On Sun, 2024-03-24 at 11:40 +1300, Barry Song wrote:
> On Sat, Mar 23, 2024 at 7:16 PM Joe Perches <joe@xxxxxxxxxxx> wrote:
> >
> > On Sat, 2024-03-23 at 17:14 +1300, Barry Song wrote:
> > > > > The patch titled
> > > > > Subject: scripts: checkpatch: check unused parameters for function-like macro
> > > > > has been added to the -mm mm-nonmm-unstable branch. Its filename is
> > > > > scripts-checkpatch-check-unused-parameters-for-function-like-macro.patch
> > > >
> > > > Conceptually I don't disagree but I wonder what the false
> > > > positive rate is.
> > >
> > > Hi Joe,
> > >
> > > If your concern is whether this script might produce inaccurate reports, I'm confident that
> > > the false positive rate is 0%. Xining, please correct me if I'm mistaken.
> >
> > I presume it's low but non-zero.
> >
> > I think the false positives are going to be
> > in some cases where a type is used as a macro
> > argument in #ifdef/#else blocks
> >
> > #ifdef CONFIG_<FOO>
> >
> > #define foo(bar, baz, ...) \
> > static inline func##bar(baz, ...) \
> > { \
> > bar baz = ... \
> > ... \
> > }
> >
> > #else
> >
> > #define foo(bar, baz, ...)
> >
> > #endif
> >
> > where it's not reasonably possible to use a
> > static inline.
>
> I do find some exceptions such as include/uapi/linux/const.h
>
> #ifdef __ASSEMBLY__
> #define _AC(X,Y) X
> #define _AT(T,X) X
> #else
> #define __AC(X,Y) (X##Y)
> #define _AC(X,Y) __AC(X,Y)
> #define _AT(T,X) ((T)(X))
> #endif
[]
> I'll check if Xining has some approach to detect this kind of ifdef/else/endif,
> and remove those false positives.
checkpatch isn't expected to be perfect.
So I wouldn't worry about it, I just wondered if
the exception rate is near the warning rate.
I expect it's not, but if it is, then perhaps
the warning isn't useful.
Do you know how many existing cases there are in
the kernel that might trip this warning?
