[PATCH v4 03/14] modpost: split the section mismatch checks into section-check.c

[Jeff Johnson]

On 5/11/2022 12:27 PM, Masahiro Yamada wrote:
> On Thu, May 12, 2022 at 3:48 AM Nick Desaulniers
> <ndesaulniers at google.com> wrote:
>>
>> On Mon, May 9, 2022 at 11:57 PM Masahiro Yamada <masahiroy at kernel.org> wrote:
>>>
>>>>> diff --git a/scripts/mod/modpost.c b/scripts/mod/modpost.c
>>>>> index a78b75f0eeb0..e7e2c70a98f5 100644
>>>>> --- a/scripts/mod/modpost.c
>>>>> +++ b/scripts/mod/modpost.c
>>>>> @@ -31,7 +31,7 @@ static bool external_module;
>>>>>   /* Only warn about unresolved symbols */
>>>>>   static bool warn_unresolved;
>>>>>
>>>>> -static int sec_mismatch_count;
>>>>> +int sec_mismatch_count;
>>>>
>>>> ^ this should go in modpost.h if it is to be used in two translation
>>>> units, rather than forward declaring it in section-check.c.  You did
>>>> this for the functions.
>>>
>>>
>>> Sorry, I do not understand.
>>>
>>>
>>> In modpost.h, I put the declaration:
>>>
>>>    extern int sec_mismatch_count;
>>>
>>> If I moved it to the header without 'extern'
>>> I would get multiple definitions.
>>
>> Yeah, you need to _declare_ it w/ extern in the header, then _define_
>> it in one source file.
>>
>> That way, if the type ever changes, the sources will agree on type in
>> all source files. You will get a redefinition error if the definition
>> changes the type of the variable since the last declaration.
>>
>> What you're doing is forward declaring, which works, and is a common
>> pattern for (bloated) C++, but is less type safe than sharing a single
>> common declaration between multiple source files via a single common
>> shared header. (Sorry I didn't respond before you sent v5)
> 
> Sorry, I still do not understand your suggestion.
> 
> 
> Could you provide me with a code diff
> showing how to do this better?

I think you are doing exactly what he's asking for:
declare it with extern in the header (modpost.h change)
define it in one source file (modpost.c change)