On 10/20/17 17:09, Paolo Bonzini wrote:
> On 20/10/2017 17:08, Laszlo Ersek wrote:
>> On 10/20/17 13:02, Paolo Bonzini wrote:
>>> On 20/10/2017 12:48, Peter Maydell wrote:
>>>> On 20 October 2017 at 11:12, Paolo Bonzini <pbonz...@redhat.com>
>>>> wrote:
>>>>> GCC 4.9 and newer stopped warning for missing braces around the
>>>>> "universal" C zero initializer {0}. One such initializer sneaked
>>>>> into scsi/qemu-pr-helper.c and is breaking the build with such
>>>>> older GCC versions.
>>>>>
>>>>> Detect the lack of support for the idiom, and disable the warning
>>>>> in that case.
>>>>>
>>>>> Signed-off-by: Paolo Bonzini <pbonz...@redhat.com>
>>>>
>>>> AFAIK "{}" will work and not be warned about anywhere (we
>>>> use it extensively already) so you don't need to fall back to
>>>> memset...
>>>
>>> Doh, of course that will work in scsi/qemu-pr-helper.c. In general
>>> you can use {0} but not {} to initialize a scalar, like
>>>
>>> IDontKnowIfItsAnArrayOrPointer x = {}; //might fail
>>> IDontKnowIfItsAnArrayOrPointer x = {0}; //always works
>>>
>>> I'm not sure if that matters.
>>
>> My remark below might matter even less, but:
>>
>> I'd find it regrettable if we suppressed a wrong gcc warning about a
>> valid C construct by replacing the construct with a GNU-ism that is
>> *not* standard C. :/
>
> Is {} to initialize structs a GNUism?
It is;
> struct {
> int a, b;
> } x = {};
$ gcc -fsyntax-only -std=c99 -pedantic -Wall -Wextra x.c
> x.c:3:7: warning: ISO C forbids empty initializer braces [-Wpedantic]
> } x = {};
> ^
> x.c:3:1: warning: missing initializer for field 'a' of 'struct <anonymous>'
> [-Wmissing-field-initializers]
> } x = {};
> ^
> x.c:2:7: note: 'a' declared here
> int a, b;
> ^
If you replace "{}" with "{ 0 }", the same command line stays quiet.
(I used: gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-16).)
> (Initializing a scalar with {} is a hard error with GCC).
Thanks,
Laszlo
