Re: Hi, Ingo, would you please help drop the TSC MSR calibration patch

From: Jacob Pan
Date: Mon Jul 11 2016 - 17:20:16 EST


On Mon, 11 Jul 2016 21:31:04 +0200
Ingo Molnar <mingo@xxxxxxxxxx> wrote:

>
> * Jacob Pan <jacob.jun.pan@xxxxxxxxx> wrote:
>
> > On Mon, 11 Jul 2016 07:59:19 -0700
> > "Chen, Yu C" <yu.c.chen@xxxxxxxxx> wrote:
> >
> > > Currently it is in your x86/timer tree:
> > >
> > > commit fc273eeef314cdaf0ac992b400d126f8184a4d1c
> > > Author: Len Brown <len.brown@xxxxxxxxx>
> > > Date: Fri Jun 17 01:22:49 2016 -0400
> > >
> > > x86/tsc_msr: Extend to include Intel Core Architecture
> > >
> > >
> > > Previously we found this patch might decrease the performance on
> > > one of our servers, due to the small gap between using old PIT
> > > calibration and new MSR calibration method, so we currently would
> > > like to hold this patch for now, until we got a clear answer from
> > > our architect. Would you please help revert this patch (the other
> > > patches are safe and can be merged), sorry for the inconvenience.
> > >
> > I modified the subject slightly to be more specific.
> > Adding lkml and x86 list, and a few more people.
> > This commit is also affected, won't compile if we revert the one
> > above.
> >
> > 37c528e... x86/tsc_msr: Fix rdmsr(MSR_PLATFORM_INFO) unsafe warning
> > in KVM guest
>
> Ok, I've rebased tip:x86/timers, it now includes the following
> commits:
>
> ff4c86635ee1 x86/tsc: Enumerate BXT tsc_khz via CPUID
> aa297292d708 x86/tsc: Enumerate SKL cpu_khz and tsc_khz via CPUID
> 02c0cd2dcf7f x86/tsc_msr: Remove irqoff around MSR-based TSC
> enumeration 6fcb41cdaee5 x86/tsc_msr: Add Airmont reference clock
> values 05680e7fa8a4 x86/tsc_msr: Correct Silvermont reference clock
> values 9e0cae9f6227 x86/tsc_msr: Update comments, expand definitions
> 14bb4e34860a x86/tsc_msr: Remove debugging messages
> ba8268330dc1 x86/tsc_msr: Identify Intel-specific code
> fc5f3ac24720 Revert "x86/tsc: Add missing Cherrytrail frequency to
> the table"
>
> Does that work for you?
>
works for me but Yu has to confirm.

> Thanks,
>
> Ingo