Why "'X' used but never defined" is a warning and not error in gcc?

Why "'X' used but never defined" is a warning and not error in gcc?

[Ilya Basin]

test.c:
    static void foo();
    
    void bar() {
        foo();
    }

$ gcc -c test.c
test.c:1: warning: 'foo' used but never defined

Why warning and not error? Another *.o can refer this static function?

Re: Why "'X' used but never defined" is a warning and not error in gcc?

[Oleg Endo]

On Mon, 2012-10-08 at 14:19 +0400, Ilya Basin wrote:
> test.c:
>     static void foo();
>     
>     void bar() {
>         foo();
>     }
> 
> $ gcc -c test.c
> test.c:1: warning: 'foo' used but never defined
> 
> Why warning and not error? Another *.o can refer this static function?
> 

No, another .o can't refer to this static function.  Static functions
are visible only in the translation unit where they are defined.

The code above will generate a symbol reference to 'foo', i.e. a
function call to a non-static function.  If it is defined in some
other .o and linked together, the 'foo' from the other .o will be used.

Cheers,
Oleg

Re[2]: Why "'X' used but never defined" is a warning and not error in gcc?

[Ilya Basin]

OE> On Mon, 2012-10-08 at 14:19 +0400, Ilya Basin wrote:
>> test.c:
>>     static void foo();
>>     
>>     void bar() {
>>         foo();
>>     }
>> 
>> $ gcc -c test.c
>> test.c:1: warning: 'foo' used but never defined
>> 
>> Why warning and not error? Another *.o can refer this static function?
>> 

OE> No, another .o can't refer to this static function.  Static functions
OE> are visible only in the translation unit where they are defined.

OE> The code above will generate a symbol reference to 'foo', i.e. a
OE> function call to a non-static function.  If it is defined in some
OE> other .o and linked together, the 'foo' from the other .o will be used.

OE> Cheers,
OE> Oleg


Is there a flag to turn this warning into an error? Is there a common
way to find the warning flag from a message?


-- 

Re: Re[2]: Why "'X' used but never defined" is a warning and not error in gcc?

[Oleg Endo]

On Mon, 2012-10-08 at 14:32 +0400, Ilya Basin wrote:
> OE> On Mon, 2012-10-08 at 14:19 +0400, Ilya Basin wrote:
> >> test.c:
> >>     static void foo();
> >>     
> >>     void bar() {
> >>         foo();
> >>     }
> >> 
> >> $ gcc -c test.c
> >> test.c:1: warning: 'foo' used but never defined
> >> 
> >> Why warning and not error? Another *.o can refer this static function?
> >> 
> 
> OE> No, another .o can't refer to this static function.  Static functions
> OE> are visible only in the translation unit where they are defined.
> 
> OE> The code above will generate a symbol reference to 'foo', i.e. a
> OE> function call to a non-static function.  If it is defined in some
> OE> other .o and linked together, the 'foo' from the other .o will be used.
> 
> OE> Cheers,
> OE> Oleg
> 
> 
> Is there a flag to turn this warning into an error? 

Doesn't look so.  You can turn all warnings into errors by -Werror, but
there's no option to control this particular warning individually.

> Is there a common
> way to find the warning flag from a message?

grep ;)

Cheers,
Oleg

Re: Re[2]: Why "'X' used but never defined" is a warning and not error in gcc?

[Jonathan Wakely]

On 8 October 2012 11:32, Ilya Basin wrote:
> Is there a flag to turn this warning into an error? Is there a common
> way to find the warning flag from a message?

If you use a modern version of GCC it tells you the flag that controls
each warning.

In this case it tells you it is "enabled by default" so there is no
flag controlling this warning.

Re: Re[2]: Why "'X' used but never defined" is a warning and not error in gcc?

[Vincent Lefevre]

On 2012-10-08 12:43:09 +0200, Oleg Endo wrote:
> Doesn't look so.  You can turn all warnings into errors by -Werror, but
> there's no option to control this particular warning individually.

It is possible to turn *particular* warnings into errors, not just
all warnings, with -Werror=<list_of_warnings>.

-- 
Vincent Lefèvre <vincent@vinc17.net> - Web: <http://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <http://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Re: Re[2]: Why "'X' used but never defined" is a warning and not error in gcc?

[Oleg Endo]

On Mon, 2012-10-08 at 17:30 +0200, Vincent Lefevre wrote:
> On 2012-10-08 12:43:09 +0200, Oleg Endo wrote:
> > Doesn't look so.  You can turn all warnings into errors by -Werror, but
> > there's no option to control this particular warning individually.
> 
> It is possible to turn *particular* warnings into errors, not just
> all warnings, with -Werror=<list_of_warnings>.

Yes, that is true.  But there is no way to turn off or turn into an
error the warning that is output in _this_particular_case_, because
there's no flag in GCC's code to do so.

gcc/toplev.c, function check_global_declaration_1, around line 470:

&& (warn_unused_function
    || TREE_SYMBOL_REFERENCED (DECL_ASSEMBLER_NAME (decl))))
  {
    if (TREE_SYMBOL_REFERENCED (DECL_ASSEMBLER_NAME (decl)))
      pedwarn (input_location, 0, "%q+F used but never defined", decl);

This particular warning can be turned into an error by specifying
-pedantic-errors

Cheers,
Oleg