16 files changed, 791 insertions, 135 deletions

diff --git a/flang/test/Examples/omp-atomic.f90 b/flang/test/Examples/omp-atomic.f90
index 934f84f..5695b62 100644
--- a/flang/test/Examples/omp-atomic.f90
+++ b/flang/test/Examples/omp-atomic.f90
@@ -31,13 +31,13 @@ end
 ! CHECK-NEXT:    - clause:      read
 ! CHECK-NEXT:      details:     ''
 ! CHECK-NEXT:    - clause:      seq_cst
-! CHECK-NEXT:      details:     'name_modifier=atomic;'
+! CHECK-NEXT:      details:     ''
 ! CHECK-NEXT:- file:            '{{[^"]*}}omp-atomic.f90'
 ! CHECK-NEXT:  line:            12
 ! CHECK-NEXT:  construct:       atomic
 ! CHECK-NEXT:  clauses:
 ! CHECK-NEXT:    - clause:      seq_cst
-! CHECK-NEXT:      details:     'name_modifier=atomic;'
+! CHECK-NEXT:      details:     ''
 ! CHECK-NEXT:    - clause:      write
 ! CHECK-NEXT:      details:     ''
 ! CHECK-NEXT:- file:            '{{[^"]*}}omp-atomic.f90'
@@ -45,7 +45,7 @@ end
 ! CHECK-NEXT:  construct:       atomic
 ! CHECK-NEXT:  clauses:
 ! CHECK-NEXT:    - clause:      capture
-! CHECK-NEXT:      details:     'name_modifier=atomic;name_modifier=atomic;'
+! CHECK-NEXT:      details:     ''
 ! CHECK-NEXT:    - clause:      seq_cst
 ! CHECK-NEXT:      details:     ''
 ! CHECK-NEXT:- file:            '{{[^"]*}}omp-atomic.f90'
diff --git a/flang/test/Examples/omp-sections.f90 b/flang/test/Examples/omp-sections.f90
index 41e6e8f..a6d2806 100644
--- a/flang/test/Examples/omp-sections.f90
+++ b/flang/test/Examples/omp-sections.f90
@@ -13,11 +13,11 @@ subroutine omp_sections()
 end subroutine omp_sections
 
 !CHECK: - file:            {{.*}}
-!CHECK:   line:            9
+!CHECK:   line:            8
 !CHECK:   construct:       section
 !CHECK:   clauses:         []
 !CHECK: - file:            {{.*}}
-!CHECK:   line:            11
+!CHECK:   line:            10
 !CHECK:   construct:       section
 !CHECK:   clauses:         []
 !CHECK: - file:            {{.*}}
diff --git a/flang/test/Lower/OpenACC/Todo/do-loops-to-acc-loops-todo.f90 b/flang/test/Lower/OpenACC/Todo/do-loops-to-acc-loops-todo.f90
new file mode 100644
index 0000000..aa1d443
--- /dev/null
+++ b/flang/test/Lower/OpenACC/Todo/do-loops-to-acc-loops-todo.f90
@@ -0,0 +1,91 @@
+! RUN: split-file %s %t
+! RUN: %not_todo_cmd bbc -fopenacc -emit-hlfir %t/do_loop_with_stop.f90 -o - 2>&1 | FileCheck %s --check-prefix=CHECK1
+! RUN: %not_todo_cmd bbc -fopenacc -emit-hlfir %t/do_loop_with_cycle_goto.f90 -o - 2>&1 | FileCheck %s --check-prefix=CHECK2
+! RUN: %not_todo_cmd bbc -fopenacc -emit-hlfir %t/nested_goto_loop.f90 -o - 2>&1 | FileCheck %s --check-prefix=CHECK3
+! RUN: %not_todo_cmd bbc -fopenacc -emit-hlfir %t/nested_loop_with_inner_goto.f90 -o - 2>&1 | FileCheck %s --check-prefix=CHECK4
+
+//--- do_loop_with_stop.f90
+
+subroutine do_loop_with_stop()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  !$acc kernels
+  do i = 1, n
+    a(i) = b(i) + 1.0
+    if (i == 5) stop
+  end do
+  !$acc end kernels
+
+! CHECK1: not yet implemented: unstructured do loop in acc kernels
+
+end subroutine
+
+//--- do_loop_with_cycle_goto.f90
+
+subroutine do_loop_with_cycle_goto()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  ! Do loop with cycle and goto - unstructured control flow is not converted.
+  !$acc kernels
+  do i = 1, n
+    if (i == 3) cycle
+    a(i) = b(i) + 1.0
+    if (i == 7) goto 200
+    a(i) = a(i) * 2.0
+  end do
+200 continue
+  !$acc end kernels
+
+! CHECK2: not yet implemented: unstructured do loop in acc kernels
+
+end subroutine
+
+//--- nested_goto_loop.f90
+
+subroutine nested_goto_loop()
+  integer :: i, j
+  integer, parameter :: n = 10, m = 5
+  real, dimension(n,m) :: a, b
+
+  ! Nested loop with goto from inner to outer - should NOT convert to acc.loop
+  !$acc kernels
+  do i = 1, n
+    do j = 1, m
+      a(i,j) = b(i,j) + 1.0
+      if (i * j > 20) goto 300  ! Exit both loops
+    end do
+  end do
+300 continue
+  !$acc end kernels
+
+! CHECK3: not yet implemented: unstructured do loop in acc kernels
+
+end subroutine
+
+//--- nested_loop_with_inner_goto.f90
+
+subroutine nested_loop_with_inner_goto()
+  integer :: ii = 0, jj = 0
+  integer, parameter :: nn = 3
+  real, dimension(nn, nn) :: aa
+  
+  aa = -1
+  
+  ! Nested loop with goto from inner loop - unstructured control flow is not converted.
+  !$acc kernels
+  do ii = 1, nn
+    do jj = 1, nn
+      if (jj > 1) goto 300
+      aa(jj, ii) = 1337
+    end do
+    300 continue
+  end do
+  !$acc end kernels
+
+! CHECK4: not yet implemented: unstructured do loop in acc kernels
+
+end subroutine
\ No newline at end of file
diff --git a/flang/test/Lower/OpenACC/acc-atomic-capture.f90 b/flang/test/Lower/OpenACC/acc-atomic-capture.f90
index ee38ab6..30e60e3 100644
--- a/flang/test/Lower/OpenACC/acc-atomic-capture.f90
+++ b/flang/test/Lower/OpenACC/acc-atomic-capture.f90
@@ -123,17 +123,20 @@ end subroutine
 ! CHECK: }
 
 subroutine capture_with_convert_i32_to_f64()
-  real(8) :: x
-  integer :: v
+  real(8) :: x 
+  integer :: v, u 
   x = 1.0
   v = 0
+  u = 1
   !$acc atomic capture
   v = x
-  x = v
+  x = u
   !$acc end atomic
 end subroutine capture_with_convert_i32_to_f64
 
 ! CHECK-LABEL: func.func @_QPcapture_with_convert_i32_to_f64()
+! CHECK: %[[U:.*]] = fir.alloca i32 {bindc_name = "u", uniq_name = "_QFcapture_with_convert_i32_to_f64Eu"}
+! CHECK: %[[U_DECL:.*]]:2 = hlfir.declare %[[U]] {uniq_name = "_QFcapture_with_convert_i32_to_f64Eu"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
 ! CHECK: %[[V:.*]] = fir.alloca i32 {bindc_name = "v", uniq_name = "_QFcapture_with_convert_i32_to_f64Ev"}
 ! CHECK: %[[V_DECL:.*]]:2 = hlfir.declare %[[V]] {uniq_name = "_QFcapture_with_convert_i32_to_f64Ev"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
 ! CHECK: %[[X:.*]] = fir.alloca f64 {bindc_name = "x", uniq_name = "_QFcapture_with_convert_i32_to_f64Ex"}
@@ -142,7 +145,9 @@ end subroutine capture_with_convert_i32_to_f64
 ! CHECK: hlfir.assign %[[CST]] to %[[X_DECL]]#0 : f64, !fir.ref<f64>
 ! CHECK: %c0_i32 = arith.constant 0 : i32
 ! CHECK: hlfir.assign %c0_i32 to %[[V_DECL]]#0 : i32, !fir.ref<i32>
-! CHECK: %[[LOAD:.*]] = fir.load %[[V_DECL]]#0 : !fir.ref<i32>
+! CHECK: %c1_i32 = arith.constant 1 : i32
+! CHECK: hlfir.assign %c1_i32 to %[[U_DECL]]#0 : i32, !fir.ref<i32>
+! CHECK: %[[LOAD:.*]] = fir.load %[[U_DECL]]#0 : !fir.ref<i32>
 ! CHECK: %[[CONV:.*]] = fir.convert %[[LOAD]] : (i32) -> f64
 ! CHECK: acc.atomic.capture {
 ! CHECK:   acc.atomic.read %[[V_DECL]]#0 = %[[X_DECL]]#0 : !fir.ref<i32>, !fir.ref<f64>, f64
@@ -155,7 +160,7 @@ subroutine capture_with_convert_f64_to_i32()
   x = 1
   v = 0
   !$acc atomic capture
-  x = v * v
+  x = x * 2.0_8
   v = x
   !$acc end atomic
 end subroutine capture_with_convert_f64_to_i32
@@ -167,15 +172,14 @@ end subroutine capture_with_convert_f64_to_i32
 ! CHECK: %[[X_DECL:.*]]:2 = hlfir.declare %[[X]] {uniq_name = "_QFcapture_with_convert_f64_to_i32Ex"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
 ! CHECK: %c1_i32 = arith.constant 1 : i32
 ! CHECK: hlfir.assign %c1_i32 to %[[X_DECL]]#0 : i32, !fir.ref<i32>
-! CHECK: %[[CST:.*]] = arith.constant 0.000000e+00 : f64
-! CHECK: hlfir.assign %[[CST]] to %[[V_DECL]]#0 : f64, !fir.ref<f64>
-! CHECK: %[[LOAD:.*]] = fir.load %[[V_DECL]]#0 : !fir.ref<f64>
+! CHECK: %[[CST:.*]] = arith.constant 2.000000e+00 : f64
 ! CHECK: acc.atomic.capture {
 ! CHECK:   acc.atomic.update %[[X_DECL]]#0 : !fir.ref<i32> {
-! CHECK:   ^bb0(%arg0: i32):
-! CHECK:     %[[MUL:.*]] = arith.mulf %[[LOAD]], %[[LOAD]] fastmath<contract> : f64
-! CHECK:     %[[CONV:.*]] = fir.convert %[[MUL]] : (f64) -> i32
-! CHECK:     acc.yield %[[CONV]] : i32
+! CHECK:   ^bb0(%[[ARG:.*]]: i32):
+! CHECK:     %[[CONV_ARG:.*]] = fir.convert %[[ARG]] : (i32) -> f64
+! CHECK:     %[[MUL:.*]] = arith.mulf %[[CONV_ARG]], %[[CST]] fastmath<contract> : f64
+! CHECK:     %[[CONV_MUL:.*]] = fir.convert %[[MUL]] : (f64) -> i32
+! CHECK:     acc.yield %[[CONV_MUL]] : i32
 ! CHECK:   }
 ! CHECK:   acc.atomic.read %[[V_DECL]]#0 = %[[X_DECL]]#0 : !fir.ref<f64>, !fir.ref<i32>, i32
 ! CHECK: }
diff --git a/flang/test/Lower/OpenACC/do-loops-to-acc-loops.f90 b/flang/test/Lower/OpenACC/do-loops-to-acc-loops.f90
new file mode 100644
index 0000000..5f8ea03
--- /dev/null
+++ b/flang/test/Lower/OpenACC/do-loops-to-acc-loops.f90
@@ -0,0 +1,267 @@
+! This test checks lowering of Fortran do loops and do concurrent loops to OpenACC loop constructs.
+! Tests the new functionality that converts Fortran iteration constructs to acc.loop with proper IV handling.
+
+! RUN: bbc -fopenacc -emit-hlfir %s -o - | FileCheck %s
+
+! CHECK-LABEL: func.func @_QPbasic_do_loop
+subroutine basic_do_loop()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  ! Basic do loop that should be converted to acc.loop
+  !$acc kernels
+  do i = 1, n
+    a(i) = b(i) + 1.0
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {auto_ = [#acc.device_type<none>], inclusiveUpperbound = array<i1: true>}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPbasic_do_concurrent
+subroutine basic_do_concurrent()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  ! Basic do concurrent loop
+  !$acc kernels
+  do concurrent (i = 1:n)
+    a(i) = b(i) + 1.0
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {auto_ = [#acc.device_type<none>], inclusiveUpperbound = array<i1: true>}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPbasic_do_loop_parallel
+subroutine basic_do_loop_parallel()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  ! Basic do loop with acc parallel that should be converted to acc.loop
+  !$acc parallel
+  do i = 1, n
+    a(i) = b(i) + 1.0
+  end do
+  !$acc end parallel
+
+! CHECK: acc.parallel {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {auto_ = [#acc.device_type<none>], inclusiveUpperbound = array<i1: true>}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPbasic_do_loop_serial
+subroutine basic_do_loop_serial()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  ! Basic do loop with acc serial that should be converted to acc.loop
+  !$acc serial
+  do i = 1, n
+    a(i) = b(i) + 1.0
+  end do
+  !$acc end serial
+
+! CHECK: acc.serial {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {inclusiveUpperbound = array<i1: true>, seq = [#acc.device_type<none>]}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPbasic_do_concurrent_parallel
+subroutine basic_do_concurrent_parallel()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  ! Basic do concurrent loop with acc parallel
+  !$acc parallel
+  do concurrent (i = 1:n)
+    a(i) = b(i) + 1.0
+  end do
+  !$acc end parallel
+
+! CHECK: acc.parallel {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {inclusiveUpperbound = array<i1: true>, independent = [#acc.device_type<none>]}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPbasic_do_concurrent_serial
+subroutine basic_do_concurrent_serial()
+  integer :: i
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b
+
+  ! Basic do concurrent loop with acc serial
+  !$acc serial
+  do concurrent (i = 1:n)
+    a(i) = b(i) + 1.0
+  end do
+  !$acc end serial
+
+! CHECK: acc.serial {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {inclusiveUpperbound = array<i1: true>, seq = [#acc.device_type<none>]}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPmulti_dimension_do_concurrent
+subroutine multi_dimension_do_concurrent()
+  integer :: i, j, k
+  integer, parameter :: n = 10, m = 20, l = 5
+  real, dimension(n,m,l) :: a, b
+
+  ! Multi-dimensional do concurrent with multiple iteration variables
+  !$acc kernels
+  do concurrent (i = 1:n, j = 1:m, k = 1:l)
+    a(i,j,k) = b(i,j,k) * 2.0
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32, %{{.*}} : i32, %{{.*}} : i32) = (%c1{{.*}}, %c1{{.*}}, %c1{{.*}} : i32, i32, i32) to (%{{.*}}, %{{.*}}, %{{.*}} : i32, i32, i32) step (%c1{{.*}}, %c1{{.*}}, %c1{{.*}} : i32, i32, i32)
+! CHECK: acc.yield
+! CHECK: attributes {auto_ = [#acc.device_type<none>], inclusiveUpperbound = array<i1: true, true, true>}
+end subroutine
+
+
+! CHECK-LABEL: func.func @_QPnested_do_loops
+subroutine nested_do_loops()
+  integer :: i, j
+  integer, parameter :: n = 10, m = 20
+  real, dimension(n,m) :: a, b
+
+  ! Nested do loops
+  !$acc kernels
+  do i = 1, n
+    do j = 1, m
+      a(i,j) = b(i,j) + i + j
+    end do
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {auto_ = [#acc.device_type<none>], inclusiveUpperbound = array<i1: true>}
+! CHECK: acc.yield
+! CHECK: attributes {auto_ = [#acc.device_type<none>], inclusiveUpperbound = array<i1: true>}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPvariable_bounds_and_step
+subroutine variable_bounds_and_step(n, start_val, step_val)
+  integer, intent(in) :: n, start_val, step_val
+  integer :: i
+  real, dimension(n) :: a, b
+
+  ! Do loop with variable bounds and step
+  !$acc kernels
+  do i = start_val, n, step_val
+    a(i) = b(i) * 2.0
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.yield
+! CHECK: attributes {auto_ = [#acc.device_type<none>], inclusiveUpperbound = array<i1: true>}
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPdifferent_iv_types
+subroutine different_iv_types()
+  integer(kind=8) :: i8
+  integer(kind=4) :: i4
+  integer(kind=2) :: i2
+  integer, parameter :: n = 10
+  real, dimension(n) :: a, b, c, d
+
+  ! Test different iteration variable types
+  !$acc kernels
+  do i8 = 1_8, int(n,8)
+    a(i8) = b(i8) + 1.0
+  end do
+  !$acc end kernels
+
+  !$acc kernels
+  do i4 = 1, n
+    b(i4) = c(i4) + 1.0
+  end do
+  !$acc end kernels
+
+  !$acc kernels
+  do i2 = 1_2, int(n,2)
+    c(i2) = d(i2) + 1.0
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i64) = (%{{.*}} : i64) to (%{{.*}} : i64) step (%{{.*}} : i64)
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i32) = (%{{.*}} : i32) to (%{{.*}} : i32) step (%{{.*}} : i32)
+! CHECK: acc.kernels {
+! CHECK: acc.loop {{.*}} control(%{{.*}} : i16) = (%{{.*}} : i16) to (%{{.*}} : i16) step (%{{.*}} : i16)
+
+end subroutine
+
+! -----------------------------------------------------------------------------------------
+! Tests for loops that should NOT be converted to acc.loop due to unstructured control flow
+
+! CHECK-LABEL: func.func @_QPinfinite_loop_no_iv
+subroutine infinite_loop_no_iv()
+  integer :: i
+  logical :: condition
+
+  ! Infinite loop with no induction variable - should NOT convert to acc.loop
+  !$acc kernels
+  do
+    i = i + 1
+    if (i > 100) exit
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK-NOT: acc.loop
+
+end subroutine
+
+! CHECK-LABEL: func.func @_QPwhile_like_loop
+subroutine while_like_loop()
+  integer :: i
+  logical :: condition
+
+  i = 1
+  condition = .true.
+
+  ! While-like infinite loop - should NOT convert to acc.loop
+  !$acc kernels
+  do while (condition)
+    i = i + 1
+    if (i > 100) condition = .false.
+  end do
+  !$acc end kernels
+
+! CHECK: acc.kernels {
+! CHECK-NOT: acc.loop
+
+end subroutine
diff --git a/flang/test/Lower/OpenMP/unroll-heuristic01.f90 b/flang/test/Lower/OpenMP/unroll-heuristic01.f90
index a5f5c00..34020eb 100644
--- a/flang/test/Lower/OpenMP/unroll-heuristic01.f90
+++ b/flang/test/Lower/OpenMP/unroll-heuristic01.f90
@@ -13,27 +13,42 @@ subroutine omp_unroll_heuristic01(lb, ub, inc)
 end subroutine omp_unroll_heuristic01
 
 
-!CHECK-LABEL: func.func @_QPomp_unroll_heuristic01(
-!CHECK:      %c0_i32 = arith.constant 0 : i32
-!CHECK-NEXT: %c1_i32 = arith.constant 1 : i32
-!CHECK-NEXT: %13 = arith.cmpi slt, %12, %c0_i32 : i32
-!CHECK-NEXT: %14 = arith.subi %c0_i32, %12 : i32
-!CHECK-NEXT: %15 = arith.select %13, %14, %12 : i32
-!CHECK-NEXT: %16 = arith.select %13, %11, %10 : i32
-!CHECK-NEXT: %17 = arith.select %13, %10, %11 : i32
-!CHECK-NEXT: %18 = arith.subi %17, %16 overflow<nuw> : i32
-!CHECK-NEXT: %19 = arith.divui %18, %15 : i32
-!CHECK-NEXT: %20 = arith.addi %19, %c1_i32 overflow<nuw> : i32
-!CHECK-NEXT: %21 = arith.cmpi slt, %17, %16 : i32
-!CHECK-NEXT: %22 = arith.select %21, %c0_i32, %20 : i32
-!CHECK-NEXT: %canonloop_s0 = omp.new_cli
-!CHECK-NEXT: omp.canonical_loop(%canonloop_s0) %iv : i32 in range(%22) {
-!CHECK-NEXT:   %23 = arith.muli %iv, %12 : i32
-!CHECK-NEXT:   %24 = arith.addi %10, %23 : i32
-!CHECK-NEXT:   hlfir.assign %24 to %9#0 : i32, !fir.ref<i32>
-!CHECK-NEXT:   %25 = fir.load %9#0 : !fir.ref<i32>
-!CHECK-NEXT:   hlfir.assign %25 to %6#0 : i32, !fir.ref<i32>
-!CHECK-NEXT:   omp.terminator
-!CHECK-NEXT: }
-!CHECK-NEXT: omp.unroll_heuristic(%canonloop_s0)
-!CHECK-NEXT: return
+! CHECK-LABEL:   func.func @_QPomp_unroll_heuristic01(
+! CHECK-SAME:      %[[ARG0:.*]]: !fir.ref<i32> {fir.bindc_name = "lb"},
+! CHECK-SAME:      %[[ARG1:.*]]: !fir.ref<i32> {fir.bindc_name = "ub"},
+! CHECK-SAME:      %[[ARG2:.*]]: !fir.ref<i32> {fir.bindc_name = "inc"}) {
+! CHECK:           %[[VAL_0:.*]] = fir.dummy_scope : !fir.dscope
+! CHECK:           %[[VAL_1:.*]] = fir.alloca i32 {bindc_name = "i", uniq_name = "_QFomp_unroll_heuristic01Ei"}
+! CHECK:           %[[VAL_2:.*]]:2 = hlfir.declare %[[VAL_1]] {uniq_name = "_QFomp_unroll_heuristic01Ei"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_3:.*]]:2 = hlfir.declare %[[ARG2]] dummy_scope %[[VAL_0]] {uniq_name = "_QFomp_unroll_heuristic01Einc"} : (!fir.ref<i32>, !fir.dscope) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_4:.*]]:2 = hlfir.declare %[[ARG0]] dummy_scope %[[VAL_0]] {uniq_name = "_QFomp_unroll_heuristic01Elb"} : (!fir.ref<i32>, !fir.dscope) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_5:.*]] = fir.alloca i32 {bindc_name = "res", uniq_name = "_QFomp_unroll_heuristic01Eres"}
+! CHECK:           %[[VAL_6:.*]]:2 = hlfir.declare %[[VAL_5]] {uniq_name = "_QFomp_unroll_heuristic01Eres"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_7:.*]]:2 = hlfir.declare %[[ARG1]] dummy_scope %[[VAL_0]] {uniq_name = "_QFomp_unroll_heuristic01Eub"} : (!fir.ref<i32>, !fir.dscope) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_8:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<i32>
+! CHECK:           %[[VAL_9:.*]] = fir.load %[[VAL_7]]#0 : !fir.ref<i32>
+! CHECK:           %[[VAL_10:.*]] = fir.load %[[VAL_3]]#0 : !fir.ref<i32>
+! CHECK:           %[[VAL_11:.*]] = arith.constant 0 : i32
+! CHECK:           %[[VAL_12:.*]] = arith.constant 1 : i32
+! CHECK:           %[[VAL_13:.*]] = arith.cmpi slt, %[[VAL_10]], %[[VAL_11]] : i32
+! CHECK:           %[[VAL_14:.*]] = arith.subi %[[VAL_11]], %[[VAL_10]] : i32
+! CHECK:           %[[VAL_15:.*]] = arith.select %[[VAL_13]], %[[VAL_14]], %[[VAL_10]] : i32
+! CHECK:           %[[VAL_16:.*]] = arith.select %[[VAL_13]], %[[VAL_9]], %[[VAL_8]] : i32
+! CHECK:           %[[VAL_17:.*]] = arith.select %[[VAL_13]], %[[VAL_8]], %[[VAL_9]] : i32
+! CHECK:           %[[VAL_18:.*]] = arith.subi %[[VAL_17]], %[[VAL_16]] overflow<nuw> : i32
+! CHECK:           %[[VAL_19:.*]] = arith.divui %[[VAL_18]], %[[VAL_15]] : i32
+! CHECK:           %[[VAL_20:.*]] = arith.addi %[[VAL_19]], %[[VAL_12]] overflow<nuw> : i32
+! CHECK:           %[[VAL_21:.*]] = arith.cmpi slt, %[[VAL_17]], %[[VAL_16]] : i32
+! CHECK:           %[[VAL_22:.*]] = arith.select %[[VAL_21]], %[[VAL_11]], %[[VAL_20]] : i32
+! CHECK:           %[[VAL_23:.*]] = omp.new_cli
+! CHECK:           omp.canonical_loop(%[[VAL_23]]) %[[VAL_24:.*]] : i32 in range(%[[VAL_22]]) {
+! CHECK:             %[[VAL_25:.*]] = arith.muli %[[VAL_24]], %[[VAL_10]] : i32
+! CHECK:             %[[VAL_26:.*]] = arith.addi %[[VAL_8]], %[[VAL_25]] : i32
+! CHECK:             hlfir.assign %[[VAL_26]] to %[[VAL_2]]#0 : i32, !fir.ref<i32>
+! CHECK:             %[[VAL_27:.*]] = fir.load %[[VAL_2]]#0 : !fir.ref<i32>
+! CHECK:             hlfir.assign %[[VAL_27]] to %[[VAL_6]]#0 : i32, !fir.ref<i32>
+! CHECK:             omp.terminator
+! CHECK:           }
+! CHECK:           omp.unroll_heuristic(%[[VAL_23]])
+! CHECK:           return
+! CHECK:         }
\ No newline at end of file
diff --git a/flang/test/Lower/OpenMP/unroll-heuristic02.f90 b/flang/test/Lower/OpenMP/unroll-heuristic02.f90
index 14f694d..fdb1366 100644
--- a/flang/test/Lower/OpenMP/unroll-heuristic02.f90
+++ b/flang/test/Lower/OpenMP/unroll-heuristic02.f90
@@ -37,61 +37,55 @@ end subroutine omp_unroll_heuristic_nested02
 !CHECK:           %[[VAL_10:.*]]:2 = hlfir.declare %[[ARG1]] dummy_scope %[[VAL_0]] {uniq_name = "_QFomp_unroll_heuristic_nested02Eouter_ub"} : (!fir.ref<i32>, !fir.dscope) -> (!fir.ref<i32>, !fir.ref<i32>)
 !CHECK:           %[[VAL_11:.*]] = fir.alloca i32 {bindc_name = "res", uniq_name = "_QFomp_unroll_heuristic_nested02Eres"}
 !CHECK:           %[[VAL_12:.*]]:2 = hlfir.declare %[[VAL_11]] {uniq_name = "_QFomp_unroll_heuristic_nested02Eres"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
-!CHECK:           %[[VAL_13:.*]] = fir.alloca i32 {bindc_name = "i", pinned, uniq_name = "_QFomp_unroll_heuristic_nested02Ei"}
-!CHECK:           %[[VAL_14:.*]]:2 = hlfir.declare %[[VAL_13]] {uniq_name = "_QFomp_unroll_heuristic_nested02Ei"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
-!CHECK:           %[[VAL_15:.*]] = fir.alloca i32 {bindc_name = "j", pinned, uniq_name = "_QFomp_unroll_heuristic_nested02Ej"}
-!CHECK:           %[[VAL_16:.*]]:2 = hlfir.declare %[[VAL_15]] {uniq_name = "_QFomp_unroll_heuristic_nested02Ej"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
-!CHECK:           %[[VAL_17:.*]] = fir.load %[[VAL_9]]#0 : !fir.ref<i32>
-!CHECK:           %[[VAL_18:.*]] = fir.load %[[VAL_10]]#0 : !fir.ref<i32>
-!CHECK:           %[[VAL_19:.*]] = fir.load %[[VAL_8]]#0 : !fir.ref<i32>
-!CHECK:           %[[VAL_20:.*]] = arith.constant 0 : i32
-!CHECK:           %[[VAL_21:.*]] = arith.constant 1 : i32
-!CHECK:           %[[VAL_22:.*]] = arith.cmpi slt, %[[VAL_19]], %[[VAL_20]] : i32
-!CHECK:           %[[VAL_23:.*]] = arith.subi %[[VAL_20]], %[[VAL_19]] : i32
-!CHECK:           %[[VAL_24:.*]] = arith.select %[[VAL_22]], %[[VAL_23]], %[[VAL_19]] : i32
-!CHECK:           %[[VAL_25:.*]] = arith.select %[[VAL_22]], %[[VAL_18]], %[[VAL_17]] : i32
-!CHECK:           %[[VAL_26:.*]] = arith.select %[[VAL_22]], %[[VAL_17]], %[[VAL_18]] : i32
-!CHECK:           %[[VAL_27:.*]] = arith.subi %[[VAL_26]], %[[VAL_25]] overflow<nuw> : i32
-!CHECK:           %[[VAL_28:.*]] = arith.divui %[[VAL_27]], %[[VAL_24]] : i32
-!CHECK:           %[[VAL_29:.*]] = arith.addi %[[VAL_28]], %[[VAL_21]] overflow<nuw> : i32
-!CHECK:           %[[VAL_30:.*]] = arith.cmpi slt, %[[VAL_26]], %[[VAL_25]] : i32
-!CHECK:           %[[VAL_31:.*]] = arith.select %[[VAL_30]], %[[VAL_20]], %[[VAL_29]] : i32
-!CHECK:           %[[VAL_32:.*]] = omp.new_cli
-!CHECK:           omp.canonical_loop(%[[VAL_32]]) %[[VAL_33:.*]] : i32 in range(%[[VAL_31]]) {
-!CHECK:             %[[VAL_34:.*]] = arith.muli %[[VAL_33]], %[[VAL_19]] : i32
-!CHECK:             %[[VAL_35:.*]] = arith.addi %[[VAL_17]], %[[VAL_34]] : i32
-!CHECK:             hlfir.assign %[[VAL_35]] to %[[VAL_14]]#0 : i32, !fir.ref<i32>
-!CHECK:             %[[VAL_36:.*]] = fir.alloca i32 {bindc_name = "j", pinned, uniq_name = "_QFomp_unroll_heuristic_nested02Ej"}
-!CHECK:             %[[VAL_37:.*]]:2 = hlfir.declare %[[VAL_36]] {uniq_name = "_QFomp_unroll_heuristic_nested02Ej"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
-!CHECK:             %[[VAL_38:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<i32>
-!CHECK:             %[[VAL_39:.*]] = fir.load %[[VAL_5]]#0 : !fir.ref<i32>
-!CHECK:             %[[VAL_40:.*]] = fir.load %[[VAL_3]]#0 : !fir.ref<i32>
-!CHECK:             %[[VAL_41:.*]] = arith.constant 0 : i32
-!CHECK:             %[[VAL_42:.*]] = arith.constant 1 : i32
-!CHECK:             %[[VAL_43:.*]] = arith.cmpi slt, %[[VAL_40]], %[[VAL_41]] : i32
-!CHECK:             %[[VAL_44:.*]] = arith.subi %[[VAL_41]], %[[VAL_40]] : i32
-!CHECK:             %[[VAL_45:.*]] = arith.select %[[VAL_43]], %[[VAL_44]], %[[VAL_40]] : i32
-!CHECK:             %[[VAL_46:.*]] = arith.select %[[VAL_43]], %[[VAL_39]], %[[VAL_38]] : i32
-!CHECK:             %[[VAL_47:.*]] = arith.select %[[VAL_43]], %[[VAL_38]], %[[VAL_39]] : i32
-!CHECK:             %[[VAL_48:.*]] = arith.subi %[[VAL_47]], %[[VAL_46]] overflow<nuw> : i32
-!CHECK:             %[[VAL_49:.*]] = arith.divui %[[VAL_48]], %[[VAL_45]] : i32
-!CHECK:             %[[VAL_50:.*]] = arith.addi %[[VAL_49]], %[[VAL_42]] overflow<nuw> : i32
-!CHECK:             %[[VAL_51:.*]] = arith.cmpi slt, %[[VAL_47]], %[[VAL_46]] : i32
-!CHECK:             %[[VAL_52:.*]] = arith.select %[[VAL_51]], %[[VAL_41]], %[[VAL_50]] : i32
-!CHECK:             %[[VAL_53:.*]] = omp.new_cli
-!CHECK:             omp.canonical_loop(%[[VAL_53]]) %[[VAL_54:.*]] : i32 in range(%[[VAL_52]]) {
-!CHECK:               %[[VAL_55:.*]] = arith.muli %[[VAL_54]], %[[VAL_40]] : i32
-!CHECK:               %[[VAL_56:.*]] = arith.addi %[[VAL_38]], %[[VAL_55]] : i32
-!CHECK:               hlfir.assign %[[VAL_56]] to %[[VAL_37]]#0 : i32, !fir.ref<i32>
-!CHECK:               %[[VAL_57:.*]] = fir.load %[[VAL_14]]#0 : !fir.ref<i32>
-!CHECK:               %[[VAL_58:.*]] = fir.load %[[VAL_37]]#0 : !fir.ref<i32>
-!CHECK:               %[[VAL_59:.*]] = arith.addi %[[VAL_57]], %[[VAL_58]] : i32
-!CHECK:               hlfir.assign %[[VAL_59]] to %[[VAL_12]]#0 : i32, !fir.ref<i32>
+!CHECK:           %[[VAL_13:.*]] = fir.load %[[VAL_9]]#0 : !fir.ref<i32>
+!CHECK:           %[[VAL_14:.*]] = fir.load %[[VAL_10]]#0 : !fir.ref<i32>
+!CHECK:           %[[VAL_15:.*]] = fir.load %[[VAL_8]]#0 : !fir.ref<i32>
+!CHECK:           %[[VAL_16:.*]] = arith.constant 0 : i32
+!CHECK:           %[[VAL_17:.*]] = arith.constant 1 : i32
+!CHECK:           %[[VAL_18:.*]] = arith.cmpi slt, %[[VAL_15]], %[[VAL_16]] : i32
+!CHECK:           %[[VAL_19:.*]] = arith.subi %[[VAL_16]], %[[VAL_15]] : i32
+!CHECK:           %[[VAL_20:.*]] = arith.select %[[VAL_18]], %[[VAL_19]], %[[VAL_15]] : i32
+!CHECK:           %[[VAL_21:.*]] = arith.select %[[VAL_18]], %[[VAL_14]], %[[VAL_13]] : i32
+!CHECK:           %[[VAL_22:.*]] = arith.select %[[VAL_18]], %[[VAL_13]], %[[VAL_14]] : i32
+!CHECK:           %[[VAL_23:.*]] = arith.subi %[[VAL_22]], %[[VAL_21]] overflow<nuw> : i32
+!CHECK:           %[[VAL_24:.*]] = arith.divui %[[VAL_23]], %[[VAL_20]] : i32
+!CHECK:           %[[VAL_25:.*]] = arith.addi %[[VAL_24]], %[[VAL_17]] overflow<nuw> : i32
+!CHECK:           %[[VAL_26:.*]] = arith.cmpi slt, %[[VAL_22]], %[[VAL_21]] : i32
+!CHECK:           %[[VAL_27:.*]] = arith.select %[[VAL_26]], %[[VAL_16]], %[[VAL_25]] : i32
+!CHECK:           %[[VAL_28:.*]] = omp.new_cli
+!CHECK:           omp.canonical_loop(%[[VAL_28]]) %[[VAL_29:.*]] : i32 in range(%[[VAL_27]]) {
+!CHECK:             %[[VAL_30:.*]] = arith.muli %[[VAL_29]], %[[VAL_15]] : i32
+!CHECK:             %[[VAL_31:.*]] = arith.addi %[[VAL_13]], %[[VAL_30]] : i32
+!CHECK:             hlfir.assign %[[VAL_31]] to %[[VAL_2]]#0 : i32, !fir.ref<i32>
+!CHECK:             %[[VAL_32:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<i32>
+!CHECK:             %[[VAL_33:.*]] = fir.load %[[VAL_5]]#0 : !fir.ref<i32>
+!CHECK:             %[[VAL_34:.*]] = fir.load %[[VAL_3]]#0 : !fir.ref<i32>
+!CHECK:             %[[VAL_35:.*]] = arith.constant 0 : i32
+!CHECK:             %[[VAL_36:.*]] = arith.constant 1 : i32
+!CHECK:             %[[VAL_37:.*]] = arith.cmpi slt, %[[VAL_34]], %[[VAL_35]] : i32
+!CHECK:             %[[VAL_38:.*]] = arith.subi %[[VAL_35]], %[[VAL_34]] : i32
+!CHECK:             %[[VAL_39:.*]] = arith.select %[[VAL_37]], %[[VAL_38]], %[[VAL_34]] : i32
+!CHECK:             %[[VAL_40:.*]] = arith.select %[[VAL_37]], %[[VAL_33]], %[[VAL_32]] : i32
+!CHECK:             %[[VAL_41:.*]] = arith.select %[[VAL_37]], %[[VAL_32]], %[[VAL_33]] : i32
+!CHECK:             %[[VAL_42:.*]] = arith.subi %[[VAL_41]], %[[VAL_40]] overflow<nuw> : i32
+!CHECK:             %[[VAL_43:.*]] = arith.divui %[[VAL_42]], %[[VAL_39]] : i32
+!CHECK:             %[[VAL_44:.*]] = arith.addi %[[VAL_43]], %[[VAL_36]] overflow<nuw> : i32
+!CHECK:             %[[VAL_45:.*]] = arith.cmpi slt, %[[VAL_41]], %[[VAL_40]] : i32
+!CHECK:             %[[VAL_46:.*]] = arith.select %[[VAL_45]], %[[VAL_35]], %[[VAL_44]] : i32
+!CHECK:             %[[VAL_47:.*]] = omp.new_cli
+!CHECK:             omp.canonical_loop(%[[VAL_47]]) %[[VAL_48:.*]] : i32 in range(%[[VAL_46]]) {
+!CHECK:               %[[VAL_49:.*]] = arith.muli %[[VAL_48]], %[[VAL_34]] : i32
+!CHECK:               %[[VAL_50:.*]] = arith.addi %[[VAL_32]], %[[VAL_49]] : i32
+!CHECK:               hlfir.assign %[[VAL_50]] to %[[VAL_7]]#0 : i32, !fir.ref<i32>
+!CHECK:               %[[VAL_51:.*]] = fir.load %[[VAL_2]]#0 : !fir.ref<i32>
+!CHECK:               %[[VAL_52:.*]] = fir.load %[[VAL_7]]#0 : !fir.ref<i32>
+!CHECK:               %[[VAL_53:.*]] = arith.addi %[[VAL_51]], %[[VAL_52]] : i32
+!CHECK:               hlfir.assign %[[VAL_53]] to %[[VAL_12]]#0 : i32, !fir.ref<i32>
 !CHECK:               omp.terminator
 !CHECK:             }
-!CHECK:             omp.unroll_heuristic(%[[VAL_53]])
+!CHECK:             omp.unroll_heuristic(%[[VAL_47]])
 !CHECK:             omp.terminator
 !CHECK:           }
-!CHECK:           omp.unroll_heuristic(%[[VAL_32]])
+!CHECK:           omp.unroll_heuristic(%[[VAL_28]])
 !CHECK:           return
 !CHECK:         }
diff --git a/flang/test/Lower/OpenMP/unroll-heuristic03.f90 b/flang/test/Lower/OpenMP/unroll-heuristic03.f90
new file mode 100644
index 0000000..308c149
--- /dev/null
+++ b/flang/test/Lower/OpenMP/unroll-heuristic03.f90
@@ -0,0 +1,61 @@
+! RUN: %flang_fc1 -emit-hlfir -fopenmp -fopenmp-version=51 -o - %s 2>&1 | FileCheck %s
+
+! Test implicitly privatized loop variable that is affected by unrolling.
+
+subroutine omp_unroll_heuristic03(lb, ub, inc)
+  integer res, i, lb, ub, inc
+
+  !$omp parallel
+    !$omp unroll
+    do i = lb, ub, inc
+      res = i
+    end do
+    !$omp end unroll
+  !$omp end parallel
+
+end subroutine omp_unroll_heuristic03
+
+
+! CHECK-LABEL:   func.func @_QPomp_unroll_heuristic03(
+! CHECK-SAME:      %[[ARG0:.*]]: !fir.ref<i32> {fir.bindc_name = "lb"},
+! CHECK-SAME:      %[[ARG1:.*]]: !fir.ref<i32> {fir.bindc_name = "ub"},
+! CHECK-SAME:      %[[ARG2:.*]]: !fir.ref<i32> {fir.bindc_name = "inc"}) {
+! CHECK:           %[[VAL_0:.*]] = fir.dummy_scope : !fir.dscope
+! CHECK:           %[[VAL_1:.*]] = fir.alloca i32 {bindc_name = "i", uniq_name = "_QFomp_unroll_heuristic03Ei"}
+! CHECK:           %[[VAL_2:.*]]:2 = hlfir.declare %[[VAL_1]] {uniq_name = "_QFomp_unroll_heuristic03Ei"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_3:.*]]:2 = hlfir.declare %[[ARG2]] dummy_scope %[[VAL_0]] {uniq_name = "_QFomp_unroll_heuristic03Einc"} : (!fir.ref<i32>, !fir.dscope) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_4:.*]]:2 = hlfir.declare %[[ARG0]] dummy_scope %[[VAL_0]] {uniq_name = "_QFomp_unroll_heuristic03Elb"} : (!fir.ref<i32>, !fir.dscope) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_5:.*]] = fir.alloca i32 {bindc_name = "res", uniq_name = "_QFomp_unroll_heuristic03Eres"}
+! CHECK:           %[[VAL_6:.*]]:2 = hlfir.declare %[[VAL_5]] {uniq_name = "_QFomp_unroll_heuristic03Eres"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           %[[VAL_7:.*]]:2 = hlfir.declare %[[ARG1]] dummy_scope %[[VAL_0]] {uniq_name = "_QFomp_unroll_heuristic03Eub"} : (!fir.ref<i32>, !fir.dscope) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:           omp.parallel private(@_QFomp_unroll_heuristic03Ei_private_i32 %[[VAL_2]]#0 -> %[[VAL_8:.*]] : !fir.ref<i32>) {
+! CHECK:             %[[VAL_9:.*]]:2 = hlfir.declare %[[VAL_8]] {uniq_name = "_QFomp_unroll_heuristic03Ei"} : (!fir.ref<i32>) -> (!fir.ref<i32>, !fir.ref<i32>)
+! CHECK:             %[[VAL_10:.*]] = fir.load %[[VAL_4]]#0 : !fir.ref<i32>
+! CHECK:             %[[VAL_11:.*]] = fir.load %[[VAL_7]]#0 : !fir.ref<i32>
+! CHECK:             %[[VAL_12:.*]] = fir.load %[[VAL_3]]#0 : !fir.ref<i32>
+! CHECK:             %[[VAL_13:.*]] = arith.constant 0 : i32
+! CHECK:             %[[VAL_14:.*]] = arith.constant 1 : i32
+! CHECK:             %[[VAL_15:.*]] = arith.cmpi slt, %[[VAL_12]], %[[VAL_13]] : i32
+! CHECK:             %[[VAL_16:.*]] = arith.subi %[[VAL_13]], %[[VAL_12]] : i32
+! CHECK:             %[[VAL_17:.*]] = arith.select %[[VAL_15]], %[[VAL_16]], %[[VAL_12]] : i32
+! CHECK:             %[[VAL_18:.*]] = arith.select %[[VAL_15]], %[[VAL_11]], %[[VAL_10]] : i32
+! CHECK:             %[[VAL_19:.*]] = arith.select %[[VAL_15]], %[[VAL_10]], %[[VAL_11]] : i32
+! CHECK:             %[[VAL_20:.*]] = arith.subi %[[VAL_19]], %[[VAL_18]] overflow<nuw> : i32
+! CHECK:             %[[VAL_21:.*]] = arith.divui %[[VAL_20]], %[[VAL_17]] : i32
+! CHECK:             %[[VAL_22:.*]] = arith.addi %[[VAL_21]], %[[VAL_14]] overflow<nuw> : i32
+! CHECK:             %[[VAL_23:.*]] = arith.cmpi slt, %[[VAL_19]], %[[VAL_18]] : i32
+! CHECK:             %[[VAL_24:.*]] = arith.select %[[VAL_23]], %[[VAL_13]], %[[VAL_22]] : i32
+! CHECK:             %[[VAL_25:.*]] = omp.new_cli
+! CHECK:             omp.canonical_loop(%[[VAL_25]]) %[[VAL_26:.*]] : i32 in range(%[[VAL_24]]) {
+! CHECK:               %[[VAL_27:.*]] = arith.muli %[[VAL_26]], %[[VAL_12]] : i32
+! CHECK:               %[[VAL_28:.*]] = arith.addi %[[VAL_10]], %[[VAL_27]] : i32
+! CHECK:               hlfir.assign %[[VAL_28]] to %[[VAL_9]]#0 : i32, !fir.ref<i32>
+! CHECK:               %[[VAL_29:.*]] = fir.load %[[VAL_9]]#0 : !fir.ref<i32>
+! CHECK:               hlfir.assign %[[VAL_29]] to %[[VAL_6]]#0 : i32, !fir.ref<i32>
+! CHECK:               omp.terminator
+! CHECK:             }
+! CHECK:             omp.unroll_heuristic(%[[VAL_25]])
+! CHECK:             omp.terminator
+! CHECK:           }
+! CHECK:           return
+! CHECK:         }
\ No newline at end of file
diff --git a/flang/test/Parser/OpenMP/declare-target-indirect-tree.f90 b/flang/test/Parser/OpenMP/declare-target-indirect-tree.f90
index df85942..916bd66 100644
--- a/flang/test/Parser/OpenMP/declare-target-indirect-tree.f90
+++ b/flang/test/Parser/OpenMP/declare-target-indirect-tree.f90
@@ -15,7 +15,8 @@ module functions
 contains
   function func1() result(i)
     !$omp declare target enter(func1) indirect(.true.)
-    !CHECK: | | | | | OmpDeclareTargetSpecifier -> OmpDeclareTargetWithClause -> OmpClauseList -> OmpClause -> Enter -> OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'func1'
+    !CHECK: | | | | | OmpDeclareTargetSpecifier -> OmpDeclareTargetWithClause -> OmpClauseList -> OmpClause -> Enter -> OmpEnterClause
+    !CHECK-NEXT: | | | | | | OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'func1'
     !CHECK-NEXT: | | | | | OmpClause -> Indirect -> OmpIndirectClause -> Scalar -> Logical -> Expr = '.true._4'
     !CHECK-NEXT: | | | | | | LiteralConstant -> LogicalLiteralConstant
     !CHECK-NEXT: | | | | | | | bool = 'true'
@@ -26,7 +27,8 @@ contains
 
   function func2() result(i)
     !$omp declare target enter(func2) indirect
-    !CHECK: | | | | | OmpDeclareTargetSpecifier -> OmpDeclareTargetWithClause -> OmpClauseList -> OmpClause -> Enter -> OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'func2'
+    !CHECK: | | | | | OmpDeclareTargetSpecifier -> OmpDeclareTargetWithClause -> OmpClauseList -> OmpClause -> Enter -> OmpEnterClause
+    !CHECK-NEXT: | | | | | | OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'func2'
     !CHECK-NEXT: | | | | | OmpClause -> Indirect -> OmpIndirectClause ->
     character(1) :: i
     i = 'b'
diff --git a/flang/test/Parser/OpenMP/enter-automap-modifier.f90 b/flang/test/Parser/OpenMP/enter-automap-modifier.f90
new file mode 100644
index 0000000..1f361ca5
--- /dev/null
+++ b/flang/test/Parser/OpenMP/enter-automap-modifier.f90
@@ -0,0 +1,16 @@
+!RUN: %flang_fc1 -fdebug-unparse -fopenmp -fopenmp-version=60 %s | FileCheck %s --check-prefix=UNPARSE
+!RUN: %flang_fc1 -fdebug-dump-parse-tree -fopenmp -fopenmp-version=60 %s | FileCheck %s --check-prefix=PARSE-TREE
+
+program automap
+   integer :: x
+   !$omp declare target enter(automap: x)
+end program
+
+!UNPARSE: PROGRAM AUTOMAP
+!UNPARSE:  INTEGER x
+!UNPARSE: !$OMP DECLARE TARGET  ENTER(AUTOMAP: x)
+!UNPARSE: END PROGRAM
+
+!PARSE-TREE: OmpClauseList -> OmpClause -> Enter -> OmpEnterClause
+!PARSE-TREE-NEXT: | Modifier -> OmpAutomapModifier -> Value = Automap
+!PARSE-TREE-NEXT: | OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'x'
diff --git a/flang/test/Parser/OpenMP/sections.f90 b/flang/test/Parser/OpenMP/sections.f90
index 3752cef..8ba2294 100644
--- a/flang/test/Parser/OpenMP/sections.f90
+++ b/flang/test/Parser/OpenMP/sections.f90
@@ -10,32 +10,41 @@ subroutine openmp_sections(x, y)
 !==============================================================================
 !CHECK: !$omp sections
 !$omp sections
-  !CHECK: !$omp section
 !CHECK: !$omp end sections
 !$omp end sections
 
-!PARSE-TREE: OpenMPConstruct -> OpenMPSectionsConstruct
-!PARSE-TREE: OmpBeginSectionsDirective
-!PARSE-TREE-NOT: ExecutionPartConstruct
-!PARSE-TREE: OmpEndSectionsDirective
+!PARSE-TREE: ExecutionPartConstruct -> ExecutableConstruct -> OpenMPConstruct -> OpenMPSectionsConstruct
+!PARSE-TREE: | OmpBeginSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | Block
+!PARSE-TREE: | OmpEndSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
 
 !==============================================================================
 ! single section, without `!$omp section`
 !==============================================================================
 !CHECK: !$omp sections
 !$omp sections
-  !CHECK: !$omp section
     !CHECK: CALL
     call F1()
 !CHECK: !$omp end sections
 !$omp end sections
 
-!PARSE-TREE: OpenMPConstruct -> OpenMPSectionsConstruct
-!PARSE-TREE:  OmpBeginSectionsDirective
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE-NOT: ExecutionPartConstruct
-!PARSE-TREE:  OmpEndSectionsDirective
+!PARSE-TREE: ExecutionPartConstruct -> ExecutableConstruct -> OpenMPConstruct -> OpenMPSectionsConstruct
+!PARSE-TREE: | OmpBeginSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f1()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f1'
+!PARSE-TREE: | OmpEndSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
 
 !==============================================================================
 ! single section with `!$omp section`
@@ -49,12 +58,22 @@ subroutine openmp_sections(x, y)
 !CHECK: !$omp end sections
 !$omp end sections
 
-!PARSE-TREE: OpenMPConstruct -> OpenMPSectionsConstruct
-!PARSE-TREE:  OmpBeginSectionsDirective
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE-NOT: ExecutionPartConstruct
-!PARSE-TREE:  OmpEndSectionsDirective
+!PARSE-TREE: ExecutionPartConstruct -> ExecutableConstruct -> OpenMPConstruct -> OpenMPSectionsConstruct
+!PARSE-TREE: | OmpBeginSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | OmpDirectiveSpecification
+!PARSE-TREE: | | | OmpDirectiveName -> llvm::omp::Directive = section
+!PARSE-TREE: | | | OmpClauseList ->
+!PARSE-TREE: | | | Flags = None
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f1()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f1'
+!PARSE-TREE: | OmpEndSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
 
 !==============================================================================
 ! multiple sections
@@ -76,16 +95,40 @@ subroutine openmp_sections(x, y)
 !CHECK: !$omp end sections
 !$omp end sections
 
-!PARSE-TREE: OpenMPConstruct -> OpenMPSectionsConstruct
-!PARSE-TREE:  OmpBeginSectionsDirective
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE-NOT: ExecutionPartConstruct
-!PARSE-TREE:  OmpEndSectionsDirective
+!PARSE-TREE: ExecutionPartConstruct -> ExecutableConstruct -> OpenMPConstruct -> OpenMPSectionsConstruct
+!PARSE-TREE: | OmpBeginSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | OmpDirectiveSpecification
+!PARSE-TREE: | | | OmpDirectiveName -> llvm::omp::Directive = section
+!PARSE-TREE: | | | OmpClauseList ->
+!PARSE-TREE: | | | Flags = None
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f1()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f1'
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | OmpDirectiveSpecification
+!PARSE-TREE: | | | OmpDirectiveName -> llvm::omp::Directive = section
+!PARSE-TREE: | | | OmpClauseList ->
+!PARSE-TREE: | | | Flags = None
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f2()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f2'
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | OmpDirectiveSpecification
+!PARSE-TREE: | | | OmpDirectiveName -> llvm::omp::Directive = section
+!PARSE-TREE: | | | OmpClauseList ->
+!PARSE-TREE: | | | Flags = None
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f3()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f3'
+!PARSE-TREE: | OmpEndSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList ->
 
 !==============================================================================
 ! multiple sections with clauses
@@ -107,15 +150,40 @@ subroutine openmp_sections(x, y)
 !CHECK: !$omp end sections NOWAIT
 !$omp end sections NOWAIT
 
-!PARSE-TREE: OpenMPConstruct -> OpenMPSectionsConstruct
-!PARSE-TREE:  OmpBeginSectionsDirective
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE:   OpenMPConstruct -> OpenMPSectionConstruct -> Block
-!PARSE-TREE:    CallStmt
-!PARSE-TREE-NOT: ExecutionPartConstruct
-!PARSE-TREE:  OmpEndSectionsDirective
+!PARSE-TREE: ExecutionPartConstruct -> ExecutableConstruct -> OpenMPConstruct -> OpenMPSectionsConstruct
+!PARSE-TREE: | OmpBeginSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList -> OmpClause -> Private -> OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'x'
+!PARSE-TREE: | | OmpClause -> Firstprivate -> OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'y'
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | OmpDirectiveSpecification
+!PARSE-TREE: | | | OmpDirectiveName -> llvm::omp::Directive = section
+!PARSE-TREE: | | | OmpClauseList ->
+!PARSE-TREE: | | | Flags = None
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f1()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f1'
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | OmpDirectiveSpecification
+!PARSE-TREE: | | | OmpDirectiveName -> llvm::omp::Directive = section
+!PARSE-TREE: | | | OmpClauseList ->
+!PARSE-TREE: | | | Flags = None
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f2()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f2'
+!PARSE-TREE: | OpenMPConstruct -> OpenMPSectionConstruct
+!PARSE-TREE: | | OmpDirectiveSpecification
+!PARSE-TREE: | | | OmpDirectiveName -> llvm::omp::Directive = section
+!PARSE-TREE: | | | OmpClauseList ->
+!PARSE-TREE: | | | Flags = None
+!PARSE-TREE: | | Block
+!PARSE-TREE: | | | ExecutionPartConstruct -> ExecutableConstruct -> ActionStmt -> CallStmt = 'CALL f3()'
+!PARSE-TREE: | | | | Call
+!PARSE-TREE: | | | | | ProcedureDesignator -> Name = 'f3'
+!PARSE-TREE: | OmpEndSectionsDirective
+!PARSE-TREE: | | OmpSectionsDirective -> llvm::omp::Directive = sections
+!PARSE-TREE: | | OmpClauseList -> OmpClause -> Nowait
 
 END subroutine openmp_sections
diff --git a/flang/test/Semantics/OpenACC/acc-atomic-validity.f90 b/flang/test/Semantics/OpenACC/acc-atomic-validity.f90
index 07fb864..5c8d33f 100644
--- a/flang/test/Semantics/OpenACC/acc-atomic-validity.f90
+++ b/flang/test/Semantics/OpenACC/acc-atomic-validity.f90
@@ -54,11 +54,38 @@ program openacc_atomic_validity
   i = c(i)
   !$acc end atomic
 
+  !TODO: Should error because c(i) references i which is the atomic update variable.
   !$acc atomic capture
   c(i) = i
   i = i + 1
   !$acc end atomic
 
+  !ERROR: The variables assigned in this atomic capture construct must be distinct
+  !$acc atomic capture
+  c(1) = c(2)
+  c(1) = c(3)
+  !$acc end atomic
+  
+  !ERROR: The assignments in this atomic capture construct do not update a variable and capture either its initial or final value
+  !$acc atomic capture
+  c(1) = c(2)
+  c(2) = c(2)
+  !$acc end atomic
+
+  !ERROR: The assignments in this atomic capture construct do not update a variable and capture either its initial or final value
+  !$acc atomic capture
+  c(1) = c(2)
+  c(2) = c(1)
+  !$acc end atomic
+
+  !ERROR: The assignments in this atomic capture construct do not update a variable and capture either its initial or final value
+  !$acc atomic capture
+  c(1) = c(2)
+  c(3) = c(2)
+  !$acc end atomic
+
+
+
   !$acc atomic capture if(l .EQV. .false.)
   c(i) = i
   i = i + 1
@@ -79,3 +106,45 @@ program openacc_atomic_validity
   !$acc end parallel
 
 end program openacc_atomic_validity
+
+subroutine capture_with_convert_f64_to_i32()
+  integer :: x
+  real(8) :: v, w
+  x = 1
+  v = 0
+  w = 2
+
+  !$acc atomic capture
+  x = x * 2.5_8
+  v = x
+  !$acc end atomic
+
+  !$acc atomic capture
+  !TODO: The rhs side of this update statement cannot reference v.
+  x = x * v
+  v = x
+  !$acc end atomic
+
+  !$acc atomic capture
+  !TODO: The rhs side of this update statement cannot reference v.
+  x = v * x
+  v = x
+  !$acc end atomic
+
+  !$acc atomic capture
+  !ERROR: The RHS of this atomic update statement must reference the updated variable: x
+  x = v * v
+  v = x
+  !$acc end atomic
+
+  !$acc atomic capture
+  x = v
+  !ERROR: The updated variable, v, cannot appear more than once in the atomic update operation
+  v = v * v
+  !$acc end atomic
+
+  !$acc atomic capture
+  v = x
+  x = w * w
+  !$acc end atomic
+end subroutine capture_with_convert_f64_to_i32
\ No newline at end of file
diff --git a/flang/test/Semantics/OpenACC/acc-default-none-function.f90 b/flang/test/Semantics/OpenACC/acc-default-none-function.f90
new file mode 100644
index 0000000..f0a697f
--- /dev/null
+++ b/flang/test/Semantics/OpenACC/acc-default-none-function.f90
@@ -0,0 +1,20 @@
+! RUN: %python %S/../test_errors.py %s %flang -fopenacc -pedantic
+
+module mm_acc_rout_function
+contains
+    integer function dosomething(res)
+        !$acc routine seq
+        integer :: res
+        dosomething = res + 1
+    end function
+end module
+    
+program main
+    use mm_acc_rout_function
+    implicit none
+    integer :: res = 1
+    !$acc serial default(none) copy(res)
+    res = dosomething(res)
+    !$acc end serial
+end program
+
diff --git a/flang/test/Semantics/assign02.f90 b/flang/test/Semantics/assign02.f90
index f998197..c447078 100644
--- a/flang/test/Semantics/assign02.f90
+++ b/flang/test/Semantics/assign02.f90
@@ -139,7 +139,7 @@ contains
     real, target :: x
     real, pointer :: p
     p => f1()
-    !ERROR: pointer 'p' is associated with the result of a reference to function 'f2' that is a not a pointer
+    !ERROR: pointer 'p' is associated with the result of a reference to function 'f2' that is not a pointer
     p => f2()
   contains
     function f1()
diff --git a/flang/test/Semantics/bug1214.cuf b/flang/test/Semantics/bug1214.cuf
new file mode 100644
index 0000000..114fad1
--- /dev/null
+++ b/flang/test/Semantics/bug1214.cuf
@@ -0,0 +1,49 @@
+! RUN: %flang_fc1 -fdebug-unparse %s 2>&1 | FileCheck %s
+module overrides
+  type realResult
+    real a
+  end type
+  interface operator(*)
+    procedure :: multHostDevice, multDeviceHost
+  end interface
+  interface assignment(=)
+    procedure :: assignHostResult, assignDeviceResult
+  end interface
+ contains
+  elemental function multHostDevice(x, y) result(result)
+    real, intent(in) :: x
+    real, intent(in), device :: y
+    type(realResult) result
+    result%a = x * y
+  end
+  elemental function multDeviceHost(x, y) result(result)
+    real, intent(in), device :: x
+    real, intent(in) :: y
+    type(realResult) result
+    result%a = x * y
+  end
+  elemental subroutine assignHostResult(lhs, rhs)
+    real, intent(out) :: lhs
+    type(realResult), intent(in) :: rhs
+    lhs = rhs%a
+  end
+  elemental subroutine assignDeviceResult(lhs, rhs)
+    real, intent(out), device :: lhs
+    type(realResult), intent(in) :: rhs
+    lhs = rhs%a
+  end
+end
+
+program p
+  use overrides
+  real, device :: da, db
+  real :: ha, hb
+!CHECK: CALL assigndeviceresult(db,multhostdevice(2._4,da))
+  db = 2. * da
+!CHECK: CALL assigndeviceresult(db,multdevicehost(da,2._4))
+  db = da * 2.
+!CHECK: CALL assignhostresult(ha,multhostdevice(2._4,da))
+  ha = 2. * da
+!CHECK: CALL assignhostresult(ha,multdevicehost(da,2._4))
+  ha = da * 2.
+end
diff --git a/flang/test/Semantics/cuf11.cuf b/flang/test/Semantics/cuf11.cuf
index 554ac25..1f5beb0 100644
--- a/flang/test/Semantics/cuf11.cuf
+++ b/flang/test/Semantics/cuf11.cuf
@@ -16,7 +16,7 @@ subroutine sub1()
   real, device :: adev(10), bdev(10)
   real :: ahost(10)
 
-!ERROR: More than one reference to a CUDA object on the right hand side of the assigment
+!ERROR: More than one reference to a CUDA object on the right hand side of the assignment
   ahost = adev + bdev
 
   ahost = adev + adev