[flang-commits] [flang] 6f0041e - [flang][lowering] Add lowering for the `size` intrinsic
Andrzej Warzynski via flang-commits
flang-commits at lists.llvm.org
Wed Mar 16 09:31:05 PDT 2022
Author: Andrzej Warzynski
Date: 2022-03-16T16:30:12Z
New Revision: 6f0041ea44f5e0eded2b05026242f2ccef7aa90a
URL: https://github.com/llvm/llvm-project/commit/6f0041ea44f5e0eded2b05026242f2ccef7aa90a
DIFF: https://github.com/llvm/llvm-project/commit/6f0041ea44f5e0eded2b05026242f2ccef7aa90a.diff
LOG: [flang][lowering] Add lowering for the `size` intrinsic
This patch adds support for lowering of the `size` intrinsic from Fortran
to the FIR dialect of MLIR.
This is part of the upstreaming effort from the `fir-dev` branch in [1].
[1] https://github.com/flang-compiler/f18-llvm-project
Differential Revision: https://reviews.llvm.org/D121803
Co-authored-by: Jean Perier <jperier at nvidia.com>
Co-authored-by: Eric Schweitz <eschweitz at nvidia.com>
Co-authored-by: Peter Steinfeld <psteinfeld at nvidia.com>
Added:
flang/test/Lower/Intrinsics/size.f90
Modified:
flang/lib/Lower/IntrinsicCall.cpp
Removed:
################################################################################
diff --git a/flang/lib/Lower/IntrinsicCall.cpp b/flang/lib/Lower/IntrinsicCall.cpp
index 856e8716a5767..4e24eb4c17ad6 100644
--- a/flang/lib/Lower/IntrinsicCall.cpp
+++ b/flang/lib/Lower/IntrinsicCall.cpp
@@ -690,6 +690,12 @@ static constexpr IntrinsicHandler handlers[]{
{{{"size", asBox}, {"put", asBox}, {"get", asBox}}},
/*isElemental=*/false},
{"set_exponent", &I::genSetExponent},
+ {"size",
+ &I::genSize,
+ {{{"array", asBox},
+ {"dim", asAddr, handleDynamicOptional},
+ {"kind", asValue}}},
+ /*isElemental=*/false},
{"sum",
&I::genSum,
{{{"array", asBox},
diff --git a/flang/test/Lower/Intrinsics/size.f90 b/flang/test/Lower/Intrinsics/size.f90
new file mode 100644
index 0000000000000..f3b565f4560af
--- /dev/null
+++ b/flang/test/Lower/Intrinsics/size.f90
@@ -0,0 +1,99 @@
+! RUN: bbc -emit-fir %s -o - | FileCheck %s
+! RUN: %flang_fc1 -emit-fir %s -o - | FileCheck %s
+
+! CHECK-LABEL: func @_QPsize_test() {
+subroutine size_test()
+ real, dimension(1:10, -10:10) :: a
+ integer :: dim = 1
+ integer :: iSize
+! CHECK: %[[VAL_c1:.*]] = arith.constant 1 : index
+! CHECK: %[[VAL_c10:.*]] = arith.constant 10 : index
+! CHECK: %[[VAL_cneg10:.*]] = arith.constant -10 : index
+! CHECK: %[[VAL_c21:.*]] = arith.constant 21 : index
+! CHECK: %[[VAL_0:.*]] = fir.alloca !fir.array<10x21xf32> {bindc_name = "a", uniq_name = "_QFsize_testEa"}
+! CHECK: %[[VAL_1:.*]] = fir.address_of(@_QFsize_testEdim) : !fir.ref<i32>
+! CHECK: %[[VAL_2:.*]] = fir.alloca i32 {bindc_name = "isize", uniq_name = "_QFsize_testEisize"}
+! CHECK: %[[VAL_c2_i64:.*]] = arith.constant 2 : i64
+! CHECK: %[[VAL_3:.*]] = fir.convert %[[VAL_c2_i64]] : (i64) -> index
+! CHECK: %[[VAL_c1_i64:.*]] = arith.constant 1 : i64
+! CHECK: %[[VAL_4:.*]] = fir.convert %[[VAL_c1_i64]] : (i64) -> index
+! CHECK: %[[VAL_c5_i64:.*]] = arith.constant 5 : i64
+! CHECK: %[[VAL_5:.*]] = fir.convert %[[VAL_c5_i64]] : (i64) -> index
+! CHECK: %[[VAL_neg1_i64:.*]] = arith.constant -1 : i64
+! CHECK: %[[VAL_6:.*]] = fir.convert %[[VAL_neg1_i64]] : (i64) -> index
+! CHECK: %[[VAL_c1_i64_0:.*]] = arith.constant 1 : i64
+! CHECK: %[[VAL_7:.*]] = fir.convert %[[VAL_c1_i64_0]] : (i64) -> index
+! CHECK: %[[VAL_c1_i64_1:.*]] = arith.constant 1 : i64
+! CHECK: %[[VAL_8:.*]] = fir.convert %[[VAL_c1_i64_1]] : (i64) -> index
+! CHECK: %[[VAL_9:.*]] = fir.shape_shift %[[VAL_c1]], %[[VAL_c10]], %[[VAL_cneg10]], %[[VAL_c21]] : (index, index, index, index) -> !fir.shapeshift<2>
+! CHECK: %[[VAL_10:.*]] = fir.slice %[[VAL_3]], %[[VAL_5]], %[[VAL_4]], %[[VAL_6]], %[[VAL_8]], %[[VAL_7]] : (index, index, index, index, index, index) -> !fir.slice<2>
+! CHECK: %[[VAL_11:.*]] = fir.embox %[[VAL_0]](%[[VAL_9]]) [%[[VAL_10]]] : (!fir.ref<!fir.array<10x21xf32>>, !fir.shapeshift<2>, !fir.slice<2>) -> !fir.box<!fir.array<?x?xf32>>
+! CHECK: %[[VAL_12:.*]] = fir.convert %[[VAL_1]] : (!fir.ref<i32>) -> i64
+! CHECK: %[[c0_i64:.*]] = arith.constant 0 : i64
+! CHECK: %[[VAL_13:.*]] = arith.cmpi eq, %[[VAL_12]], %[[c0_i64]] : i64
+! CHECK: %[[VAL_14:.*]] = fir.if %[[VAL_13]] -> (i64) {
+! CHECK: %[[VAL_17:.*]] = fir.convert %[[VAL_11]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
+! CHECK: %[[VAL_19:.*]] = fir.call @_FortranASize(%[[VAL_17]], %{{.*}}, %{{.*}}) : (!fir.box<none>, !fir.ref<i8>, i32) -> i64
+! CHECK: fir.result %[[VAL_19]] : i64
+! CHECK: } else {
+! CHECK: %[[VAL_16:.*]] = fir.load %[[VAL_1]] : !fir.ref<i32>
+! CHECK: %[[VAL_18:.*]] = fir.convert %[[VAL_11]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
+! CHECK: %[[VAL_20:.*]] = fir.call @_FortranASizeDim(%[[VAL_18]], %[[VAL_16]], %{{.*}}, %{{.*}}) : (!fir.box<none>, i32, !fir.ref<i8>, i32) -> i64
+! CHECK: fir.result %[[VAL_20]] : i64
+! CHECK: }
+! CHECK: %[[VAL_21:.*]] = fir.convert %[[VAL_14]] : (i64) -> i32
+! CHECK: fir.store %[[VAL_21]] to %[[VAL_2]] : !fir.ref<i32>
+ iSize = size(a(2:5, -1:1), dim, 8)
+end subroutine size_test
+
+! CHECK-LABEL: func @_QPsize_optional_dim_1(
+! CHECK-SAME: %[[VAL_0:.*]]: !fir.box<!fir.array<?x?xf32>> {fir.bindc_name = "array"},
+! CHECK-SAME: %[[VAL_1:.*]]: !fir.ref<i32> {fir.bindc_name = "dim", fir.optional},
+! CHECK-SAME: %[[VAL_2:.*]]: !fir.ref<i64> {fir.bindc_name = "isize"}) {
+subroutine size_optional_dim_1(array, dim, iSize)
+ real, dimension(:,:) :: array
+ integer, optional :: dim
+ integer(8) :: iSize
+ iSize = size(array, dim, 8)
+! CHECK: %[[VAL_3:.*]] = fir.convert %[[VAL_1]] : (!fir.ref<i32>) -> i64
+! CHECK: %[[VAL_4:.*]] = arith.constant 0 : i64
+! CHECK: %[[VAL_5:.*]] = arith.cmpi eq, %[[VAL_3]], %[[VAL_4]] : i64
+! CHECK: %[[VAL_6:.*]] = fir.if %[[VAL_5]] -> (i64) {
+! CHECK: %[[VAL_9:.*]] = fir.convert %[[VAL_0]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
+! CHECK: %[[VAL_11:.*]] = fir.call @_FortranASize(%[[VAL_9]], %{{.*}}, %{{.*}}) : (!fir.box<none>, !fir.ref<i8>, i32) -> i64
+! CHECK: fir.result %[[VAL_11]] : i64
+! CHECK: } else {
+! CHECK: %[[VAL_12:.*]] = fir.load %[[VAL_1]] : !fir.ref<i32>
+! CHECK: %[[VAL_15:.*]] = fir.convert %[[VAL_0]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
+! CHECK: %[[VAL_17:.*]] = fir.call @_FortranASizeDim(%[[VAL_15]], %[[VAL_12]], %{{.*}}, %{{.*}}) : (!fir.box<none>, i32, !fir.ref<i8>, i32) -> i64
+! CHECK: fir.result %[[VAL_17]] : i64
+! CHECK: }
+! CHECK: fir.store %[[VAL_6]] to %[[VAL_2]] : !fir.ref<i64>
+end subroutine
+
+! CHECK-LABEL: func @_QPsize_optional_dim_2(
+! CHECK-SAME: %[[VAL_0:.*]]: !fir.box<!fir.array<?x?xf32>> {fir.bindc_name = "array"},
+! CHECK-SAME: %[[VAL_1:.*]]: !fir.ref<!fir.box<!fir.ptr<i32>>> {fir.bindc_name = "dim"},
+! CHECK-SAME: %[[VAL_2:.*]]: !fir.ref<i64> {fir.bindc_name = "isize"}) {
+subroutine size_optional_dim_2(array, dim, iSize)
+ real, dimension(:,:) :: array
+ integer, pointer :: dim
+ integer(8) :: iSize
+ iSize = size(array, dim, 8)
+! CHECK: %[[VAL_3:.*]] = fir.load %[[VAL_1]] : !fir.ref<!fir.box<!fir.ptr<i32>>>
+! CHECK: %[[VAL_4:.*]] = fir.box_addr %[[VAL_3]] : (!fir.box<!fir.ptr<i32>>) -> !fir.ptr<i32>
+! CHECK: %[[VAL_5:.*]] = fir.convert %[[VAL_4]] : (!fir.ptr<i32>) -> i64
+! CHECK: %[[VAL_6:.*]] = arith.constant 0 : i64
+! CHECK: %[[VAL_7:.*]] = arith.cmpi eq, %[[VAL_5]], %[[VAL_6]] : i64
+! CHECK: %[[VAL_8:.*]] = fir.if %[[VAL_7]] -> (i64) {
+! CHECK: %[[VAL_11:.*]] = fir.convert %[[VAL_0]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
+! CHECK: %[[VAL_13:.*]] = fir.call @_FortranASize(%[[VAL_11]], %{{.*}}, %{{.*}}) : (!fir.box<none>, !fir.ref<i8>, i32) -> i64
+! CHECK: fir.result %[[VAL_13]] : i64
+! CHECK: } else {
+! CHECK: %[[VAL_14:.*]] = fir.load %[[VAL_4]] : !fir.ptr<i32>
+! CHECK: %[[VAL_17:.*]] = fir.convert %[[VAL_0]] : (!fir.box<!fir.array<?x?xf32>>) -> !fir.box<none>
+! CHECK: %[[VAL_19:.*]] = fir.call @_FortranASizeDim(%[[VAL_17]], %[[VAL_14]], %{{.*}}, %{{.*}}) : (!fir.box<none>, i32, !fir.ref<i8>, i32) -> i64
+! CHECK: fir.result %[[VAL_19]] : i64
+! CHECK: }
+! CHECK: fir.store %[[VAL_8]] to %[[VAL_2]] : !fir.ref<i64>
+end subroutine
More information about the flang-commits
mailing list