| 1 |
// Test array-copy-value pass (copy elision) with fir.array_modify |
| 2 |
// RUN: fir-opt %s --array-value-copy | FileCheck %s |
| 3 |
// RUN: fir-opt %s --array-value-copy="optimize-conflicts=true" | FileCheck %s |
| 4 |
|
| 5 |
// Test user_defined_assignment(arg0(:), arg1(:)) |
| 6 |
func.func @no_overlap(%arg0: !fir.ref<!fir.array<100xf32>>, %arg1: !fir.ref<!fir.array<100xf32>>) { |
| 7 |
%c100 = arith.constant 100 : index |
| 8 |
%c99 = arith.constant 99 : index |
| 9 |
%c1 = arith.constant 1 : index |
| 10 |
%c0 = arith.constant 0 : index |
| 11 |
%0 = fir.alloca f32 |
| 12 |
%1 = fir.shape %c100 : (index) -> !fir.shape<1> |
| 13 |
%2 = fir.array_load %arg0(%1) : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>) -> !fir.array<100xf32> |
| 14 |
%3 = fir.array_load %arg1(%1) : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>) -> !fir.array<100xf32> |
| 15 |
%4 = fir.do_loop %arg2 = %c0 to %c99 step %c1 unordered iter_args(%arg3 = %2) -> (!fir.array<100xf32>) { |
| 16 |
%5 = fir.array_fetch %3, %arg2 : (!fir.array<100xf32>, index) -> f32 |
| 17 |
%6:2 = fir.array_modify %arg3, %arg2 : (!fir.array<100xf32>, index) -> (!fir.ref<f32>, !fir.array<100xf32>) |
| 18 |
fir.store %5 to %0 : !fir.ref<f32> |
| 19 |
fir.call @user_defined_assignment(%6#0, %0) : (!fir.ref<f32>, !fir.ref<f32>) -> () |
| 20 |
fir.result %6#1 : !fir.array<100xf32> |
| 21 |
} |
| 22 |
fir.array_merge_store %2, %4 to %arg0 : !fir.array<100xf32>, !fir.array<100xf32>, !fir.ref<!fir.array<100xf32>> |
| 23 |
return |
| 24 |
} |
| 25 |
// CHECK-LABEL: func @no_overlap( |
| 26 |
// CHECK-SAME: %[[VAL_0:.*]]: !fir.ref<!fir.array<100xf32>>, |
| 27 |
// CHECK-SAME: %[[VAL_1:.*]]: !fir.ref<!fir.array<100xf32>>) { |
| 28 |
// CHECK-DAG: %[[VAL_2:.*]] = arith.constant 100 : index |
| 29 |
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 99 : index |
| 30 |
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 1 : index |
| 31 |
// CHECK-DAG: %[[VAL_5:.*]] = arith.constant 0 : index |
| 32 |
// CHECK: %[[VAL_6:.*]] = fir.alloca f32 |
| 33 |
// CHECK: %[[VAL_7:.*]] = fir.shape %[[VAL_2]] : (index) -> !fir.shape<1> |
| 34 |
// CHECK: %[[VAL_8:.*]] = fir.undefined !fir.array<100xf32> |
| 35 |
// CHECK: %[[VAL_9:.*]] = fir.undefined !fir.array<100xf32> |
| 36 |
// CHECK: %[[VAL_10:.*]] = fir.do_loop %[[VAL_11:.*]] = %[[VAL_5]] to %[[VAL_3]] step %[[VAL_4]] unordered iter_args(%[[VAL_12:.*]] = %[[VAL_8]]) -> (!fir.array<100xf32>) { |
| 37 |
// CHECK: %[[VAL_13:.*]] = arith.constant 1 : index |
| 38 |
// CHECK: %[[VAL_14:.*]] = arith.addi %[[VAL_11]], %[[VAL_13]] : index |
| 39 |
// CHECK: %[[VAL_15:.*]] = fir.array_coor %[[VAL_1]](%[[VAL_7]]) %[[VAL_14]] : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>, index) -> !fir.ref<f32> |
| 40 |
// CHECK: %[[VAL_16:.*]] = fir.load %[[VAL_15]] : !fir.ref<f32> |
| 41 |
// CHECK: %[[VAL_17:.*]] = arith.constant 1 : index |
| 42 |
// CHECK: %[[VAL_18:.*]] = arith.addi %[[VAL_11]], %[[VAL_17]] : index |
| 43 |
// CHECK: %[[VAL_19:.*]] = fir.array_coor %[[VAL_0]](%[[VAL_7]]) %[[VAL_18]] : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>, index) -> !fir.ref<f32> |
| 44 |
// CHECK: fir.store %[[VAL_16]] to %[[VAL_6]] : !fir.ref<f32> |
| 45 |
// CHECK: fir.call @user_defined_assignment(%[[VAL_19]], %[[VAL_6]]) : (!fir.ref<f32>, !fir.ref<f32>) -> () |
| 46 |
// CHECK: fir.result %[[VAL_8]] : !fir.array<100xf32> |
| 47 |
// CHECK: } |
| 48 |
// CHECK: return |
| 49 |
// CHECK: } |
| 50 |
|
| 51 |
|
| 52 |
// Test user_defined_assignment(arg0(:), arg0(100:1:-1)) |
| 53 |
func.func @overlap(%arg0: !fir.ref<!fir.array<100xf32>>) { |
| 54 |
%c100 = arith.constant 100 : index |
| 55 |
%c99 = arith.constant 99 : index |
| 56 |
%c1 = arith.constant 1 : index |
| 57 |
%c-1 = arith.constant -1 : index |
| 58 |
%c0 = arith.constant 0 : index |
| 59 |
%0 = fir.alloca f32 |
| 60 |
%1 = fir.shape %c100 : (index) -> !fir.shape<1> |
| 61 |
%2 = fir.array_load %arg0(%1) : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>) -> !fir.array<100xf32> |
| 62 |
%3 = fir.slice %c100, %c1, %c-1 : (index, index, index) -> !fir.slice<1> |
| 63 |
%4 = fir.array_load %arg0(%1) [%3] : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>, !fir.slice<1>) -> !fir.array<100xf32> |
| 64 |
%5 = fir.do_loop %arg1 = %c0 to %c99 step %c1 unordered iter_args(%arg2 = %2) -> (!fir.array<100xf32>) { |
| 65 |
%6 = fir.array_fetch %4, %arg1 : (!fir.array<100xf32>, index) -> f32 |
| 66 |
%7:2 = fir.array_modify %arg2, %arg1 : (!fir.array<100xf32>, index) -> (!fir.ref<f32>, !fir.array<100xf32>) |
| 67 |
fir.store %6 to %0 : !fir.ref<f32> |
| 68 |
fir.call @user_defined_assignment(%7#0, %0) : (!fir.ref<f32>, !fir.ref<f32>) -> () |
| 69 |
fir.result %7#1 : !fir.array<100xf32> |
| 70 |
} |
| 71 |
fir.array_merge_store %2, %5 to %arg0 : !fir.array<100xf32>, !fir.array<100xf32>, !fir.ref<!fir.array<100xf32>> |
| 72 |
return |
| 73 |
} |
| 74 |
// CHECK-LABEL: func @overlap( |
| 75 |
// CHECK-SAME: %[[VAL_0:.*]]: !fir.ref<!fir.array<100xf32>>) { |
| 76 |
// CHECK-DAG: %[[VAL_1:.*]] = arith.constant 100 : index |
| 77 |
// CHECK-DAG: %[[VAL_2:.*]] = arith.constant 99 : index |
| 78 |
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 1 : index |
| 79 |
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant -1 : index |
| 80 |
// CHECK-DAG: %[[VAL_5:.*]] = arith.constant 0 : index |
| 81 |
// CHECK: %[[VAL_6:.*]] = fir.alloca f32 |
| 82 |
// CHECK: %[[VAL_7:.*]] = fir.shape %[[VAL_1]] : (index) -> !fir.shape<1> |
| 83 |
// CHECK: %[[VAL_8:.*]] = fir.allocmem !fir.array<100xf32>{{$}} |
| 84 |
// CHECK: %[[VAL_9:.*]] = fir.convert %[[VAL_1]] : (index) -> index |
| 85 |
// CHECK: %[[VAL_10:.*]] = arith.constant 0 : index |
| 86 |
// CHECK: %[[VAL_11:.*]] = arith.constant 1 : index |
| 87 |
// CHECK: %[[VAL_12:.*]] = arith.subi %[[VAL_9]], %[[VAL_11]] : index |
| 88 |
// CHECK: fir.do_loop %[[VAL_13:.*]] = %[[VAL_10]] to %[[VAL_12]] step %[[VAL_11]] { |
| 89 |
// CHECK: %[[VAL_14:.*]] = arith.constant 1 : index |
| 90 |
// CHECK: %[[VAL_15:.*]] = arith.addi %[[VAL_13]], %[[VAL_14]] : index |
| 91 |
// CHECK: %[[VAL_16:.*]] = fir.array_coor %[[VAL_0]](%[[VAL_7]]) %[[VAL_15]] : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>, index) -> !fir.ref<f32> |
| 92 |
// CHECK: %[[VAL_18:.*]] = arith.constant 1 : index |
| 93 |
// CHECK: %[[VAL_19:.*]] = arith.addi %[[VAL_13]], %[[VAL_18]] : index |
| 94 |
// CHECK: %[[VAL_20:.*]] = fir.array_coor %[[VAL_8]](%[[VAL_7]]) %[[VAL_19]] : (!fir.heap<!fir.array<100xf32>>, !fir.shape<1>, index) -> !fir.ref<f32> |
| 95 |
// CHECK: %[[VAL_17:.*]] = fir.load %[[VAL_16]] : !fir.ref<f32> |
| 96 |
// CHECK: fir.store %[[VAL_17]] to %[[VAL_20]] : !fir.ref<f32> |
| 97 |
// CHECK: } |
| 98 |
// CHECK: %[[VAL_21:.*]] = fir.undefined !fir.array<100xf32> |
| 99 |
// CHECK: %[[VAL_22:.*]] = fir.slice %[[VAL_1]], %[[VAL_3]], %[[VAL_4]] : (index, index, index) -> !fir.slice<1> |
| 100 |
// CHECK: %[[VAL_23:.*]] = fir.undefined !fir.array<100xf32> |
| 101 |
// CHECK: %[[VAL_24:.*]] = fir.do_loop %[[VAL_25:.*]] = %[[VAL_5]] to %[[VAL_2]] step %[[VAL_3]] unordered iter_args(%[[VAL_26:.*]] = %[[VAL_21]]) -> (!fir.array<100xf32>) { |
| 102 |
// CHECK: %[[VAL_27:.*]] = arith.constant 1 : index |
| 103 |
// CHECK: %[[VAL_28:.*]] = arith.addi %[[VAL_25]], %[[VAL_27]] : index |
| 104 |
// CHECK: %[[VAL_29:.*]] = fir.array_coor %[[VAL_0]](%[[VAL_7]]) {{\[}}%[[VAL_22]]] %[[VAL_28]] : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>, !fir.slice<1>, index) -> !fir.ref<f32> |
| 105 |
// CHECK: %[[VAL_30:.*]] = fir.load %[[VAL_29]] : !fir.ref<f32> |
| 106 |
// CHECK: %[[VAL_31:.*]] = arith.constant 1 : index |
| 107 |
// CHECK: %[[VAL_32:.*]] = arith.addi %[[VAL_25]], %[[VAL_31]] : index |
| 108 |
// CHECK: %[[VAL_33:.*]] = fir.array_coor %[[VAL_8]](%[[VAL_7]]) %[[VAL_32]] : (!fir.heap<!fir.array<100xf32>>, !fir.shape<1>, index) -> !fir.ref<f32> |
| 109 |
// CHECK: fir.store %[[VAL_30]] to %[[VAL_6]] : !fir.ref<f32> |
| 110 |
// CHECK: fir.call @user_defined_assignment(%[[VAL_33]], %[[VAL_6]]) : (!fir.ref<f32>, !fir.ref<f32>) -> () |
| 111 |
// CHECK: fir.result %[[VAL_21]] : !fir.array<100xf32> |
| 112 |
// CHECK: } |
| 113 |
// CHECK: %[[VAL_34:.*]] = fir.convert %[[VAL_1]] : (index) -> index |
| 114 |
// CHECK: %[[VAL_35:.*]] = arith.constant 0 : index |
| 115 |
// CHECK: %[[VAL_36:.*]] = arith.constant 1 : index |
| 116 |
// CHECK: %[[VAL_37:.*]] = arith.subi %[[VAL_34]], %[[VAL_36]] : index |
| 117 |
// CHECK: fir.do_loop %[[VAL_38:.*]] = %[[VAL_35]] to %[[VAL_37]] step %[[VAL_36]] { |
| 118 |
// CHECK: %[[VAL_39:.*]] = arith.constant 1 : index |
| 119 |
// CHECK: %[[VAL_40:.*]] = arith.addi %[[VAL_38]], %[[VAL_39]] : index |
| 120 |
// CHECK: %[[VAL_41:.*]] = fir.array_coor %[[VAL_8]](%[[VAL_7]]) %[[VAL_40]] : (!fir.heap<!fir.array<100xf32>>, !fir.shape<1>, index) -> !fir.ref<f32> |
| 121 |
// CHECK: %[[VAL_43:.*]] = arith.constant 1 : index |
| 122 |
// CHECK: %[[VAL_44:.*]] = arith.addi %[[VAL_38]], %[[VAL_43]] : index |
| 123 |
// CHECK: %[[VAL_45:.*]] = fir.array_coor %[[VAL_0]](%[[VAL_7]]) %[[VAL_44]] : (!fir.ref<!fir.array<100xf32>>, !fir.shape<1>, index) -> !fir.ref<f32> |
| 124 |
// CHECK: %[[VAL_42:.*]] = fir.load %[[VAL_41]] : !fir.ref<f32> |
| 125 |
// CHECK: fir.store %[[VAL_42]] to %[[VAL_45]] : !fir.ref<f32> |
| 126 |
// CHECK: } |
| 127 |
// CHECK: fir.freemem %[[VAL_8]] : !fir.heap<!fir.array<100xf32>> |
| 128 |
// CHECK: return |
| 129 |
// CHECK: } |
| 130 |
|
| 131 |
func.func private @user_defined_assignment(!fir.ref<f32>, !fir.ref<f32>) |