[flang] Fix length handling in character kind implicit conversion

[jeanPerier]

When assigning to a whole allocatable, lowering is dealing with the implicit conversion to preserve the RHS lower bounds. In case of character KIND mismatch, the code was setting the new RHS length to the one from the LHS, which is wrong for two reasons:

• no padding/truncation was actually done in the conversion
• the RHS length should anyway not be touched since the one from the allocatable LHS may change to become the one of the RHS.

Update the code to preserve the RHS type length when materializing the implicit character KIND conversion.

[llvmbot]

@llvm/pr-subscribers-flang-fir-hlfir

Author: None (jeanPerier)

Changes

When assigning to a whole allocatable, lowering is dealing with the implicit conversion to preserve the RHS lower bounds. In case of character KIND mismatch, the code was setting the new RHS length to the one from the LHS, which is wrong for two reasons:

• no padding/truncation was actually done in the conversion
• the RHS length should anyway not be touched since the one from the allocatable LHS may change to become the one of the RHS.

Update the code to preserve the RHS type length when materializing the implicit character KIND conversion.

---

Full diff: https://github.com/llvm/llvm-project/pull/74586.diff

2 Files Affected:

• (modified) flang/lib/Optimizer/Builder/HLFIRTools.cpp (+6-1)
• (modified) flang/test/Lower/HLFIR/implicit-type-conversion-allocatable.f90 (+26)

diff --git a/flang/lib/Optimizer/Builder/HLFIRTools.cpp b/flang/lib/Optimizer/Builder/HLFIRTools.cpp
index 0b1e57e8f6c37..17efa45b8667d 100644
--- a/flang/lib/Optimizer/Builder/HLFIRTools.cpp
+++ b/flang/lib/Optimizer/Builder/HLFIRTools.cpp
@@ -1132,8 +1132,13 @@ hlfir::genTypeAndKindConvert(mlir::Location loc, fir::FirOpBuilder &builder,
   std::optional<int> toKindCharConvert;
   if (auto toCharTy = mlir::dyn_cast<fir::CharacterType>(toType)) {
     if (auto fromCharTy = mlir::dyn_cast<fir::CharacterType>(fromType))
-      if (toCharTy.getFKind() != fromCharTy.getFKind())
+      if (toCharTy.getFKind() != fromCharTy.getFKind()) {
         toKindCharConvert = toCharTy.getFKind();
+        // Preserve source length (padding/truncation will occur in assignment
+        // if needed).
+        toType = fir::CharacterType::get(
+            fromType.getContext(), toCharTy.getFKind(), fromCharTy.getLen());
+      }
     // Do not convert in case of character length mismatch only, hlfir.assign
     // deals with it.
     if (!toKindCharConvert)
diff --git a/flang/test/Lower/HLFIR/implicit-type-conversion-allocatable.f90 b/flang/test/Lower/HLFIR/implicit-type-conversion-allocatable.f90
index 7083a825dfd3b..361cd61adea2d 100644
--- a/flang/test/Lower/HLFIR/implicit-type-conversion-allocatable.f90
+++ b/flang/test/Lower/HLFIR/implicit-type-conversion-allocatable.f90
@@ -38,3 +38,29 @@ subroutine preserve_lbounds(x, y)
 ! CHECK:           hlfir.destroy %[[VAL_8]] : !hlfir.expr<?xi32>
 ! CHECK:           return
 ! CHECK:         }
+
+! Test that RHS character length is preserved in a character KIND
+! conversion before the assignment.
+subroutine kind_and_length(a, b)
+  character(len=4,kind=4), allocatable :: a(:)
+  character(len=2,kind=1) :: b(:)
+  a = b
+end subroutine
+! CHECK-LABEL:   func.func @_QPkind_and_length(
+! CHECK:         %[[VAL_2:.*]] = arith.constant 4 : index
+! CHECK:         %[[VAL_3:.*]]:2 = hlfir.declare %[[VAL_0:[a-z0-9]*]] typeparams %[[VAL_2:[a-z0-9]*]]  {{.*}}Ea
+! CHECK:         %[[VAL_4:.*]] = arith.constant 2 : index
+! CHECK:         %[[VAL_5:.*]]:2 = hlfir.declare %[[VAL_1:[a-z0-9]*]] typeparams %[[VAL_4:[a-z0-9]*]]  {{.*}}Eb
+! CHECK:         %[[VAL_6:.*]] = arith.constant 0 : index
+! CHECK:         %[[VAL_7:.*]]:3 = fir.box_dims %[[VAL_5]]#0, %[[VAL_6]] : (!fir.box<!fir.array<?x!fir.char<1,2>>>, index) -> (index, index, index)
+! CHECK:         %[[VAL_8:.*]] = fir.shape %[[VAL_7]]#1 : (index) -> !fir.shape<1>
+! CHECK:         %[[VAL_9:.*]] = hlfir.elemental %[[VAL_8]] typeparams %[[VAL_4]] unordered : (!fir.shape<1>, index) -> !hlfir.expr<?x!fir.char<4,2>> {
+! CHECK:         ^bb0(%[[VAL_10:.*]]: index):
+! CHECK:           %[[VAL_11:.*]] = hlfir.designate %[[VAL_5]]#0 (%[[VAL_10]])  typeparams %[[VAL_4]] : (!fir.box<!fir.array<?x!fir.char<1,2>>>, index, index) -> !fir.ref<!fir.char<1,2>>
+! CHECK:           %[[VAL_12:.*]] = fir.alloca !fir.char<4,?>(%[[VAL_4]] : index)
+! CHECK:           fir.char_convert %[[VAL_11]] for %[[VAL_4]] to %[[VAL_12]] : !fir.ref<!fir.char<1,2>>, index, !fir.ref<!fir.char<4,?>>
+! CHECK:           %[[VAL_13:.*]]:2 = hlfir.declare %[[VAL_12:[a-z0-9]*]] typeparams %[[VAL_4:[a-z0-9]*]] {uniq_name = ".temp.kindconvert"} : (!fir.ref<!fir.char<4,?>>, index) -> (!fir.boxchar<4>, !fir.ref<!fir.char<4,?>>)
+! CHECK:           hlfir.yield_element %[[VAL_13]]#0 : !fir.boxchar<4>
+! CHECK:         }
+! CHECK:         hlfir.assign %[[VAL_9]] to %[[VAL_3]]#0 realloc keep_lhs_len : !hlfir.expr<?x!fir.char<4,2>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?x!fir.char<4,4>>>>>
+

[psteinfeld]

All builds and tests correctly and looks good.

I wonder about the update message, which discusses setting the new RHS length. Should this say the new LHS length?

[jeanPerier]

I wonder about the update message, which discusses setting the new RHS length. Should this say the new LHS length?

Thanks for the review! The "new RHS" is a new value converted from the RHS of the original fortran code. The updated code is producing this "new RHS" and was incorrectly setting its MLIR type length to the one of the assignment LHS (instead of preserving the one from the original RHS).