Re: [C PATCH] Don't leak C_MAYBE_CONST_EXPRs into fold() (PR c/68412)

[Bernd Schmidt <bschmidt at redhat dot com>]

On 11/18/2015 05:16 PM, Marek Polacek wrote:
> Actually, no, I think we should do this instead.
>
> +++ gcc/c-family/c-common.c
> @@ -1924,7 +1924,7 @@ warn_tautological_cmp (location_t loc, enum tree_code code, tree lhs, tree rhs)
>
>     /* We do not warn for constants because they are typical of macro
>        expansions that test for features, sizeof, and similar.  */
> -  if (CONSTANT_CLASS_P (fold (lhs)) || CONSTANT_CLASS_P (fold (rhs)))
> +  if (CONSTANT_CLASS_P (lhs) || CONSTANT_CLASS_P (rhs))
>       return;
>
>     /* Don't warn for e.g.
> diff --git gcc/cp/call.c gcc/cp/call.c
> index 8cdda62..77c2936 100644
> --- gcc/cp/call.c
> +++ gcc/cp/call.c
> @@ -5741,7 +5741,7 @@ build_new_op_1 (location_t loc, enum tree_code code, int flags, tree arg1,
>   	maybe_warn_bool_compare (loc, code, fold (arg1),
>   				 fold (arg2));
>         if (complain & tf_warning && warn_tautological_compare)
> -	warn_tautological_cmp (loc, code, arg1, arg2);
> +	warn_tautological_cmp (loc, code, fold (arg1), fold (arg2));
>         /* Fall through.  */
>       case PLUS_EXPR:
>       case MINUS_EXPR:

That seems to change the behaviour of the code though. Most of the code 
in warn_tautological_cmp only looks at the unfolded form. Are you sure 
this is safe, for example wrt. the from_macro_expansion_at tests or the 
CONVERT_EXPR_P ones?


Bernd