Re: [musl] [RFC PATCH] time.h: drop non-standard and out-of-date CLOCK_SGI_CYCLE

[Yao Zi <ziyao@disroot.org>]

On Fri, Feb 21, 2025 at 07:27:25PM -0500, Rich Felker wrote:
> On Thu, Nov 21, 2024 at 05:40:16PM +0000, Yao Zi wrote:
> > This definition was introduced in commit d74e462a ("add CLOCK_TAI (and
> > CLOCK_SGI_CYCLE) clock ids to time.h added in linux-v3.10 commit
> > 1ff3c9677bff7e468e0c487d0ffefe4e901d33f4"). However, it's actually a
> > platform specific feature, available only on SGI SN2 IA64 systems. Its
> > Linux support has been removed later in commit 07903ada9613
> > ("mmtimer: Remove the SGI SN2 mmtimer driver").
> > 
> > Having this definition is known to cause applications misdetect the
> > environment. There's no reason to keep a non-standard, seldom-available
> > and removed-since-4.13-kernel macro in a standard header.
> > 
> > Signed-off-by: Yao Zi <ziyao@disroot.org>
> > ---
> >  include/time.h | 1 -
> >  1 file changed, 1 deletion(-)
> > 
> > diff --git a/include/time.h b/include/time.h
> > index 3d948372..0fe02111 100644
> > --- a/include/time.h
> > +++ b/include/time.h
> > @@ -94,7 +94,6 @@ struct itimerspec {
> >  #define CLOCK_BOOTTIME           7
> >  #define CLOCK_REALTIME_ALARM     8
> >  #define CLOCK_BOOTTIME_ALARM     9
> > -#define CLOCK_SGI_CYCLE         10
> >  #define CLOCK_TAI               11
> >  
> >  #define TIMER_ABSTIME 1
> > -- 
> > 2.47.0
> 
> I think this looks ok. Any objections?
> 

Should I send it as formal patch again, or are you okay with picking the
RFC one directly?

> Rich

Thanks,
Yao Zi