Re: Re: [PATCH] staging:vt6656:baseband.h: fix function definition argument without identifier name issue

[Joe Perches]

On Thu, 2016-10-13 at 18:49 +0200, Greg KH wrote:
> On Thu, Oct 13, 2016 at 09:37:23AM -0700, Joe Perches wrote:
> > On Thu, 2016-10-13 at 16:57 +0200, Greg KH wrote:
> > > On Thu, Oct 13, 2016 at 05:23:45PM +0300, Mikhail Golubev wrote:
> > > > On Thu, Oct 13, 2016 at 02:06:02PM +0200, Greg KH wrote:
> > > > > On Thu, Oct 13, 2016 at 02:50:18PM +0300, Mikhail Golubev wrote:
> > > > > > Function definitions arguments should also have an identifier name as reported by checkpatch.pl.
> > 
> > []
> > > > > > diff --git a/drivers/staging/vt6656/baseband.h b/drivers/staging/vt6656/baseband.h
> > 
> > []
> > > > > > @@ -86,15 +86,15 @@ struct vnt_phy_field {
> > > > > >  unsigned int vnt_get_frame_time(u8 preamble_type, u8 pkt_type,
> > > > > >  				unsigned int frame_length, u16 tx_rate);p
> > > > > >  
> > > > > > -void vnt_get_phy_field(struct vnt_private *, u32 frame_length,
> > > > > > -		       u16 tx_rate, u8 pkt_type, struct vnt_phy_field *);
> > > > > > -
> > > > > > -void vnt_set_short_slot_time(struct vnt_private *);
> > > > > > -void vnt_set_vga_gain_offset(struct vnt_private *, u8);
> > > > > > -void vnt_set_antenna_mode(struct vnt_private *, u8);
> > > > > > -int vnt_vt3184_init(struct vnt_private *);
> > > > > > -void vnt_set_deep_sleep(struct vnt_private *);
> > > > > > -void vnt_exit_deep_sleep(struct vnt_private *);
> > > > > > -void vnt_update_pre_ed_threshold(struct vnt_private *, int scanning);
> > > > > > +void vnt_get_phy_field(struct vnt_private *priv, u32 frame_length,
> > > > > > +		       u16 tx_rate, u8 pkt_type, struct vnt_phy_field *phy);
> > > > > > +
> > > > > 
> > > > > Really?  Since when is this a coding style requirement?
> > > > 
> > > > This requirement is really new. It was proposed by Joe Perches at 26 Sep 2016:
> > > > [PATCH] checkpatch: Add warning for unnamed function definition.
> > > > 
> > > > Should this type of warnings be fixed here?
> > > 
> > > Ugh, Joe, why did you add this option?
> > 
> > 
> > 1. Most all kernel prototypes use named arguments.
> > 2. It helps make header files easier to read/lookup with grep.
> > 
> > int func(int, int, int)
> > vs
> > int func(int weight, int density, int mass)
> > 
> > which is easier for humans to use?
> 
> 
> Yes, which is why I use that format, but is it something we are now
> going to suddenly require?
> 
> Also, this is going to take a lot more work to review patches like this,
> to match up the variable names to ensure that the developer got it
> right...

Coccinelle to the rescue...

http://www.spinics.net/lists/linux-kernel-janitors/msg28450.html

Julia Lawall wrote a script for that.
Unfortunately, it doesn't currently span .h and .c files

---

@r@
identifier f;
position p;
type T, t;
parameter list[n] ps;
@@

T f@p(ps,t,...);

@s@
identifier r.f,x;
type r.T, r.t;
parameter list[r.n] ps;
@@

T f(ps,t x,...) { ... }

@@
identifier r.f, s.x;
position r.p;
type r.T, r.t;
parameter list[r.n] ps;
@@

T f@p(ps,t
+ x
  ,...);