================
@@ -13861,6 +13897,36 @@ Sema::CheckReturnValExpr(Expr *RetValExp, QualType 
lhsType,
     CheckPPCMMAType(RetValExp->getType(), ReturnLoc);
 }
 
+/// Diagnose comparison to NAN or INFINITY in fast math modes.
+/// The comparison to NaN or INFINITY is always false in
+/// fast modes: float evaluation will not result in inf or nan.
+void Sema::CheckInfNaNFloatComparison(SourceLocation Loc, Expr *LHS, Expr *RHS,
----------------
AaronBallman wrote:

The basic idea I have is that it's UB to generate a constant infinity or NaN if 
infinity/NaN is disabled. Instead of performing special logic for float 
comparisons, I think the logic should live in `Sema::ActOnNumericConstant()` so 
that we capture any time a float or infinity constant is used. e.g. `inf + 1` 
should also diagnose.

Then the tautological comparison component to this could be implemented by 
looking for calls to `__builtin_isinf` or `__builtin_isnan` and claim those are 
tautologically false (as a follow-up patch). WDYT?

https://github.com/llvm/llvm-project/pull/76873
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to