Re: [PATCH v2 5/5] Linux: Consolidate auxiliary vector parsing

[Adhemerval Zanella <adhemerval.zanella@linaro.org>]

On 10/02/2022 07:54, Florian Weimer wrote:
> * Adhemerval Zanella:
> 
>>>>> +  auxv_values[AT_PAGESZ] = EXEC_PAGESIZE;
>>>>> +  auxv_values[AT_FPUCW] = _FPU_DEFAULT;
>>>>> +
>>>>> +  /* NB: Default to a constant CONSTANT_MINSIGSTKSZ.  */
>>>>> +  _Static_assert (__builtin_constant_p (CONSTANT_MINSIGSTKSZ),
>>>>> +                  "CONSTANT_MINSIGSTKSZ is constant");
>>>>
>>>> Shouldn't it warn that CONSTANT_MINSIGSTKSZ is not a constant?
>>>
>>> Sorry, would please rephrase?  Should I change this to
>>> ""CONSTANT_MINSIGSTKSZ is not constant"?
>>
>> Yeah because on a the _Static_assert failure the resulting message seems
>> to the opposite of what triggered it.
> 
> The GCC error message doesn't suggest a strong preference either way:
> 
> /tmp/t.c:1:1: error: static assertion failed: "failed"
>     1 | _Static_assert (0, "failed");
>       | ^~~~~~~~~~~~~~
> 
> I think in the existing sources, the message is sometimes indicating the
> failure, or what is being asserted (as is the case here).

Compiler is will just emit the message if condition is met, my point is
emitting 'CONSTANT_MINSIGSTKSZ is constant' on an error is the opposite
of what triggered it (CONSTANT_MINSIGSTKSZ is really *not* a constant).