On 18 November 2014 17:09, Christopher Li <sparse@xxxxxxxxxxx> wrote:
> On Tue, Nov 18, 2014 at 12:24 AM, Ard Biesheuvel
> <ard.biesheuvel@xxxxxxxxxx> wrote:
>>
>> No, the test case looks correct now. You may want to add a conflicting
>> definition, for completeness, as that is an important case for sparse
>> to verify.
>> So you could just add
>>
>> static void *pure1(void); // conflicting declaration
>
> Just to confirm. If function declare as non pure then second time
> declare as pure, that is a conflict error. That is very different from other
> C attributes. C allow function declare attribute incrementally.
>
> If what you said is true, then pure attribute is actually acting like a modifier
> which don't fix not other non pure attribute.
>
Well, that is how sparse currently treats it, and I think that is correct.
If an extern declaration in a .h file specifies __pure, then GCC may
legally emit fewer calls to that function than appear in the source.
So if the actual implementation in the .c file is not __pure, then you
have an error.
>> at the bottom, and ensure it is flagged by sparse as an error after
>> you make your changes.
>
> I am curios how does GCC behave on this incremental declare of pure
> attribute. I just test it. GCC does not warn on the conflicting declare.
>
> GCC seems warn on some thing shouldn't warn on the test case:
>
> function "f2_ok" and "f5" shouldn't been warned. Strange.
>
> validation/pure-function.c:14:1: warning: '__pure__' attribute ignored
> [-Wattributes]
> static __attribute__((__pure__)) void*(*f2_ok)(void) = pure1;
> ^
> validation/pure-function.c:15:1: warning: '__pure__' attribute ignored
> [-Wattributes]
> static __attribute__((__pure__)) void*(*f3_error)(void) = non_pure1;
> ^
> validation/pure-function.c:29:1: warning: '__pure__' attribute ignored
> [-Wattributes]
> static __attribute__((__pure__)) int(*f5)(void) = pure2;
> ^
> validation/pure-function.c:30:1: warning: '__pure__' attribute ignored
> [-Wattributes]
> static __attribute__((__pure__)) int(*f6_error)(void) = non_pure2;
>
My apologies. It appears GCC ignores pure attributes on function
pointers, so the assignments are all legal. Sorry to have wasted your
time on this.
So what remains imo is
"""
static __attribute__((__pure__)) void *pure1(void)
{
void *i = (void *)0;
return i;
}
static __attribute__((__pure__)) int pure2(void)
{
int i = 0;
return i;
}
static int pure2(void); // error here
"""
and all the other stuff can be dropped.
Regards,
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
