[Mlir-commits] [flang] [mlir] [MLIR][OpenMP] Fix type mismatch in linear clause for INTEGER(8) variables (PR #173982)

[llvmlistbot at llvm.org]

================
@@ -0,0 +1,79 @@
+! Test lowering of OpenMP parallel do simd with linear clause on INTEGER(8) variable
+! This is a regression test for issue #173332
+! At MLIR level, the step type may differ from the variable type - this is legal and handled during translation
+
+! RUN: %flang_fc1 -fopenmp -emit-hlfir %s -o - 2>&1 | FileCheck %s
+
+! CHECK-LABEL: func.func @_QPtest_linear_i8
+subroutine test_linear_i8
+    implicit none
+    integer(8) :: i, j
+    
+    ! CHECK-DAG: %[[J_alloca:.*]] = fir.alloca i64 {bindc_name = "j"
+    ! CHECK-DAG: %[[J:.*]]:2 = hlfir.declare %[[J_alloca]]
+    ! CHECK-DAG: %[[STEP:.*]] = arith.constant 1 : i32
+    ! CHECK: omp.wsloop
+    ! CHECK: omp.simd linear(%[[J]]#0 = %[[STEP]] : !fir.ref<i64>)
+    !$omp parallel do simd linear(j)
+    do i = 1,100,1
+    end do
+    !$omp end parallel do simd
+end subroutine
+
+! CHECK-LABEL: func.func @_QPtest_linear_i8_with_step
+subroutine test_linear_i8_with_step
+    implicit none
+    integer(8) :: i, j
+    
+    ! CHECK-DAG: %[[J:.*]]:2 = hlfir.declare {{.*}} {uniq_name = "_QFtest_linear_i8_with_stepEj"}
+    ! CHECK: omp.wsloop
+    ! CHECK: omp.simd linear(%[[J]]#0 = {{.*}} : !fir.ref<i64>)
+    !$omp parallel do simd linear(j:2)
+    do i = 1,100,1
+    end do
+    !$omp end parallel do simd
+end subroutine
+
+! CHECK-LABEL: func.func @_QPtest_simd_linear_i8
+subroutine test_simd_linear_i8
+    implicit none
+    integer(8) :: i, j
+    
+    ! CHECK-DAG: %[[J:.*]]:2 = hlfir.declare {{.*}} {uniq_name = "_QFtest_simd_linear_i8Ej"}
+    ! CHECK-DAG: %[[STEP:.*]] = arith.constant 1 : i32
+    ! CHECK: omp.simd linear(%[[J]]#0 = %[[STEP]] : !fir.ref<i64>)
+    !$omp simd linear(j)
+    do i = 1,100,1
+    end do
+    !$omp end simd
+end subroutine
+
+! CHECK-LABEL: func.func @_QPtest_do_linear_i8
+subroutine test_do_linear_i8
+    implicit none
+    integer(8) :: i, j
+    
+    ! CHECK-DAG: %[[J:.*]]:2 = hlfir.declare {{.*}} {uniq_name = "_QFtest_do_linear_i8Ej"}
+    ! CHECK-DAG: %[[STEP:.*]] = arith.constant 1 : i32
+    ! CHECK: omp.wsloop linear(%[[J]]#0 = %[[STEP]] : !fir.ref<i64>)
+    !$omp do linear(j)
+    do i = 1,100,1
+    end do
+    !$omp end do
+end subroutine
+
+! Test with multiple INTEGER(8) linear variables
+! CHECK-LABEL: func.func @_QPtest_multiple_i8
+subroutine test_multiple_i8
+    implicit none
+    integer(8) :: i, j, k
+    
+    ! CHECK-DAG: %[[J:.*]]:2 = hlfir.declare {{.*}} {uniq_name = "_QFtest_multiple_i8Ej"}
+    ! CHECK-DAG: %[[K:.*]]:2 = hlfir.declare {{.*}} {uniq_name = "_QFtest_multiple_i8Ek"}
+    ! CHECK-DAG: %[[STEP:.*]] = arith.constant 1 : i32
+    ! CHECK: omp.simd linear(%[[J]]#0 = %[[STEP]] : !fir.ref<i64>, %[[K]]#0 = %[[STEP]] : !fir.ref<i64>)
----------------
NimishMishra wrote:

Is there something specific you are checking during FIR lowering? Since the truncation / sign-ext is happening during translation of the MLIR op, it is best to have MLIR tests. I think we can do away with this test explicitly.

https://github.com/llvm/llvm-project/pull/173982