Re: [PATCH net-next 06/10] can: rcar_canfd: Repurpose f_dcfg base for other registers

[Jakub Kicinski <kuba@kernel.org>]

On Thu, 19 Jun 2025 12:16:00 +0200 Geert Uytterhoeven wrote:
> On Thu, 19 Jun 2025 at 06:43, Vincent Mailhol <mailhol.vincent@wanadoo.fr> wrote:
> > On Thu. 19 Jun. 2025 at 10:38, Jakub Kicinski <kuba@kernel.org> wrote:  
> > > On Wed, 18 Jun 2025 11:20:00 +0200 Marc Kleine-Budde wrote:  
> > > > +static inline unsigned int rcar_canfd_f_cfdcrc(struct rcar_canfd_global *gpriv,
> > > > +                                            unsigned int ch)
> > > > +{
> > > > +     return gpriv->info->regs->coffset + 0x10 + 0x20 * ch;
> > > > +}  
> > >
> > > clang is no longer fooled by static inline, it identifies that 4 out of  
> 
> Oh well, that explains why someone pointed to a CI log showing more
> unused functions in a different driver.  I hope it only does that
> for unused functions in .c files, not in header files?

Yes, AFAIU it's clever enough to distinguish what came in from 
the headers.

> > > these functions are never called. I think one ends up getting used in
> > > patch 10 (just looking at warning counts), but the other 3 remain dead
> > > code. Geert, do you have a strong attachment to having all helpers
> > > defined or can we trim this, please?  
> 
> I would like to keep them (or at least the information), as it serves
> as register documentation, just like the macros they replaced....

Okay, we'll pull, but we really should try to keep the tree free of W=1
warnings. The CI can deal with existing warnings but they will annoy
humans doing development. Maybe there is a way to disable the warning
selectively for rcar if you find it unhelpful? And then we'll see if
some well meaning code janitor sends a patch to delete them anyway ;)