Re: [PATCH v5 1/8] selftests: Add printf attribute to kselftest prints

[Ilpo Järvinen]

On Thu, 12 Oct 2023, Maciej Wieczór-Retman wrote:

> On 2023-10-11 at 13:40:48 -0600, Shuah wrote:
> >On 10/11/23 02:23, Maciej Wieczor-Retman wrote:
> >> Kselftest header defines multiple variadic functions that use printf
> >> along with other logic.
> >> 
> >> There is no format checking for the variadic functions that use
> >> printing inside kselftest.h. Because of this the compiler won't
> >> be able to catch instances of mismatched printf formats and debugging
> >> tests might be more difficult.
> >> 
> >> Add the common __printf attribute macro to kselftest.h.
> >> 
> >> Add __printf attribute to every function using formatted printing with
> >> variadic arguments.
> >> 
> >> Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@xxxxxxxxx>
> >> Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@xxxxxxxxxxxxxxx>
> >> Reviewed-by: Reinette Chatre <reinette.chatre@xxxxxxxxx>
> >> ---
> >> Changelog v4:
> >> - Fix typo in patch subject. (Reinette)
> >> - Add Reinette's reviewed-by tag.
> >> 
> >
> >I still need information on how you found these problems. Please
> >add it to change log for each of these patches.
> 
> Sure, I'll add notes on methodology to patches 2-8. I understand that
> this patch (1/8) message doesn't need that addition since the problems
> it exposes are in separate patches.
> 
> Or would you like me to also note here more specifically what effect it
> has in the rest of the series?
> 
> >I am seeing checkpatch warning:
> >
> >WARNING: Prefer __printf(a, b) over __attribute__((format(printf, a, b)))
> >#102: FILE: tools/testing/selftests/kselftest.h:81:
> >+#define __printf(a, b)   __attribute__((format(printf, a, b)))
> 
> Running checkpatch.pl with --show-types shows the
> PREFER_DEFINED_ATTRIBUTE_MACRO is raised. From looking at the error
> message in the script it looks like a false positive:
> 	"Prefer $new over __attribute__(($orig_attr$params))\n"
> 
> Please correct me if my train of thought is wrong but I think checkpatch
> sees __printf() macro defined and it sees it's raw version
> "__attribute__((format(printf, a, b)))" which it wants to replace with
> the macro. But since the raw version is found in the define line that is
> obviously not possible.

Yes, this is clearly a false positive from checkpatch. Checkpatch's logic 
cannot differentiate the definition from the use of __printf(), it just 
assumes __printf() is there already, which is not true for selftests.

The patch adds the capability to use __printf() elsewhere in the 
selftests code but of course the definition of __printf() itself has to 
use __attribute__().

-- 
 i.