On 7 September 2014 19:14, Josh Triplett <josh@xxxxxxxxxxxxxxxx> wrote:
> On Sun, Sep 07, 2014 at 02:36:53PM +0200, Ard Biesheuvel wrote:
>> This code snippet:
>>
>> static void bar(void const *arg)
>> {
>> int (*foo)(void) = arg;
>> }
>>
>> produces the following warning:
>>
>> test.c:4:28: warning: incorrect type in initializer (different modifiers)
>> test.c:4:28: expected int ( *foo )( ... )
>> test.c:4:28: got void const *arg
>>
>> which is caused by the fact that the function pointer 'foo' is not annotated
>> as being a pointer to const data. However, dereferencing a function pointer
>> does not produce an lvalue, so a function pointer points to const data by
>> definition, and we should treat it accordingly.
>>
>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx>
>
> As a data point: gcc does not warn about this case either, whereas it
> does warn about "int *foo = arg". So, this seems fine to me.
>
> Reviewed-by: Josh Triplett <josh@xxxxxxxxxxxxxxxx>
>
>> OK, so while my v1 did solve the example case, it turns out universally treating
>> function pointers as pointers to const data produces so much fallout that it
>> does more harm than good.
>
> Can you elaborate on the fallout, and ideally provide that rationale in
> the commit message?
>
Treating a function pointer as pointer-to-const-data in every
occurrence results in sparse triggering on the inverse case, i.e.
extern int (*foo)(void);
void *arg = foo;
which is a pattern that is widely used in the kernel, arguably
incorrect (considering the nature of a function pointer) but not the
case I intended to address.
I will put something to that effect in the v3 commit message.
Cheers,
Ard.
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
