Re: [PATCH] rtc: adapt allowed RTC update error

[Jason Gunthorpe <jgg@ziepe.ca>]

On Wed, Dec 02, 2020 at 11:08:38PM +0100, Thomas Gleixner wrote:
> >
> > arch/x86/kernel/kvmclock.c:     x86_platform.set_wallclock = kvm_set_wallclock;
> > arch/x86/kernel/x86_init.c:             x86_platform.set_wallclock = set_rtc_noop;
> > arch/x86/xen/time.c:            x86_platform.set_wallclock = xen_set_wallclock;
> > arch/x86/xen/time.c:    x86_platform.set_wallclock = xen_set_wallclock;
> >   All returns -ENODEV/EINVAL
> 
> You forgot to stare at the .get_wallclock() functions. That's the more
> interesting part, i.e. what's behind read_persistent_clock64().

Small steps! I was only looking at deleting the legacy
CONFIG_GENERIC_CMOS_UPDATE from x86 which only controls the
update_persistent_clock64()

Yes, there is a similar redundancy with rtclib on the
read_persistant_clock side, and that does looks much further..

> > arch/x86/kernel/x86_init.c:     .set_wallclock                  = mach_set_rtc_mmss,
> >   This is already rtclib under drivers/rtc/rtc-mc146818-lib.c
> 
> That's the shared library function for setting the darn thing.

Yes, but if a PNP entry is present then rtc-cmos will load and call
that function through the rtclib path instead of the
update_persistent_clock64() path

Jason