Re: [PATCH 3/4] Constify struct kset_uevent_ops for 2.6.32-git-053fe57ac v2

[Emese Revfy <re.emese@gmail.com>]

David Rientjes wrote:
> On Sat, 26 Dec 2009, Emese Revfy wrote:
> 
>>>> diff --git a/mm/slub.c b/mm/slub.c
>>>> index 4996fc7..fb63aca 100644
>>>> --- a/mm/slub.c
>>>> +++ b/mm/slub.c
>>>> @@ -4522,7 +4522,7 @@ static int uevent_filter(struct kset *kset, struct kobject *kobj)
>>>>        return 0;
>>>>  }
>>>>
>>>> -static struct kset_uevent_ops slab_uevent_ops = {
>>>> +static const struct kset_uevent_ops slab_uevent_ops = {
>>>>        .filter = uevent_filter,
>>>>  };
>>>   CC      mm/slub.o
>>> mm/slub.c: In function 'slab_sysfs_init':
>>> mm/slub.c:4679: warning: passing argument 2 of 'kset_create_and_add'
>>> discards qualifiers from pointer target type
>>> include/linux/kobject.h:164: note: expected 'struct kset_uevent_ops *'
>>> but argument is of type 'const struct kset_uevent_ops *'
>>>
>>> Hmm?
>>>
>> I double checked both the declaration and definitions of the affected
>> function/structure and they are consistently const here.
>> Can you tell me what patch/tree combination you encountered this warning with?
>> Thanks, Emese
>>
> 
> The warning emitted by gcc is pretty explicit; the second formal of 
> kset_create_and_add() takes a 'struct kset_uevent_ops *' type while you're 
> now passing 'const struct kset_uevent_ops *' as the result of your change.  
> That said, kset_create() could probably be modified with the const 
> qualifier, but that's outside the scope of your patchset and would have to 
> be proposed seperately.

I see what you mean, this was part of [PATCH 1/4] that I forgot to CC to you.
Emese