Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Fri, 06 Dec 2019 16:13:05 +0000
To:     Peter Zijlstra <peterz@infradead.org>
From:   Krzysztof Piecuch <piecuch@protonmail.com>
Cc:     "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "tglx@linutronix.de" <tglx@linutronix.de>,
        "mingo@redhat.com" <mingo@redhat.com>,
        "bp@alien8.de" <bp@alien8.de>, "hpa@zytor.com" <hpa@zytor.com>,
        "x86@kernel.org" <x86@kernel.org>,
        "viresh.kumar@linaro.org" <viresh.kumar@linaro.org>,
        "rafael.j.wysocki@intel.com" <rafael.j.wysocki@intel.com>,
        "drake@endlessm.com" <drake@endlessm.com>,
        "malat@debian.org" <malat@debian.org>,
        "juri.lelli@redhat.com" <juri.lelli@redhat.com>
Reply-To: Krzysztof Piecuch <piecuch@protonmail.com>
Subject: Re: [PATCH] x86/tsc: Don't use cpuid 0x16 leaf to determine cpu speed.
Message-ID: <s2tCh5akte6KmZimjnjJ0dn_zhWfLr_DEXJ6je2DFCUwZwrF9wqS4kPcJ-j1LOtP8rmkjR6oMg58anmztMJMd0e1Dfe8CvtTZudBXRT4gik=@protonmail.com>
In-Reply-To: <20191206103942.GE2844@hirez.programming.kicks-ass.net>
References: <SoluZg51N39Rx0tDCSJFbEvvgMrDnJ_g0RdRdN5mtCfag4GahIOPfok7UbkyeO5Qpl3wUHp8H8y73JtClcZvr1ARSIOBIFQAne0Z712el8M=@protonmail.com>
 <20191206103942.GE2844@hirez.programming.kicks-ass.net>
Feedback-ID: krphKiiPlx_XKIryTSpdJ_XtBwogkHXWA-Us-PsTeaBSrzOTAKWxwbFkseT4Z85b_7PMRvSnq3Ah7f9INXrOMw==:Ext:ProtonMail
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Thank you for your reply.

I experience 2% tsc clock-drift (671s ahead of my local NTP server after 9.=
5h)
on my machine with Supermicro's Hyperspeed turned on. There's no clock
drift when I turn Hyperspeed off.

As far as I know Hyperspeed increases base clock frequency[1].

That's what CPUID says about my overclocked Intel Xeon Gold 6146:

   Time Stamp Counter/Core Crystal Clock Information (0x15):
      TSC/clock ratio =3D 256/2
      nominal core crystal clock =3D 0 Hz
   Processor Frequency Information (0x16):
      Core Base Frequency (MHz) =3D 0xc80 (3200)
      Core Maximum Frequency (MHz) =3D 0x1068 (4200)
      Bus (Reference) Frequency (MHz) =3D 0x64 (100)

tsc_refine_calibration_work never corrects the early calibration
because it calculates a tsc frequency beyond 1% tolerance.

I've bumped the tsc_refine_calibration_work's tolerance to 3% and made it w=
ork:

Hyperspeed:
[    8.571471] tsc: Refined TSC clocksource calibration: 3264.012 MHz
No hyperspeed:
[    8.506009] tsc: Refined TSC clocksource calibration: 3200.013 MHz

Increasing the tolerance to 3% would work in my case but apparently some
servers can increase the base-clock frequency to 6%. [2]
At this point in order to completely eliminate this bug we would need to
significantly increase the tolerance which might introduce other bugs.

[1]: https://www.supermicro.com/support/faqs/faq.cfm?faq=3D21337
[2]: https://www.servethehome.com/supermicro-hyper-speed-server-overclockin=
g-bios/

Kind regards,
Krzysztof Piecuch