Re: [RFC PATCH v2 02/14] x86/hpet: Expose more functions to read and write registers

[Ricardo Neri]

On Tue, Mar 26, 2019 at 10:00:24PM +0100, Thomas Gleixner wrote:
> On Wed, 27 Feb 2019, Ricardo Neri wrote:
> >  struct irq_data;
> > @@ -109,6 +114,11 @@ extern void hpet_unregister_irq_handler(rtc_irq_handler handler);
> >  static inline int hpet_enable(void) { return 0; }
> >  static inline int is_hpet_enabled(void) { return 0; }
> >  #define hpet_readl(a) 0
> > +#define hpet_writel(d, a)
> 
> What for?
> 
> > +#ifdef CONFIG_X86_64
> > +#define hpet_readq(a) 0
> > +#define hpet_writeq(d, a)
> > +#endif
> 
> Ditto.
> 
> There are no users outside of HPET and your new HPET watchdog code for
> those. And both are not compiled when CONFIG_HPET=n.

I'll remove these unneeded defintions.
> 
> The only reason to have the hpet_readl() define, which btw. should be an
> inline, is to avoid massive ifdeffery in the TSC calibration code.

May I ask what is the problem with the #define hpet_readl()? Commit
bfc0f5947afa("x86: merge tsc calibration") changed it from inline to
#define. Should I change it back?

Thanks and BR,
Ricardo