[flang-commits] [flang] 982527e - [flang] Use saturated intrinsics for floating point to integer conversions (#130686)

[via flang-commits]

Author: Asher Mancinelli
Date: 2025-03-12T08:14:46-07:00
New Revision: 982527eef0f284161dd5e09a9b4fc952b332ec92

URL: https://github.com/llvm/llvm-project/commit/982527eef0f284161dd5e09a9b4fc952b332ec92
DIFF: https://github.com/llvm/llvm-project/commit/982527eef0f284161dd5e09a9b4fc952b332ec92.diff

LOG: [flang] Use saturated intrinsics for floating point to integer conversions (#130686)

The saturated floating point conversion intrinsics match the semantics in the standard more closely than the fptosi/fptoui instructions.

Case 2 of 16.9.100 is

> INT (A [, KIND])
> If A is of type real, there are two cases: if |A| < 1, INT (A) has the
value 0; if |A| ≥ 1, INT (A) is the integer whose magnitude is the
largest integer that does not exceed the magnitude of A and whose sign
is the same as the sign of A.

Currently, converting a floating point value into an integer type too
small to hold the constant will be converted to poison in opt, leaving
us with garbage:

```
> cat t.f90
program main
  real(kind=16)   :: f
  integer(kind=4) :: i
  f=huge(f)
  i=f
  print *, i
end program main

# current upstream
> for i in `seq 10`; do; ./a.out; done
 -862156992
 -1497393344
 -739096768
 -1649494208
 1761228608
 -1959270592
 -746244288
 -1629194432
 -231217344
 382322496
```

With the saturated fptoui/fptosi intrinsics, we get the appropriate
values

```
# mine
> flang -O2 ./t.f90 && ./a.out
 2147483647

> perl -e 'printf "%d\n", (2 ** 31) - 1'
2147483647
```

One notable difference: NaNs being converted to ints will become zero, unlike current flang (and some other compilers). Newer versions of GCC have this behavior.

Added: 
    

Modified: 
    flang/lib/Optimizer/CodeGen/CodeGen.cpp
    flang/test/Fir/convert-to-llvm.fir

Removed: 
    


################################################################################
diff  --git a/flang/lib/Optimizer/CodeGen/CodeGen.cpp b/flang/lib/Optimizer/CodeGen/CodeGen.cpp
index 5548f5f981b12..2cb4cea58c2b0 100644
--- a/flang/lib/Optimizer/CodeGen/CodeGen.cpp
+++ b/flang/lib/Optimizer/CodeGen/CodeGen.cpp
@@ -835,10 +835,20 @@ struct ConvertOpConversion : public fir::FIROpConversion<fir::ConvertOp> {
         return mlir::success();
       }
       if (mlir::isa<mlir::IntegerType>(toTy)) {
-        if (toTy.isUnsignedInteger())
-          rewriter.replaceOpWithNewOp<mlir::LLVM::FPToUIOp>(convert, toTy, op0);
-        else
-          rewriter.replaceOpWithNewOp<mlir::LLVM::FPToSIOp>(convert, toTy, op0);
+        // NOTE: We are checking the fir type here because toTy is an LLVM type
+        // which is signless, and we need to use the intrinsic that matches the
+        // sign of the output in fir.
+        if (toFirTy.isUnsignedInteger()) {
+          auto intrinsicName =
+              mlir::StringAttr::get(convert.getContext(), "llvm.fptoui.sat");
+          rewriter.replaceOpWithNewOp<mlir::LLVM::CallIntrinsicOp>(
+              convert, toTy, intrinsicName, op0);
+        } else {
+          auto intrinsicName =
+              mlir::StringAttr::get(convert.getContext(), "llvm.fptosi.sat");
+          rewriter.replaceOpWithNewOp<mlir::LLVM::CallIntrinsicOp>(
+              convert, toTy, intrinsicName, op0);
+        }
         return mlir::success();
       }
     } else if (mlir::isa<mlir::IntegerType>(fromTy)) {

diff  --git a/flang/test/Fir/convert-to-llvm.fir b/flang/test/Fir/convert-to-llvm.fir
index c7037019ee701..2960528fb6c24 100644
--- a/flang/test/Fir/convert-to-llvm.fir
+++ b/flang/test/Fir/convert-to-llvm.fir
@@ -701,6 +701,10 @@ func.func @convert_from_float(%arg0 : f32) {
   %7 = fir.convert %arg0 : (f32) -> i16
   %8 = fir.convert %arg0 : (f32) -> i32
   %9 = fir.convert %arg0 : (f32) -> i64
+  %10 = fir.convert %arg0 : (f32) -> ui8
+  %11 = fir.convert %arg0 : (f32) -> ui16
+  %12 = fir.convert %arg0 : (f32) -> ui32
+  %13 = fir.convert %arg0 : (f32) -> ui64
   return
 }
 
@@ -711,11 +715,15 @@ func.func @convert_from_float(%arg0 : f32) {
 // CHECK:         %{{.*}} = llvm.fpext %[[ARG0]] : f32 to f64
 // CHECK:         %{{.*}} = llvm.fpext %[[ARG0]] : f32 to f80
 // CHECK:         %{{.*}} = llvm.fpext %[[ARG0]] : f32 to f128
-// CHECK:         %{{.*}} = llvm.fptosi %[[ARG0]] : f32 to i1
-// CHECK:         %{{.*}} = llvm.fptosi %[[ARG0]] : f32 to i8
-// CHECK:         %{{.*}} = llvm.fptosi %[[ARG0]] : f32 to i16
-// CHECK:         %{{.*}} = llvm.fptosi %[[ARG0]] : f32 to i32
-// CHECK:         %{{.*}} = llvm.fptosi %[[ARG0]] : f32 to i64
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptosi.sat"(%[[ARG0]]) : (f32) -> i1
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptosi.sat"(%[[ARG0]]) : (f32) -> i8
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptosi.sat"(%[[ARG0]]) : (f32) -> i16
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptosi.sat"(%[[ARG0]]) : (f32) -> i32
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptosi.sat"(%[[ARG0]]) : (f32) -> i64
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptoui.sat"(%[[ARG0]]) : (f32) -> i8
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptoui.sat"(%[[ARG0]]) : (f32) -> i16
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptoui.sat"(%[[ARG0]]) : (f32) -> i32
+// CHECK:         %{{.*}} = llvm.call_intrinsic "llvm.fptoui.sat"(%[[ARG0]]) : (f32) -> i64
 
 // -----