2018-10-08 07:56:55

by Igor Stoppa

Subject: CHECKPATCH: strange warning on alignment modifier

Hi,

I have the following fragment of code:

+struct my_struct {
+ atomic_long_t l __aligned(sizeof(atomic_long_t));
+} __aligned(sizeof(atomic_long_t));


triggering this warning, when fed to checkpatch.pl:

WARNING: function definition argument 'atomic_long_t' should also have
an identifier name
#19: FILE: path/to/file.h
+ atomic_long_t l __aligned(sizeof(atomic_long_t));


gcc [(Ubuntu 7.3.0-16ubuntu3) 7.3.0] seems to be happy about it

I am using the HEAD from mainline.

My intent is to specify the alignment of both the field and the
structure (yes, probably redundant in this single-field case).

If I am doing something wrong, I can't figure out what it is, but I
don't understand why the WARNING is mentioning a function definition.

--
igor

2018-10-09 02:08:23

by Joe Perches

Subject: Re: CHECKPATCH: strange warning on alignment modifier

On Mon, 2018-10-08 at 10:56 +0300, Igor Stoppa wrote:
> Hi,
>
> I have the following fragment of code:
>
> +struct my_struct {
> + atomic_long_t l __aligned(sizeof(atomic_long_t));
> +} __aligned(sizeof(atomic_long_t));
>
>
> triggering this warning, when fed to checkpatch.pl:
>
> WARNING: function definition argument 'atomic_long_t' should also have
> an identifier name
> #19: FILE: path/to/file.h
> + atomic_long_t l __aligned(sizeof(atomic_long_t));
>
>
> gcc [(Ubuntu 7.3.0-16ubuntu3) 7.3.0] seems to be happy about it
>
> I am using the HEAD from mainline.
>
> My intent is to specify the alignment of both the field and the
> structure (yes, probably redundant in this single-field case).
>
> If I am doing something wrong, I can't figure out what it is, but I
> don't understand why the WARNING is mentioning a function definition.

It's a defect in checkpatch.
For now, just ignore the message.
I will work on it later.