Re: [PATCH for Xen 4.5] xen/arm: Add support for GICv3 for domU

[Julien Grall <julien.grall@linaro.org>]

On 11/18/2014 04:15 PM, Jan Beulich wrote:
>>>> On 18.11.14 at 16:00, <julien.grall@linaro.org> wrote:
>> On 10/31/2014 09:02 AM, Jan Beulich wrote:
>>>>>> On 30.10.14 at 19:51, <julien.grall@linaro.org> wrote:
>>> The naming suggests that the #if really should be around just the
>>> gic_version field (with a dummy field in the #else case to be C89
>>> compatible, e.g. a zero width unnamed bitfield) and the
>>> corresponding #define-s above, ...
>>
>> Not really related to this patch... but the way to improve it (via
>> extending createdomain).
>>
>> I need to create an empty structure. Is the dummy field really needed?
>> If so, did you meant?
>>
>> struct
>> {
>>    int :0;
>> }
> 
> Yes.
> 
>> The C spec declare this kind of structure as undefined.
> 
> I can't find anything saying so.

http://c0x.coding-guidelines.com/6.7.2.1.html

"1401 If the struct-declaration-list contains no named members, the
behavior is undefined."

>> Would an empty structure and used it be better?
> 
> Empty structures (and unions) aren't valid in standard C afaics, up to
> and including C11. That was the whole point of suggesting the above
> alternative, with me (maybe wrongly) believing that this would be valid.

Right, this is an extension of GCC. As neither of the 2 solutions are
valid, Ian Jackson was suggesting to use

struct {
   char dummy;
}

Would it be ok for you?

Regards,

-- 
Julien Grall