On 10/23/2014 07:09 PM, Jakub Hrozek wrote:
On Thu, Oct 23, 2014 at 07:02:56PM +0200, Lukas Slebodnik wrote:
> On (23/10/14 18:38), Jakub Hrozek wrote:
>> On Thu, Oct 23, 2014 at 06:35:07PM +0200, Jakub Hrozek wrote:
>>> On Thu, Oct 23, 2014 at 02:14:03PM +0200, Michal Židek wrote:
>>>> On 10/23/2014 11:36 AM, Lukas Slebodnik wrote:
>>>>> ehlo,
>>>>>
>>>>> src/monitor/monitor.c: In function ‘main’:
>>>>> src/monitor/monitor.c:2953:24: error: ‘monitor’ may be used
uninitialized
>>>>> in this function
[-Werror=maybe-uninitialized]
>>>>> monitor->is_daemon = !opt_interactive;
>>>>> ^
>>>>> cc1: all warnings being treated as errors
>>>>>
>>>>> This warning is caused be change
579e5d4b7a3ca161ea7518b2996905fa22c15995
>>>>> "MONITOR: Allow confdb to be accessed by nonroot user"
>>>>>
>>>>> Patch is attached.
>>>>>
>>>>> LS
>>>>>
>>>
>>> Not sure I understand, under what circumstances can errno be 0 while
>>> chown returns 0?
>>
>> Err, while chown returns -1
> Try to explain it gcc with optimization (-O2), but usually gcc does not lie.
I know, but this code just makes wonder what's going on even now, I'm
pretty sure I will have no idea a year from now...
>
> another way how to suppress warning is to initialize variabl monitor to NULL
That sounds like a more readable option to me.
Well... if it really is the case that gcc somehow reorganizes the
optimized code in a way, that errno set by chown is not catched in
the ret variable, then the initialization on NULL will cause crash
(but yes it will suppress the warning). As I said, I think it is
false positive, but I am in favour of the first solution, (the
one I acked already) because it is safer.
Michal