Re: [PATCH v3 04/14] lib/vsprintf.c: expand field_width to 24 bits

[Rasmus Villemoes]

On Fri, Dec 04 2015, Joe Perches <joe@xxxxxxxxxxx> wrote:

> On Thu, 2015-12-03 at 15:34 -0800, Andrew Morton wrote:
>> I've been fiddling with a BUILD_BUG_ON which works outside functions
>> using gcc's __COUNTER__ - something like
>> 
>> #define BBO(expr) typedef char __bbo##__COUNTER__[1-2*(!!expr)]
>
> nit: Âyou need another parenthesis around expr
>
>> BBO(1 == 1);
>> BBO(2 == 2);
>> 
>> but that comes out as
>> 
>> typedef char __bbo__COUNTER__[1-2*(!!1 == 1)];
>> typedef char __bbo__COUNTER__[1-2*(!!2 == 2)];
>> 
>> instead of
>> 
>> typedef char __bbo0[1-2*(!!1 == 1)];
>> typedef char __bbo1[1-2*(!!2 == 2)];
>> 
>> There's some trick here but I've forgotten what it is.
>
> I believe it's something like:
>
> #define __stringify_2(a, b)	a##b
> #define __stringify2(a, b)	__stringify_2(a, b)
>
> #define BBO(expr) typedef char __stringify2(bbo, __COUNTER__)[1 - 2*(!!(expr))]

Let's at least not reinvent two wheels. __UNIQUE_ID exists and does the
gluing (which we have __PASTE for, not stringify) etc., and uses
__LINE__ as a poor man's fallback for compilers without __COUNTER__ (gcc
< 4.3).

But I don't see why we even need the unique identifier. What's wrong
with 'extern char blabla[1 - 2*(!!(expr))]'? blabla can be declared
multiple times without problems - and when it fails, we either get a
'negative size' error or at least some complaint about conflicting
declarations. Maybe stick a __always_unused in to prevent gcc from
complaining if this declaration is inside a function.

Rasmus
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/