Re: [PATCH mlx5-next v4 1/4] PCI: Add sysfs callback to allow MSI-X table size change of SR-IOV VFs

[Leon Romanovsky <leon@kernel.org>]

On Tue, Jan 26, 2021 at 12:20:11AM -0800, Joe Perches wrote:
> On Tue, 2021-01-26 at 08:01 +0200, Leon Romanovsky wrote:
> > On Mon, Jan 25, 2021 at 01:52:29PM -0800, Jakub Kicinski wrote:
> > > On Sun, 24 Jan 2021 15:11:16 +0200 Leon Romanovsky wrote:
> > > > +static int pci_enable_vfs_overlay(struct pci_dev *dev) { return 0; }
> > > > +static void pci_disable_vfs_overlay(struct pci_dev *dev) {}
> > >
> > > s/static /static inline /
> >
> > Thanks a lot, I think that we should extend checkpatch.pl to catch such
> > mistakes.
>
> Who is this "we" you refer to? ;)

"We" == community :)

>
> > How hard is it to extend checkpatch.pl to do regexp and warn if in *.h file
> > someone declared function with implementation but didn't add "inline" word?
>
> Something like this seems reasonable and catches these instances in
> include/linux/*.h

Thanks

>
> $ ./scripts/checkpatch.pl -f include/linux/*.h --types=static_inline --terse --nosummary
> include/linux/dma-mapping.h:203: WARNING: static function definition might be better as static inline
> include/linux/genl_magic_func.h:55: WARNING: static function definition might be better as static inline
> include/linux/genl_magic_func.h:78: WARNING: static function definition might be better as static inline
> include/linux/kernel.h:670: WARNING: static function definition might be better as static inline
> include/linux/kprobes.h:213: WARNING: static function definition might be better as static inline
> include/linux/kprobes.h:231: WARNING: static function definition might be better as static inline
> include/linux/kprobes.h:511: WARNING: static function definition might be better as static inline
> include/linux/skb_array.h:185: WARNING: static function definition might be better as static inline
> include/linux/slab.h:606: WARNING: static function definition might be better as static inline
> include/linux/stop_machine.h:62: WARNING: static function definition might be better as static inline
> include/linux/vmw_vmci_defs.h:850: WARNING: static function definition might be better as static inline
> include/linux/zstd.h:95: WARNING: static function definition might be better as static inline
> include/linux/zstd.h:106: WARNING: static function definition might be better as static inline
>
> A false positive exists when __must_check is used between
> static and inline.  It's an unusual and IMO not a preferred use.

Maybe just filter and ignore such functions for now?
Will you send proper patch or do you want me to do it?

> ---
>  scripts/checkpatch.pl | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
>
> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> index 4f8494527139..0ac366481962 100755
> --- a/scripts/checkpatch.pl
> +++ b/scripts/checkpatch.pl
> @@ -4451,6 +4451,18 @@ sub process {
>  			}
>  		}
>
> +# check for static function definitions without inline in .h files
> +# only works for static in column 1 and avoids multiline macro definitions
> +		if ($realfile =~ /\.h$/ &&
> +		    defined($stat) &&
> +		    $stat =~ /^\+static(?!\s+(?:$Inline|union|struct))\b.*\{.*\}\s*$/s &&
> +		    $line =~ /^\+static(?!\s+(?:$Inline|union|struct))\b/ &&
> +		    $line !~ /\\$/) {
> +			WARN("STATIC_INLINE",
> +			     "static function definition might be better as static inline\n" .
> +				$herecurr);
> +		}
> +
>  # check for non-global char *foo[] = {"bar", ...} declarations.
>  		if ($line =~ /^.\s+(?:static\s+|const\s+)?char\s+\*\s*\w+\s*\[\s*\]\s*=\s*\{/) {
>  			WARN("STATIC_CONST_CHAR_ARRAY",
>
>