[clang][Interp] Note UB when converting Inf to integer

The F.isFinite() check here was introduced back when the first
floating point implementation was pushed, but I don't think it's
necessary and it fixes the attached test case.

Author: tbaederr

clang/lib/AST/Interp/Interp.h

@@ -1680,7 +1680,7 @@ bool CastFloatingIntegral(InterpState &S, CodePtr OpPC) {
     auto Status = F.convertToInteger(Result);
 
     // Float-to-Integral overflow check.
-    if ((Status & APFloat::opStatus::opInvalidOp) && F.isFinite()) {
+    if ((Status & APFloat::opStatus::opInvalidOp)) {
       const Expr *E = S.Current->getExpr(OpPC);
       QualType Type = E->getType();
 

clang/test/AST/Interp/cxx20.cpp

@@ -765,3 +765,12 @@ namespace FailingDestructor {
     f<D{0, false}>(); // both-error {{no matching function}}
   }
 }
+
+
+void overflowInSwitchCase(int n) {
+  switch (n) {
+  case (int)(float)1e300: // both-error {{constant expression}} \
+                          // both-note {{value +Inf is outside the range of representable values of type 'int'}}
+    break;
+  }
+}