On Tue, 2017-11-21 at 01:18 +0900, Masahiro Yamada wrote:
> 2017-11-17 2:01 GMT+09:00 Knut Omang <knut.om...@oracle.com>:
> > Add interpretation of a new environment variable P={1,2} in spirit of the
> > C= option, but executing checkpatch instead of sparse.
> >
> > Signed-off-by: Knut Omang <knut.om...@oracle.com>
> > Reviewed-by: Håkon Bugge <haakon.bu...@oracle.com>
> > Acked-by: Åsmund Østvold <asmund.ostv...@oracle.com>
> > ---
> > Makefile | 20 +++++++++++++++++++-
> > scripts/Makefile.build | 13 +++++++++++++
> > 2 files changed, 32 insertions(+), 1 deletion(-)
> >
> > diff --git a/Makefile b/Makefile
> > index ccd9818..eb4bca9 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -176,6 +176,20 @@ ifndef KBUILD_CHECKSRC
> > KBUILD_CHECKSRC = 0
> > endif
> >
> > +# Run scripts/checkpatch.pl with --ignore-cfg checkpatch.cfg
> > +#
> > +# Use 'make P=1' to enable checking of only re-compiled files.
> > +# Use 'make P=2' to enable checking of *all* source files, regardless
> > +#
> > +# See the file "Documentation/dev-tools/run-checkpatch.rst" for more
> > details,
> > +#
> > +ifeq ("$(origin P)", "command line")
> > + KBUILD_CHECKPATCH = $(P)
> > +endif
> > +ifndef KBUILD_CHECKPATCH
> > + KBUILD_CHECKPATCH = 0
> > +endif
>
>
> I am unhappy about adding a new interface
> for each checker.
I see your point. I wanted to extend C= to handle checkpatch as well but see no
obvious good non-intrusive solution.
> The default of CHECK is "sparse", but
> users can override it to use another checker.
>
>
>
> As Decumentation/dev-tools/coccinelle.rst says,
> if you want to use coccinelle as a checker,
>
> make C=1 CHECK="scripts/coccicheck"
That works well with coccinelle since both sparse and coccinelle rely on getting
the same command line options as what's passed to the compiler, while checkpatch
is quite different:
make C=2 CHECK="\$(srctree)/scripts/checkpatch.pl"
fails, complaining about every single compiler flag.
Thanks,
Knut
> Recently, I saw a patch to use scripts/kernel-doc as a checker.
> https://patchwork.kernel.org/patch/10030521/
>
>
>
> If I accept your patch,
> we would end up with
>
> KBUILD_CHECKPATCH,
> KBUILD_CHECKCOCCI
> KBUILD_CHECKDOC,
> ...
>
> This is ugly.
