[PATCH] D89360: Treat constant contexts as being in the default rounding mode.

Richard Smith - zygoloid via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Tue Oct 13 19:19:38 PDT 2020


rsmith created this revision.
rsmith added reviewers: rjmccall, mibintc, sepavloff.
Herald added a project: clang.
rsmith requested review of this revision.

This addresses a regression where pretty much all C++ compilations using
-frounding-math now fail, due to rounding being performed in constexpr
function definitions in the standard library.

This follows the "manifestly constant evaluated" approach described in
https://reviews.llvm.org/D87528#2270676 -- evaluations that are required
to succeed at compile time are permitted even in regions with dynamic
rounding modes, as are (unfortunately) the evaluation of the
initializers of local variables of const integral types.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D89360

Files:
  clang/lib/AST/ExprConstant.cpp
  clang/test/SemaCXX/rounding-math.cpp


Index: clang/test/SemaCXX/rounding-math.cpp
===================================================================
--- /dev/null
+++ clang/test/SemaCXX/rounding-math.cpp
@@ -0,0 +1,36 @@
+// RUN: %clang_cc1 -triple x86_64-linux -verify=norounding %s
+// RUN: %clang_cc1 -triple x86_64-linux -verify=rounding %s -frounding-math
+// rounding-no-diagnostics
+
+#define fold(x) (__builtin_constant_p(x) ? (x) : (x))
+
+constexpr double a = 1.0 / 3.0;
+
+constexpr int f(int n) { return int(n * (1.0 / 3.0)); }
+
+using T = int[f(3)];
+using T = int[1];
+
+struct Bitfield { unsigned int n : 1; };
+
+void f(Bitfield &b) {
+  b.n = int(6 * (1.0 / 3.0)); // norounding-warning {{changes value from 2 to 0}}
+}
+
+const int k = 3 * (1.0 / 3.0);
+static_assert(k == 1, "");
+
+void g() {
+  // FIXME: Constant-evaluating this initializer is surprising, and violates
+  // the recommended practice in C++ [expr.const]p12:
+  //
+  //   Implementations should provide consistent results of floating-point
+  //   evaluations, irrespective of whether the evaluation is performed during
+  //   translation or during program execution.
+  const int k = 3 * (1.0 / 3.0);
+  static_assert(k == 1, "");
+}
+
+int *h() {
+  return new int[int(-3 * (1.0 / 3.0))]; // norounding-error {{too large}}
+}
Index: clang/lib/AST/ExprConstant.cpp
===================================================================
--- clang/lib/AST/ExprConstant.cpp
+++ clang/lib/AST/ExprConstant.cpp
@@ -2528,6 +2528,11 @@
 /// Check if the given evaluation result is allowed for constant evaluation.
 static bool checkFloatingPointResult(EvalInfo &Info, const Expr *E,
                                      APFloat::opStatus St) {
+  // In a constant context, assume that any dynamic rounding mode or FP
+  // exception state matches the default floating-point environment.
+  if (Info.InConstantContext)
+    return true;
+
   FPOptions FPO = E->getFPFeaturesInEffect(Info.Ctx.getLangOpts());
   if ((St & APFloat::opInexact) &&
       FPO.getRoundingMode() == llvm::RoundingMode::Dynamic) {


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D89360.298015.patch
Type: text/x-patch
Size: 2051 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20201014/35fe400b/attachment.bin>


More information about the cfe-commits mailing list