MIME-Version: 1.0
In-Reply-To: <20150610154309.13853.qmail@ns.horizon.com>
References: <20150610092553.GA32269@gmail.com>
	<20150610154309.13853.qmail@ns.horizon.com>
Date: Wed, 10 Jun 2015 08:56:16 -0700
Message-ID: <CADyApD0n0wnT_m7vgojgJHNO7MD-q6Mcf0=hYpBMDALkB1jgLA@mail.gmail.com>
Subject: Re: Discussion: quick_pit_calibrate is slow
From: Arjan van de Ven <arjanvandeven@gmail.com>
To: George Spelvin <linux@horizon.com>
Cc: Ingo Molnar <mingo@kernel.org>, Peter Zijlstra <a.p.zijlstra@chello.nl>,
        adrian.hunter@intel.com, ak@linux.intel.com,
        Matt Morton <akpm@linux-foundation.org>,
        Arjan van de Ven <arjan@infradead.org>, Borislav Petkov <bp@alien8.de>,
        "H. Peter Anvin" <hpa@zytor.com>, LKML <linux-kernel@vger.kernel.org>,
        luto@amacapital.net, penberg@iki.fi,
        Thomas Gleixner <tglx@linutronix.de>,
        Linus Torvalds <torvalds@linux-foundation.org>
Content-Type: text/plain; charset=UTF-8
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1518
Lines: 35

> If my plan survives contact with reality, it should work better 100%
> of the time and obsolete the old code.
>
> You said it should fail back to the old code, but there's not a lot of
> difference between failures I can detect and failures I can work around.
>
> I know it's a fatal error to underestimate the perversity of PC hardware
> (virtualized hardware even more so) but I'm trying to mitigate that by
> really deeply understanding what the current code does and all the error
> conditions is can handle.
>
>> (and RTC if that can be managed.)
>


btw one thing to think about; if you calibrate VERY quickly, you need
to consider spread spectrum clocking.
various systems have different clocks on different SSC domains, and
while on a bit larger timewindow you can completely ignore SSC,
once you go into very short time windows you can't.
Now... you can do a quick calibration and then a longer calibration in
the background
(we do that already after 1 second or so.. but that's 900 msec after
userspace finished booting so probably too late if the first
calibration is too coarse)


also note that normally people kind of expect around 100ppm to not
have their clocks deviate too much during the day.
(but this can be the result of the longer term calibration)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/