Re: [PATCH] arm64: perf_event: Fix time_offset for arch timer

[Will Deacon <will@kernel.org>]

On Sat, May 02, 2020 at 12:10:50AM +0800, Leo Yan wrote:
> On Fri, May 01, 2020 at 04:26:09PM +0100, Will Deacon wrote:
> 
> [...]
> 
> > > > > Let me try and understand your particular problem better.
> > > > 
> > > > I think the long and short of it is that userspace needs a way to convert
> > > > the raw counter cycles into a ns value that can be compared against values
> > > > coming out of sched_clock. To do this accurately, I think it needs the
> > > > cycles value at the point when sched_clock was initialised.
> > > 
> > > Will's understanding is exactly what I want to resolve in this patch.
> > > 
> > > The background info is for the ARM SPE [1] decoding with perf tool, if
> > > the timestamp is enabled, it uses the generic timer's counter as
> > > timestamp source.  SPE trace data only contains the raw counter cycles,
> > > as Will mentioned, the perf tool needs to convert it to a coordinate
> > > value with sched_clock.  This is why this patch tries to calculate the
> > > offset between the raw counter's ns value and sched_clock, eventually
> > > this offset value will be used by SPE's decoding code in Perf tool to
> > > calibrate a 'correct' timestamp.
> > > 
> > > Based on your suggestions, I will use __sched_clock_offset to resolve
> > > the accuracy issue in patch v2.  (I noticed Peter suggested to use a
> > > new API for wrapping clock_data structure, IIUC, __sched_clock_offset
> > > is more straightforward for this case).
> > > 
> > > Please correct if I miss anything.  Thank you for reviewing and
> > > suggestions!
> > 
> > I don't think you can use __sched_clock_offset without selecting
> > HAVE_UNSTABLE_SCHED_CLOCK, and we really don't want to do that just
> > for this. So Peter's idea about exposing what we need is better, although
> > you'll probably need to take care with the switch-over from jiffies.
> > 
> > It needs some thought, but one possibility would be to introduce a new
> > variant of sthe ched_clock_register() function that returns the cycle
> > offset, and then we could fish that out of the timer driver.
> 
> Thanks a lot for you and Marc for correction.
> 
> > If we're
> > crossing all the 'i's and dotting all the 't's then we'd want to disable the
> > perf userpage if sched_clock changes clocksource too (a bit like we do for
> > the vDSO).
> 
> To be honest, one thing is not clear for me is how the perf tool to
> update the arch timer's parameters in the middle of tracing after
> disable and re-enable per userpage.  I will note for this and look
> into detailed implementation for this part.

I don't fully understand the concern but, generally, the seqlock should
take care of any inconsistencies in the data page.

Will