allocmem !fir.array<2650000xf32>
! CHECK: fir.array_coor %[[aa]](%[[shape]]) {{.*}} : (!fir.ref<!fir.array<2650000xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: fir.array_coor %[[temp]](%[[shape]]) {{.*}} : (!fir.heap<!fir.array<2650000xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: fir.array_coor %[[aa]](%[[shape]]) [{{.*}}] {{.*}} : (!fir.ref<!fir.array<2650000xf32>>, !fir.shape<1>, !fir.slice<1>, index) -> !fir.ref<f32>
! CHECK: fir.array_coor %[[temp]](%[[shape]]) [{{.*}}] {{.*}} : (!fir.heap<!fir.array<2650000xf32>>, !fir.shape<1>, !fir.slice<1>, index) -> !fir.ref<f32>
! CHECK: fir.array_coor %[[temp]](%[[shape]]) {{.*}} : (!fir.heap<!fir.array<2650000xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: fir.array_coor %[[aa]](%[[shape]]) {{.*}} : (!fir.ref<!fir.array<2650000xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: fir.freemem %[[temp]] : !fir.heap<!fir.array<2650000xf32>>
aa(2:N) = aa(1:N-1) + 7.0
! print*, aa(1:2), aa(N-1:N)
end
subroutine ss2(N)
real aa(N)
aa = -2
aa(2:N) = aa(1:N-1) + 7.0
print*, aa(1:2), aa(N-1:N)
end
subroutine ss3(N)
real aa(2,N)
aa = -2
aa(:,2:N) = aa(:,1:N-1) + 7.0
print*, aa(:,1:2), aa(:,N-1:N)
end
subroutine ss4(N)
real aa(N,2)
aa = -2
aa(2:N,:) = aa(1:N-1,:) + 7.0
print*, aa(1:2,:), aa(N-1:N,:)
end
! CHECK-LABEL: func @_QPss2(
! CHECK-SAME: %[[VAL_0:.*]]: !fir.ref<i32> {fir.bindc_name = "n"}) {
! CHECK-DAG: %[[VAL_1:.*]] = arith.constant -1 : index
! CHECK-DAG: %[[VAL_2:.*]] = arith.constant -2 : i32
! CHECK-DAG: %[[VAL_3:.*]] = arith.constant 1 : index
! CHECK-DAG: %[[VAL_4:.*]] = arith.constant 0 : index
! CHECK-DAG: %[[VAL_5:.*]] = arith.constant 2 : index
! CHECK-DAG: %[[VAL_6:.*]] = arith.constant 1 : i32
! CHECK-DAG: %[[VAL_7:.*]] = arith.constant 7.000000e+00 : f32
! CHECK-DAG: %[[VAL_8:.*]] = arith.constant -1 : i32
! CHECK: %[[VAL_10:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_11A:.*]] = fir.convert %[[VAL_10]] : (i32) -> index
! CHECK: %[[CMP:.*]] = arith.cmpi sgt, %[[VAL_11A]], %[[VAL_4]] : index
! CHECK: %[[VAL_11:.*]] = arith.select %[[CMP]], %[[VAL_11A]], %[[VAL_4]] : index
! CHECK: %[[VAL_12:.*]] = fir.alloca !fir.array<?xf32>, %[[VAL_11]] {bindc_name = "aa", uniq_name = "_QFss2Eaa"}
! CHECK: %[[VAL_13:.*]] = fir.shape %[[VAL_11]] : (index) -> !fir.shape<1>
! CHECK: %[[VAL_14:.*]] = fir.convert %[[VAL_2]] : (i32) -> f32
! CHECK: br ^bb1(%[[VAL_4]], %[[VAL_11]] : index, index)
! CHECK: ^bb1(%[[VAL_15:.*]]: index, %[[VAL_16:.*]]: index):
! CHECK: %[[VAL_17:.*]] = arith.cmpi sgt, %[[VAL_16]], %[[VAL_4]] : index
! CHECK: cond_br %[[VAL_17]], ^bb2, ^bb3
! CHECK: ^bb2:
! CHECK: %[[VAL_18:.*]] = arith.addi %[[VAL_15]], %[[VAL_3]] : index
! CHECK: %[[VAL_19:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_18]] : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: fir.store %[[VAL_14]] to %[[VAL_19]] : !fir.ref<f32>
! CHECK: %[[VAL_20:.*]] = arith.subi %[[VAL_16]], %[[VAL_3]] : index
! CHECK: br ^bb1(%[[VAL_18]], %[[VAL_20]] : index, index)
! CHECK: ^bb3:
! CHECK: %[[VAL_21:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_22:.*]] = fir.convert %[[VAL_21]] : (i32) -> index
! CHECK: %[[VAL_23:.*]] = arith.addi %[[VAL_22]], %[[VAL_1]] : index
! CHECK: %[[VAL_24:.*]] = arith.cmpi sgt, %[[VAL_23]], %[[VAL_4]] : index
! CHECK: %[[VAL_25:.*]] = arith.select %[[VAL_24]], %[[VAL_23]], %[[VAL_4]] : index
! CHECK: %[[VAL_26:.*]] = fir.slice %[[VAL_5]], %[[VAL_22]], %[[VAL_3]] : (index, index, index) -> !fir.slice<1>
! CHECK: %[[VAL_27:.*]] = fir.allocmem !fir.array<?xf32>, %[[VAL_11]]
! CHECK: br ^bb4(%[[VAL_4]], %[[VAL_11]] : index, index)
! CHECK: ^bb4(%[[VAL_28:.*]]: index, %[[VAL_29:.*]]: index):
! CHECK: %[[VAL_30:.*]] = arith.cmpi sgt, %[[VAL_29]], %[[VAL_4]] : index
! CHECK: cond_br %[[VAL_30]], ^bb5, ^bb6
! CHECK: ^bb5:
! CHECK: %[[VAL_31:.*]] = arith.addi %[[VAL_28]], %[[VAL_3]] : index
! CHECK: %[[VAL_32:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_31]] : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: %[[VAL_33:.*]] = fir.array_coor %[[VAL_27]](%[[VAL_13]]) %[[VAL_31]] : (!fir.heap<!fir.array<?xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: %[[VAL_34:.*]] = fir.load %[[VAL_32]] : !fir.ref<f32>
! CHECK: fir.store %[[VAL_34]] to %[[VAL_33]] : !fir.ref<f32>
! CHECK: %[[VAL_35:.*]] = arith.subi %[[VAL_29]], %[[VAL_3]] : index
! CHECK: br ^bb4(%[[VAL_31]], %[[VAL_35]] : index, index)
! CHECK: ^bb6:
! CHECK: %[[VAL_36:.*]] = arith.subi %[[VAL_21]], %[[VAL_6]] : i32
! CHECK: %[[VAL_37:.*]] = fir.convert %[[VAL_36]] : (i32) -> index
! CHECK: %[[VAL_38:.*]] = fir.slice %[[VAL_3]], %[[VAL_37]], %[[VAL_3]] : (index, index, index) -> !fir.slice<1>
! CHECK: br ^bb7(%[[VAL_4]], %[[VAL_25]] : index, index)
! CHECK: ^bb7(%[[VAL_39:.*]]: index, %[[VAL_40:.*]]: index):
! CHECK: %[[VAL_41:.*]] = arith.cmpi sgt, %[[VAL_40]], %[[VAL_4]] : index
! CHECK: cond_br %[[VAL_41]], ^bb8, ^bb9(%[[VAL_4]], %[[VAL_11]] : index, index)
! CHECK: ^bb8:
! CHECK: %[[VAL_42:.*]] = arith.addi %[[VAL_39]], %[[VAL_3]] : index
! CHECK: %[[VAL_43:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_38]]] %[[VAL_42]] : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, !fir.slice<1>, index) -> !fir.ref<f32>
! CHECK: %[[VAL_44:.*]] = fir.load %[[VAL_43]] : !fir.ref<f32>
! CHECK: %[[VAL_45:.*]] = arith.addf %[[VAL_44]], %[[VAL_7]] : f32
! CHECK: %[[VAL_46:.*]] = fir.array_coor %[[VAL_27]](%[[VAL_13]]) {{\[}}%[[VAL_26]]] %[[VAL_42]] : (!fir.heap<!fir.array<?xf32>>, !fir.shape<1>, !fir.slice<1>, index) -> !fir.ref<f32>
! CHECK: fir.store %[[VAL_45]] to %[[VAL_46]] : !fir.ref<f32>
! CHECK: %[[VAL_47:.*]] = arith.subi %[[VAL_40]], %[[VAL_3]] : index
! CHECK: br ^bb7(%[[VAL_42]], %[[VAL_47]] : index, index)
! CHECK: ^bb9(%[[VAL_48:.*]]: index, %[[VAL_49:.*]]: index):
! CHECK: %[[VAL_50:.*]] = arith.cmpi sgt, %[[VAL_49]], %[[VAL_4]] : index
! CHECK: cond_br %[[VAL_50]], ^bb10, ^bb11
! CHECK: ^bb10:
! CHECK: %[[VAL_51:.*]] = arith.addi %[[VAL_48]], %[[VAL_3]] : index
! CHECK: %[[VAL_52:.*]] = fir.array_coor %[[VAL_27]](%[[VAL_13]]) %[[VAL_51]] : (!fir.heap<!fir.array<?xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: %[[VAL_53:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_51]] : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, index) -> !fir.ref<f32>
! CHECK: %[[VAL_54:.*]] = fir.load %[[VAL_52]] : !fir.ref<f32>
! CHECK: fir.store %[[VAL_54]] to %[[VAL_53]] : !fir.ref<f32>
! CHECK: %[[VAL_55:.*]] = arith.subi %[[VAL_49]], %[[VAL_3]] : index
! CHECK: br ^bb9(%[[VAL_51]], %[[VAL_55]] : index, index)
! CHECK: ^bb11:
! CHECK: fir.freemem %[[VAL_27]] : !fir.heap<!fir.array<?xf32>>
! CHECK: %[[VAL_58:.*]] = fir.call @_FortranAioBeginExternalListOutput(%[[VAL_8]], %{{.*}}, %{{.*}}) : (i32, !fir.ref<i8>, i32) -> !fir.ref<i8>
! CHECK: %[[VAL_59:.*]] = fir.slice %[[VAL_3]], %[[VAL_5]], %[[VAL_3]] : (index, index, index) -> !fir.slice<1>
[flang] Fix one corner case in reshape intrinsic As Fortran 2018 16.9.163, the reshape is the only intrinsic which requires the shape argument to be rank-one integer array and the SIZE of it to be one constant expression. The current expression lowering converts the shape expression with slice in intrinsic into one box value with the box element type of unknown extent. However, the genReshape requires the box element type to be constant size. So, convert the box value into one with box element type of sequence of 1 x constant. This corner case is found in cam4 in SPEC 2017 https://github.com/llvm/llvm-project/issues/56140. Reviewed By: Jean Perier Differential Revision: https://reviews.llvm.org/D128597
2022-06-30 23:52:44 +08:00
! CHECK: %[[VAL_60:.*]] = fir.embox %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_59]]] : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, !fir.slice<1>) -> !fir.box<!fir.array<2xf32>>
! CHECK: %[[VAL_61:.*]] = fir.convert %[[VAL_60]] : (!fir.box<!fir.array<2xf32>>) -> !fir.box<none>
[flang][NFC] Add array lowering tests These tests were left behind during the upstreaming of parts lowering. This patch is part of the upstreaming effort from fir-dev branch. Reviewed By: jeanPerier Differential Revision: https://reviews.llvm.org/D128632 Co-authored-by: V Donaldson <vdonaldson@nvidia.com> Co-authored-by: Jean Perier <jperier@nvidia.com> Co-authored-by: Eric Schweitz <eschweitz@nvidia.com>
2022-06-27 20:19:23 +08:00
! CHECK: %[[VAL_62:.*]] = fir.call @_FortranAioOutputDescriptor(%[[VAL_58]], %[[VAL_61]]) : (!fir.ref<i8>, !fir.box<none>) -> i1
! CHECK: %[[VAL_63:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_64:.*]] = arith.subi %[[VAL_63]], %[[VAL_6]] : i32
! CHECK: %[[VAL_65:.*]] = fir.convert %[[VAL_64]] : (i32) -> index
! CHECK: %[[VAL_66:.*]] = fir.convert %[[VAL_63]] : (i32) -> index
! CHECK: %[[VAL_67:.*]] = fir.slice %[[VAL_65]], %[[VAL_66]], %[[VAL_3]] : (index, index, index) -> !fir.slice<1>
! CHECK: %[[VAL_68:.*]] = fir.embox %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_67]]] : (!fir.ref<!fir.array<?xf32>>, !fir.shape<1>, !fir.slice<1>) -> !fir.box<!fir.array<?xf32>>
! CHECK: %[[VAL_69:.*]] = fir.convert %[[VAL_68]] : (!fir.box<!fir.array<?xf32>>) -> !fir.box<none>
! CHECK: %[[VAL_70:.*]] = fir.call @_FortranAioOutputDescriptor(%[[VAL_58]], %[[VAL_69]]) : (!fir.ref<i8>, !fir.box<none>) -> i1
! CHECK: %[[VAL_71:.*]] = fir.call @_FortranAioEndIoStatement(%[[VAL_58]]) : (!fir.ref<i8>) -> i32
! CHECK: return
! CHECK: }
! CHECK-LABEL: func @_QPss3(
! CHECK-SAME: %[[VAL_0:.*]]: !fir.ref<i32> {fir.bindc_name = "n"}) {
! CHECK-DAG: %[[VAL_1:.*]] = arith.constant -1 : index
! CHECK-DAG: %[[VAL_2:.*]] = arith.constant 2 : index
! CHECK-DAG: %[[VAL_3:.*]] = arith.constant 1 : index
! CHECK-DAG: %[[VAL_4:.*]] = arith.constant -2 : i32
! CHECK-DAG: %[[VAL_5:.*]] = arith.constant 0 : index
! CHECK-DAG: %[[VAL_6:.*]] = arith.constant 1 : i32
! CHECK-DAG: %[[VAL_7:.*]] = arith.constant 7.000000e+00 : f32
! CHECK-DAG: %[[VAL_8:.*]] = arith.constant -1 : i32
! CHECK: %[[VAL_10:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_11:.*]] = fir.convert %[[VAL_10]] : (i32) -> index
! CHECK: %[[CMP:.*]] = arith.cmpi sgt, %[[VAL_11A]], %[[VAL_5]] : index
! CHECK: %[[VAL_11:.*]] = arith.select %[[CMP]], %[[VAL_11A]], %[[VAL_5]] : index
! CHECK: %[[VAL_12:.*]] = fir.alloca !fir.array<2x?xf32>, %[[VAL_11]] {bindc_name = "aa", uniq_name = "_QFss3Eaa"}
! CHECK: %[[VAL_13:.*]] = fir.shape %[[VAL_2]], %[[VAL_11]] : (index, index) -> !fir.shape<2>
! CHECK: %[[VAL_14:.*]] = fir.convert %[[VAL_4]] : (i32) -> f32
! CHECK: br ^bb1(%[[VAL_5]], %[[VAL_11]] : index, index)
! CHECK: ^bb1(%[[VAL_15:.*]]: index, %[[VAL_16:.*]]: index):
! CHECK: %[[VAL_17:.*]] = arith.cmpi sgt, %[[VAL_16]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_17]], ^bb2(%[[VAL_5]], %[[VAL_2]] : index, index), ^bb5
! CHECK: ^bb2(%[[VAL_18:.*]]: index, %[[VAL_19:.*]]: index):
! CHECK: %[[VAL_20:.*]] = arith.cmpi sgt, %[[VAL_19]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_20]], ^bb3, ^bb4
! CHECK: ^bb3:
! CHECK: %[[VAL_21:.*]] = arith.addi %[[VAL_18]], %[[VAL_3]] : index
! CHECK: %[[VAL_22:.*]] = arith.addi %[[VAL_15]], %[[VAL_3]] : index
! CHECK: %[[VAL_23:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_21]], %[[VAL_22]] : (!fir.ref<!fir.array<2x?xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: fir.store %[[VAL_14]] to %[[VAL_23]] : !fir.ref<f32>
! CHECK: %[[VAL_24:.*]] = arith.subi %[[VAL_19]], %[[VAL_3]] : index
! CHECK: br ^bb2(%[[VAL_21]], %[[VAL_24]] : index, index)
! CHECK: ^bb4:
! CHECK: %[[VAL_25:.*]] = arith.addi %[[VAL_15]], %[[VAL_3]] : index
! CHECK: %[[VAL_26:.*]] = arith.subi %[[VAL_16]], %[[VAL_3]] : index
! CHECK: br ^bb1(%[[VAL_25]], %[[VAL_26]] : index, index)
! CHECK: ^bb5:
! CHECK: %[[VAL_27:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_28:.*]] = fir.convert %[[VAL_27]] : (i32) -> index
! CHECK: %[[VAL_29:.*]] = arith.addi %[[VAL_28]], %[[VAL_1]] : index
! CHECK: %[[VAL_30:.*]] = arith.cmpi sgt, %[[VAL_29]], %[[VAL_5]] : index
! CHECK: %[[VAL_31:.*]] = arith.select %[[VAL_30]], %[[VAL_29]], %[[VAL_5]] : index
! CHECK: %[[VAL_32:.*]] = fir.slice %[[VAL_3]], %[[VAL_2]], %[[VAL_3]], %[[VAL_2]], %[[VAL_28]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
! CHECK: %[[VAL_33:.*]] = fir.allocmem !fir.array<2x?xf32>, %[[VAL_11]]
! CHECK: br ^bb6(%[[VAL_5]], %[[VAL_11]] : index, index)
! CHECK: ^bb6(%[[VAL_34:.*]]: index, %[[VAL_35:.*]]: index):
! CHECK: %[[VAL_36:.*]] = arith.cmpi sgt, %[[VAL_35]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_36]], ^bb7(%[[VAL_5]], %[[VAL_2]] : index, index), ^bb10
! CHECK: ^bb7(%[[VAL_37:.*]]: index, %[[VAL_38:.*]]: index):
! CHECK: %[[VAL_39:.*]] = arith.cmpi sgt, %[[VAL_38]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_39]], ^bb8, ^bb9
! CHECK: ^bb8:
! CHECK: %[[VAL_40:.*]] = arith.addi %[[VAL_37]], %[[VAL_3]] : index
! CHECK: %[[VAL_41:.*]] = arith.addi %[[VAL_34]], %[[VAL_3]] : index
! CHECK: %[[VAL_42:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_40]], %[[VAL_41]] : (!fir.ref<!fir.array<2x?xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_43:.*]] = fir.array_coor %[[VAL_33]](%[[VAL_13]]) %[[VAL_40]], %[[VAL_41]] : (!fir.heap<!fir.array<2x?xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_44:.*]] = fir.load %[[VAL_42]] : !fir.ref<f32>
! CHECK: fir.store %[[VAL_44]] to %[[VAL_43]] : !fir.ref<f32>
! CHECK: %[[VAL_45:.*]] = arith.subi %[[VAL_38]], %[[VAL_3]] : index
! CHECK: br ^bb7(%[[VAL_40]], %[[VAL_45]] : index, index)
! CHECK: ^bb9:
! CHECK: %[[VAL_46:.*]] = arith.addi %[[VAL_34]], %[[VAL_3]] : index
! CHECK: %[[VAL_47:.*]] = arith.subi %[[VAL_35]], %[[VAL_3]] : index
! CHECK: br ^bb6(%[[VAL_46]], %[[VAL_47]] : index, index)
! CHECK: ^bb10:
! CHECK: %[[VAL_48:.*]] = arith.subi %[[VAL_27]], %[[VAL_6]] : i32
! CHECK: %[[VAL_49:.*]] = fir.convert %[[VAL_48]] : (i32) -> index
! CHECK: %[[VAL_50:.*]] = fir.slice %[[VAL_3]], %[[VAL_2]], %[[VAL_3]], %[[VAL_3]], %[[VAL_49]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
! CHECK: br ^bb11(%[[VAL_5]], %[[VAL_31]] : index, index)
! CHECK: ^bb11(%[[VAL_51:.*]]: index, %[[VAL_52:.*]]: index):
! CHECK: %[[VAL_53:.*]] = arith.cmpi sgt, %[[VAL_52]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_53]], ^bb12(%[[VAL_5]], %[[VAL_2]] : index, index), ^bb15(%[[VAL_5]], %[[VAL_11]] : index, index)
! CHECK: ^bb12(%[[VAL_54:.*]]: index, %[[VAL_55:.*]]: index):
! CHECK: %[[VAL_56:.*]] = arith.cmpi sgt, %[[VAL_55]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_56]], ^bb13, ^bb14
! CHECK: ^bb13:
! CHECK: %[[VAL_57:.*]] = arith.addi %[[VAL_54]], %[[VAL_3]] : index
! CHECK: %[[VAL_58:.*]] = arith.addi %[[VAL_51]], %[[VAL_3]] : index
! CHECK: %[[VAL_59:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_50]]] %[[VAL_57]], %[[VAL_58]] : (!fir.ref<!fir.array<2x?xf32>>, !fir.shape<2>, !fir.slice<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_60:.*]] = fir.load %[[VAL_59]] : !fir.ref<f32>
! CHECK: %[[VAL_61:.*]] = arith.addf %[[VAL_60]], %[[VAL_7]] : f32
! CHECK: %[[VAL_62:.*]] = fir.array_coor %[[VAL_33]](%[[VAL_13]]) {{\[}}%[[VAL_32]]] %[[VAL_57]], %[[VAL_58]] : (!fir.heap<!fir.array<2x?xf32>>, !fir.shape<2>, !fir.slice<2>, index, index) -> !fir.ref<f32>
! CHECK: fir.store %[[VAL_61]] to %[[VAL_62]] : !fir.ref<f32>
! CHECK: %[[VAL_63:.*]] = arith.subi %[[VAL_55]], %[[VAL_3]] : index
! CHECK: br ^bb12(%[[VAL_57]], %[[VAL_63]] : index, index)
! CHECK: ^bb14:
! CHECK: %[[VAL_64:.*]] = arith.addi %[[VAL_51]], %[[VAL_3]] : index
! CHECK: %[[VAL_65:.*]] = arith.subi %[[VAL_52]], %[[VAL_3]] : index
! CHECK: br ^bb11(%[[VAL_64]], %[[VAL_65]] : index, index)
! CHECK: ^bb15(%[[VAL_66:.*]]: index, %[[VAL_67:.*]]: index):
! CHECK: %[[VAL_68:.*]] = arith.cmpi sgt, %[[VAL_67]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_68]], ^bb16(%[[VAL_5]], %[[VAL_2]] : index, index), ^bb19
! CHECK: ^bb16(%[[VAL_69:.*]]: index, %[[VAL_70:.*]]: index):
! CHECK: %[[VAL_71:.*]] = arith.cmpi sgt, %[[VAL_70]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_71]], ^bb17, ^bb18
! CHECK: ^bb17:
! CHECK: %[[VAL_72:.*]] = arith.addi %[[VAL_69]], %[[VAL_3]] : index
! CHECK: %[[VAL_73:.*]] = arith.addi %[[VAL_66]], %[[VAL_3]] : index
! CHECK: %[[VAL_74:.*]] = fir.array_coor %[[VAL_33]](%[[VAL_13]]) %[[VAL_72]], %[[VAL_73]] : (!fir.heap<!fir.array<2x?xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_75:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_72]], %[[VAL_73]] : (!fir.ref<!fir.array<2x?xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_76:.*]] = fir.load %[[VAL_74]] : !fir.ref<f32>
! CHECK: fir.store %[[VAL_76]] to %[[VAL_75]] : !fir.ref<f32>
! CHECK: %[[VAL_77:.*]] = arith.subi %[[VAL_70]], %[[VAL_3]] : index
! CHECK: br ^bb16(%[[VAL_72]], %[[VAL_77]] : index, index)
! CHECK: ^bb18:
! CHECK: %[[VAL_78:.*]] = arith.addi %[[VAL_66]], %[[VAL_3]] : index
! CHECK: %[[VAL_79:.*]] = arith.subi %[[VAL_67]], %[[VAL_3]] : index
! CHECK: br ^bb15(%[[VAL_78]], %[[VAL_79]] : index, index)
! CHECK: ^bb19:
! CHECK: fir.freemem %[[VAL_33]] : !fir.heap<!fir.array<2x?xf32>>
! CHECK: %[[VAL_82:.*]] = fir.call @_FortranAioBeginExternalListOutput(%[[VAL_8]], %{{.*}}, %{{.*}}) : (i32, !fir.ref<i8>, i32) -> !fir.ref<i8>
! CHECK: %[[VAL_83:.*]] = fir.slice %[[VAL_3]], %[[VAL_2]], %[[VAL_3]], %[[VAL_3]], %[[VAL_2]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
[flang] Fix one corner case in reshape intrinsic As Fortran 2018 16.9.163, the reshape is the only intrinsic which requires the shape argument to be rank-one integer array and the SIZE of it to be one constant expression. The current expression lowering converts the shape expression with slice in intrinsic into one box value with the box element type of unknown extent. However, the genReshape requires the box element type to be constant size. So, convert the box value into one with box element type of sequence of 1 x constant. This corner case is found in cam4 in SPEC 2017 https://github.com/llvm/llvm-project/issues/56140. Reviewed By: Jean Perier Differential Revision: https://reviews.llvm.org/D128597
2022-06-30 23:52:44 +08:00
! CHECK: %[[VAL_84:.*]] = fir.embox %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_83]]] : (!fir.ref<!fir.array<2x?xf32>>, !fir.shape<2>, !fir.slice<2>) -> !fir.box<!fir.array<?x2xf32>>
! CHECK: %[[VAL_85:.*]] = fir.convert %[[VAL_84]] : (!fir.box<!fir.array<?x2xf32>>) -> !fir.box<none>
[flang][NFC] Add array lowering tests These tests were left behind during the upstreaming of parts lowering. This patch is part of the upstreaming effort from fir-dev branch. Reviewed By: jeanPerier Differential Revision: https://reviews.llvm.org/D128632 Co-authored-by: V Donaldson <vdonaldson@nvidia.com> Co-authored-by: Jean Perier <jperier@nvidia.com> Co-authored-by: Eric Schweitz <eschweitz@nvidia.com>
2022-06-27 20:19:23 +08:00
! CHECK: %[[VAL_86:.*]] = fir.call @_FortranAioOutputDescriptor(%[[VAL_82]], %[[VAL_85]]) : (!fir.ref<i8>, !fir.box<none>) -> i1
! CHECK: %[[VAL_87:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_88:.*]] = arith.subi %[[VAL_87]], %[[VAL_6]] : i32
! CHECK: %[[VAL_89:.*]] = fir.convert %[[VAL_88]] : (i32) -> index
! CHECK: %[[VAL_90:.*]] = fir.convert %[[VAL_87]] : (i32) -> index
! CHECK: %[[VAL_91:.*]] = fir.slice %[[VAL_3]], %[[VAL_2]], %[[VAL_3]], %[[VAL_89]], %[[VAL_90]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
! CHECK: %[[VAL_92:.*]] = fir.embox %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_91]]] : (!fir.ref<!fir.array<2x?xf32>>, !fir.shape<2>, !fir.slice<2>) -> !fir.box<!fir.array<?x?xf32>>
! CHECK: %[[VAL_93:.*]] = fir.convert %[[VAL_92]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
! CHECK: %[[VAL_94:.*]] = fir.call @_FortranAioOutputDescriptor(%[[VAL_82]], %[[VAL_93]]) : (!fir.ref<i8>, !fir.box<none>) -> i1
! CHECK: %[[VAL_95:.*]] = fir.call @_FortranAioEndIoStatement(%[[VAL_82]]) : (!fir.ref<i8>) -> i32
! CHECK: return
! CHECK: }
! CHECK-LABEL: func @_QPss4(
! CHECK-SAME: %[[VAL_0:.*]]: !fir.ref<i32> {fir.bindc_name = "n"}) {
! CHECK-DAG: %[[VAL_1:.*]] = arith.constant -1 : index
! CHECK-DAG: %[[VAL_2:.*]] = arith.constant 2 : index
! CHECK-DAG: %[[VAL_3:.*]] = arith.constant 1 : index
! CHECK-DAG: %[[VAL_4:.*]] = arith.constant -2 : i32
! CHECK-DAG: %[[VAL_5:.*]] = arith.constant 0 : index
! CHECK-DAG: %[[VAL_6:.*]] = arith.constant 1 : i32
! CHECK-DAG: %[[VAL_7:.*]] = arith.constant 7.000000e+00 : f32
! CHECK-DAG: %[[VAL_8:.*]] = arith.constant -1 : i32
! CHECK: %[[VAL_10:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_11A:.*]] = fir.convert %[[VAL_10]] : (i32) -> index
! CHECK: %[[CMP:.*]] = arith.cmpi sgt, %[[VAL_11A]], %[[VAL_5]] : index
! CHECK: %[[VAL_11:.*]] = arith.select %[[CMP]], %[[VAL_11A]], %[[VAL_5]] : index
! CHECK: %[[VAL_12:.*]] = fir.alloca !fir.array<?x2xf32>, %[[VAL_11]] {bindc_name = "aa", uniq_name = "_QFss4Eaa"}
! CHECK: %[[VAL_13:.*]] = fir.shape %[[VAL_11]], %[[VAL_2]] : (index, index) -> !fir.shape<2>
! CHECK: %[[VAL_14:.*]] = fir.convert %[[VAL_4]] : (i32) -> f32
! CHECK: br ^bb1(%[[VAL_5]], %[[VAL_2]] : index, index)
! CHECK: ^bb1(%[[VAL_15:.*]]: index, %[[VAL_16:.*]]: index):
! CHECK: %[[VAL_17:.*]] = arith.cmpi sgt, %[[VAL_16]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_17]], ^bb2(%[[VAL_5]], %[[VAL_11]] : index, index), ^bb5
! CHECK: ^bb2(%[[VAL_18:.*]]: index, %[[VAL_19:.*]]: index):
! CHECK: %[[VAL_20:.*]] = arith.cmpi sgt, %[[VAL_19]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_20]], ^bb3, ^bb4
! CHECK: ^bb3:
! CHECK: %[[VAL_21:.*]] = arith.addi %[[VAL_18]], %[[VAL_3]] : index
! CHECK: %[[VAL_22:.*]] = arith.addi %[[VAL_15]], %[[VAL_3]] : index
! CHECK: %[[VAL_23:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_21]], %[[VAL_22]] : (!fir.ref<!fir.array<?x2xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: fir.store %[[VAL_14]] to %[[VAL_23]] : !fir.ref<f32>
! CHECK: %[[VAL_24:.*]] = arith.subi %[[VAL_19]], %[[VAL_3]] : index
! CHECK: br ^bb2(%[[VAL_21]], %[[VAL_24]] : index, index)
! CHECK: ^bb4:
! CHECK: %[[VAL_25:.*]] = arith.addi %[[VAL_15]], %[[VAL_3]] : index
! CHECK: %[[VAL_26:.*]] = arith.subi %[[VAL_16]], %[[VAL_3]] : index
! CHECK: br ^bb1(%[[VAL_25]], %[[VAL_26]] : index, index)
! CHECK: ^bb5:
! CHECK: %[[VAL_27:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_28:.*]] = fir.convert %[[VAL_27]] : (i32) -> index
! CHECK: %[[VAL_29:.*]] = arith.addi %[[VAL_28]], %[[VAL_1]] : index
! CHECK: %[[VAL_30:.*]] = arith.cmpi sgt, %[[VAL_29]], %[[VAL_5]] : index
! CHECK: %[[VAL_31:.*]] = arith.select %[[VAL_30]], %[[VAL_29]], %[[VAL_5]] : index
! CHECK: %[[VAL_32:.*]] = fir.slice %[[VAL_2]], %[[VAL_28]], %[[VAL_3]], %[[VAL_3]], %[[VAL_2]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
! CHECK: %[[VAL_33:.*]] = fir.allocmem !fir.array<?x2xf32>, %[[VAL_11]]
! CHECK: br ^bb6(%[[VAL_5]], %[[VAL_2]] : index, index)
! CHECK: ^bb6(%[[VAL_34:.*]]: index, %[[VAL_35:.*]]: index):
! CHECK: %[[VAL_36:.*]] = arith.cmpi sgt, %[[VAL_35]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_36]], ^bb7(%[[VAL_5]], %[[VAL_11]] : index, index), ^bb10
! CHECK: ^bb7(%[[VAL_37:.*]]: index, %[[VAL_38:.*]]: index):
! CHECK: %[[VAL_39:.*]] = arith.cmpi sgt, %[[VAL_38]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_39]], ^bb8, ^bb9
! CHECK: ^bb8:
! CHECK: %[[VAL_40:.*]] = arith.addi %[[VAL_37]], %[[VAL_3]] : index
! CHECK: %[[VAL_41:.*]] = arith.addi %[[VAL_34]], %[[VAL_3]] : index
! CHECK: %[[VAL_42:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_40]], %[[VAL_41]] : (!fir.ref<!fir.array<?x2xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_43:.*]] = fir.array_coor %[[VAL_33]](%[[VAL_13]]) %[[VAL_40]], %[[VAL_41]] : (!fir.heap<!fir.array<?x2xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_44:.*]] = fir.load %[[VAL_42]] : !fir.ref<f32>
! CHECK: fir.store %[[VAL_44]] to %[[VAL_43]] : !fir.ref<f32>
! CHECK: %[[VAL_45:.*]] = arith.subi %[[VAL_38]], %[[VAL_3]] : index
! CHECK: br ^bb7(%[[VAL_40]], %[[VAL_45]] : index, index)
! CHECK: ^bb9:
! CHECK: %[[VAL_46:.*]] = arith.addi %[[VAL_34]], %[[VAL_3]] : index
! CHECK: %[[VAL_47:.*]] = arith.subi %[[VAL_35]], %[[VAL_3]] : index
! CHECK: br ^bb6(%[[VAL_46]], %[[VAL_47]] : index, index)
! CHECK: ^bb10:
! CHECK: %[[VAL_48:.*]] = arith.subi %[[VAL_27]], %[[VAL_6]] : i32
! CHECK: %[[VAL_49:.*]] = fir.convert %[[VAL_48]] : (i32) -> index
! CHECK: %[[VAL_50:.*]] = fir.slice %[[VAL_3]], %[[VAL_49]], %[[VAL_3]], %[[VAL_3]], %[[VAL_2]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
! CHECK: br ^bb11(%[[VAL_5]], %[[VAL_2]] : index, index)
! CHECK: ^bb11(%[[VAL_51:.*]]: index, %[[VAL_52:.*]]: index):
! CHECK: %[[VAL_53:.*]] = arith.cmpi sgt, %[[VAL_52]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_53]], ^bb12(%[[VAL_5]], %[[VAL_31]] : index, index), ^bb15(%[[VAL_5]], %[[VAL_2]] : index, index)
! CHECK: ^bb12(%[[VAL_54:.*]]: index, %[[VAL_55:.*]]: index):
! CHECK: %[[VAL_56:.*]] = arith.cmpi sgt, %[[VAL_55]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_56]], ^bb13, ^bb14
! CHECK: ^bb13:
! CHECK: %[[VAL_57:.*]] = arith.addi %[[VAL_54]], %[[VAL_3]] : index
! CHECK: %[[VAL_58:.*]] = arith.addi %[[VAL_51]], %[[VAL_3]] : index
! CHECK: %[[VAL_59:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_50]]] %[[VAL_57]], %[[VAL_58]] : (!fir.ref<!fir.array<?x2xf32>>, !fir.shape<2>, !fir.slice<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_60:.*]] = fir.load %[[VAL_59]] : !fir.ref<f32>
! CHECK: %[[VAL_61:.*]] = arith.addf %[[VAL_60]], %[[VAL_7]] : f32
! CHECK: %[[VAL_62:.*]] = fir.array_coor %[[VAL_33]](%[[VAL_13]]) {{\[}}%[[VAL_32]]] %[[VAL_57]], %[[VAL_58]] : (!fir.heap<!fir.array<?x2xf32>>, !fir.shape<2>, !fir.slice<2>, index, index) -> !fir.ref<f32>
! CHECK: fir.store %[[VAL_61]] to %[[VAL_62]] : !fir.ref<f32>
! CHECK: %[[VAL_63:.*]] = arith.subi %[[VAL_55]], %[[VAL_3]] : index
! CHECK: br ^bb12(%[[VAL_57]], %[[VAL_63]] : index, index)
! CHECK: ^bb14:
! CHECK: %[[VAL_64:.*]] = arith.addi %[[VAL_51]], %[[VAL_3]] : index
! CHECK: %[[VAL_65:.*]] = arith.subi %[[VAL_52]], %[[VAL_3]] : index
! CHECK: br ^bb11(%[[VAL_64]], %[[VAL_65]] : index, index)
! CHECK: ^bb15(%[[VAL_66:.*]]: index, %[[VAL_67:.*]]: index):
! CHECK: %[[VAL_68:.*]] = arith.cmpi sgt, %[[VAL_67]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_68]], ^bb16(%[[VAL_5]], %[[VAL_11]] : index, index), ^bb19
! CHECK: ^bb16(%[[VAL_69:.*]]: index, %[[VAL_70:.*]]: index):
! CHECK: %[[VAL_71:.*]] = arith.cmpi sgt, %[[VAL_70]], %[[VAL_5]] : index
! CHECK: cond_br %[[VAL_71]], ^bb17, ^bb18
! CHECK: ^bb17:
! CHECK: %[[VAL_72:.*]] = arith.addi %[[VAL_69]], %[[VAL_3]] : index
! CHECK: %[[VAL_73:.*]] = arith.addi %[[VAL_66]], %[[VAL_3]] : index
! CHECK: %[[VAL_74:.*]] = fir.array_coor %[[VAL_33]](%[[VAL_13]]) %[[VAL_72]], %[[VAL_73]] : (!fir.heap<!fir.array<?x2xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_75:.*]] = fir.array_coor %[[VAL_12]](%[[VAL_13]]) %[[VAL_72]], %[[VAL_73]] : (!fir.ref<!fir.array<?x2xf32>>, !fir.shape<2>, index, index) -> !fir.ref<f32>
! CHECK: %[[VAL_76:.*]] = fir.load %[[VAL_74]] : !fir.ref<f32>
! CHECK: fir.store %[[VAL_76]] to %[[VAL_75]] : !fir.ref<f32>
! CHECK: %[[VAL_77:.*]] = arith.subi %[[VAL_70]], %[[VAL_3]] : index
! CHECK: br ^bb16(%[[VAL_72]], %[[VAL_77]] : index, index)
! CHECK: ^bb18:
! CHECK: %[[VAL_78:.*]] = arith.addi %[[VAL_66]], %[[VAL_3]] : index
! CHECK: %[[VAL_79:.*]] = arith.subi %[[VAL_67]], %[[VAL_3]] : index
! CHECK: br ^bb15(%[[VAL_78]], %[[VAL_79]] : index, index)
! CHECK: ^bb19:
! CHECK: fir.freemem %[[VAL_33]] : !fir.heap<!fir.array<?x2xf32>>
! CHECK: %[[VAL_82:.*]] = fir.call @_FortranAioBeginExternalListOutput(%[[VAL_8]], %{{.*}}, %{{.*}}) : (i32, !fir.ref<i8>, i32) -> !fir.ref<i8>
! CHECK: %[[VAL_83:.*]] = fir.slice %[[VAL_3]], %[[VAL_2]], %[[VAL_3]], %[[VAL_3]], %[[VAL_2]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
[flang] Fix one corner case in reshape intrinsic As Fortran 2018 16.9.163, the reshape is the only intrinsic which requires the shape argument to be rank-one integer array and the SIZE of it to be one constant expression. The current expression lowering converts the shape expression with slice in intrinsic into one box value with the box element type of unknown extent. However, the genReshape requires the box element type to be constant size. So, convert the box value into one with box element type of sequence of 1 x constant. This corner case is found in cam4 in SPEC 2017 https://github.com/llvm/llvm-project/issues/56140. Reviewed By: Jean Perier Differential Revision: https://reviews.llvm.org/D128597
2022-06-30 23:52:44 +08:00
! CHECK: %[[VAL_84:.*]] = fir.embox %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_83]]] : (!fir.ref<!fir.array<?x2xf32>>, !fir.shape<2>, !fir.slice<2>) -> !fir.box<!fir.array<2x?xf32>>
! CHECK: %[[VAL_85:.*]] = fir.convert %[[VAL_84]] : (!fir.box<!fir.array<2x?xf32>>) -> !fir.box<none>
[flang][NFC] Add array lowering tests These tests were left behind during the upstreaming of parts lowering. This patch is part of the upstreaming effort from fir-dev branch. Reviewed By: jeanPerier Differential Revision: https://reviews.llvm.org/D128632 Co-authored-by: V Donaldson <vdonaldson@nvidia.com> Co-authored-by: Jean Perier <jperier@nvidia.com> Co-authored-by: Eric Schweitz <eschweitz@nvidia.com>
2022-06-27 20:19:23 +08:00
! CHECK: %[[VAL_86:.*]] = fir.call @_FortranAioOutputDescriptor(%[[VAL_82]], %[[VAL_85]]) : (!fir.ref<i8>, !fir.box<none>) -> i1
! CHECK: %[[VAL_87:.*]] = fir.load %[[VAL_0]] : !fir.ref<i32>
! CHECK: %[[VAL_88:.*]] = arith.subi %[[VAL_87]], %[[VAL_6]] : i32
! CHECK: %[[VAL_89:.*]] = fir.convert %[[VAL_88]] : (i32) -> index
! CHECK: %[[VAL_90:.*]] = fir.convert %[[VAL_87]] : (i32) -> index
! CHECK: %[[VAL_91:.*]] = fir.slice %[[VAL_89]], %[[VAL_90]], %[[VAL_3]], %[[VAL_3]], %[[VAL_2]], %[[VAL_3]] : (index, index, index, index, index, index) -> !fir.slice<2>
! CHECK: %[[VAL_92:.*]] = fir.embox %[[VAL_12]](%[[VAL_13]]) {{\[}}%[[VAL_91]]] : (!fir.ref<!fir.array<?x2xf32>>, !fir.shape<2>, !fir.slice<2>) -> !fir.box<!fir.array<?x?xf32>>
! CHECK: %[[VAL_93:.*]] = fir.convert %[[VAL_92]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
! CHECK: %[[VAL_94:.*]] = fir.call @_FortranAioOutputDescriptor(%[[VAL_82]], %[[VAL_93]]) : (!fir.ref<i8>, !fir.box<none>) -> i1
! CHECK: %[[VAL_95:.*]] = fir.call @_FortranAioEndIoStatement(%[[VAL_82]]) : (!fir.ref<i8>) -> i32
! CHECK: return
! CHECK: }
! CHECK-LABEL: func @_QPtt1
subroutine tt1
! CHECK: fir.call @_FortranAioBeginExternalListOutput
! CHECK: %[[temp3:[0-9]+]] = fir.allocmem !fir.array<3xf32>
! CHECK: br ^bb1(%[[temp3]]
! CHECK-NEXT: ^bb1(%[[temp3arg:[0-9]+]]: !fir.heap<!fir.array<3xf32>>
! CHECK: %[[temp1:[0-9]+]] = fir.allocmem !fir.array<1xf32>
! CHECK: fir.call @_QFtt1Pr
! CHECK: fir.call @realloc
! CHECK: fir.freemem %[[temp1]] : !fir.heap<!fir.array<1xf32>>
! CHECK: %[[temp3x:[0-9]+]] = fir.allocmem !fir.array<3xf32>
! CHECK: fir.call @_FortranAioOutputDescriptor
! CHECK-NEXT: fir.freemem %[[temp3x]] : !fir.heap<!fir.array<3xf32>>
! CHECK-NEXT: fir.freemem %[[temp3arg]] : !fir.heap<!fir.array<3xf32>>
! CHECK-NEXT: fir.call @_FortranAioEndIoStatement
print*, [(r([7.0]),i=1,3)]
contains
! CHECK-LABEL: func @_QFtt1Pr
function r(x)
real x(:)
r = x(1)
end
end