Re: [RFC patch 0/4] TSC calibration improvements

[Ingo Molnar <mingo@elte.hu>]

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> However, this one is:
> 
> > +	/*
> > +	 * The iteration assumes that expect never goes below zero:
> > +	 */
> > +	BUILD_BUG_ON(QUICK_PIT_ITERATIONS >= 0xff);
> 
> No it doesn't. "expect" is unsigned char and will happily wrap, as 
> will the PIT timer. The fact that it is in "single shot" mode doesn't 
> actually mean that the timer stops, it just affects what happens when 
> it goes down to zero.
> 
> So that BUILD_BUG_ON() is misleading and incorrect.

ah, indeed, that bit of mine is wrong - and the period is programmed to 
0xffff so it should all work out just fine. You code in a way too tricky 
manner ;) I zapped that portion.

In fact ... shouldnt we intentionally include a 'wraparound' event in 
the test? Some of the erratums/instabilities regarding PITs happened 
around wraparounds [of the lsb] - maybe the wraparound of the MSB 
matters too. (Maybe some boards freeze the counter readout until the 
host OS ACKs the PIT irq or something - which we dont do in this 
calibration run so if there's some weirdness there we'd detect it.)

So maybe we should start with an expect value of QUICK_PIT_ITERATIONS/2, 
with a wraparound right in the middle of the measurement?

	Ingo