[llvm-branch-commits] [clang-tools-extra] [clang-tidy] fix wrong float to float conversion check when floating point type is not standard type (PR #122637)

[via llvm-branch-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-clang-tidy

Author: Congcong Cai (HerrCai0907)

<details>
<summary>Changes</summary>

compare type kind is the wrong way to compare floating point type compatibility.
more generic compatibility check is needed.

---
Full diff: https://github.com/llvm/llvm-project/pull/122637.diff


3 Files Affected:

- (modified) clang-tools-extra/clang-tidy/bugprone/NarrowingConversionsCheck.cpp (+3-1) 
- (modified) clang-tools-extra/docs/ReleaseNotes.rst (+5) 
- (modified) clang-tools-extra/test/clang-tidy/checkers/bugprone/narrowing-conversions-narrowingfloatingpoint-option.cpp (+9) 


``````````diff
diff --git a/clang-tools-extra/clang-tidy/bugprone/NarrowingConversionsCheck.cpp b/clang-tools-extra/clang-tidy/bugprone/NarrowingConversionsCheck.cpp
index 408390ebc70b64..bafcd402ca8510 100644
--- a/clang-tools-extra/clang-tidy/bugprone/NarrowingConversionsCheck.cpp
+++ b/clang-tools-extra/clang-tidy/bugprone/NarrowingConversionsCheck.cpp
@@ -513,7 +513,9 @@ void NarrowingConversionsCheck::handleFloatingCast(const ASTContext &Context,
       return;
     }
     const BuiltinType *FromType = getBuiltinType(Rhs);
-    if (ToType->getKind() < FromType->getKind())
+    if (!llvm::APFloatBase::isRepresentableBy(
+            Context.getFloatTypeSemantics(FromType->desugar()),
+            Context.getFloatTypeSemantics(ToType->desugar())))
       diagNarrowType(SourceLoc, Lhs, Rhs);
   }
 }
diff --git a/clang-tools-extra/docs/ReleaseNotes.rst b/clang-tools-extra/docs/ReleaseNotes.rst
index 835a0269a2733c..f709902cfd45e7 100644
--- a/clang-tools-extra/docs/ReleaseNotes.rst
+++ b/clang-tools-extra/docs/ReleaseNotes.rst
@@ -206,6 +206,11 @@ Changes in existing checks
   <clang-tidy/checks/bugprone/forwarding-reference-overload>` check by fixing
   a crash when determining if an ``enable_if[_t]`` was found.
 
+- Improve :doc:`bugprone-narrowing-conversions
+  <clang-tidy/checks/bugprone/narrowing-conversions>` to avoid incorrect check
+  results when floating point type is not ``float``, ``double`` and
+  ``long double``.
+
 - Improved :doc:`bugprone-optional-value-conversion
   <clang-tidy/checks/bugprone/optional-value-conversion>` to support detecting
   conversion directly by ``std::make_unique`` and ``std::make_shared``.
diff --git a/clang-tools-extra/test/clang-tidy/checkers/bugprone/narrowing-conversions-narrowingfloatingpoint-option.cpp b/clang-tools-extra/test/clang-tidy/checkers/bugprone/narrowing-conversions-narrowingfloatingpoint-option.cpp
index 9ded2f0923f4e6..180b789e45bb37 100644
--- a/clang-tools-extra/test/clang-tidy/checkers/bugprone/narrowing-conversions-narrowingfloatingpoint-option.cpp
+++ b/clang-tools-extra/test/clang-tidy/checkers/bugprone/narrowing-conversions-narrowingfloatingpoint-option.cpp
@@ -36,6 +36,15 @@ void narrow_double_to_float_not_ok(double d) {
   f = narrow_double_to_float_return();
 }
 
+float narrow_float16_to_float_return(_Float16 f) {
+  return f;
+}
+
+_Float16 narrow_float_to_float16_return(float f) {
+  return f;
+  // CHECK-MESSAGES: :[[@LINE-1]]:10: warning: narrowing conversion from 'float' to '_Float16' [bugprone-narrowing-conversions]
+}
+
 void narrow_fp_constants() {
   float f;
   f = 0.5; // [dcl.init.list] 7.2 : in-range fp constant to narrower float is not a narrowing.

``````````

</details>


https://github.com/llvm/llvm-project/pull/122637