Re: [PATCH 0/3] Reorg ppc64 pmu insn counting

[Daniel Henrique Barboza]

On 1/3/22 03:48, Cédric Le Goater wrote:
> On 12/30/21 23:12, Daniel Henrique Barboza wrote:
> > On 12/23/21 18:19, Richard Henderson wrote:
> > > On 12/23/21 12:36 PM, Daniel Henrique Barboza wrote:
> > > > This reorg is breaking PMU-EBB tests, unfortunately. These tests are run from 
> > > > the kernel
> > > > tree [1] and I test them inside a pSeries TCG guest. You'll need to apply 
> > > > patches 9 and
> > > > 10 of [2] beforehand (they apply cleanly in current master) because they aren't 
> > > > upstream
> > > > yet and EBB needs it.
> > > >
> > > > The tests that are breaking consistently with this reorg are:
> > > >
> > > > back_to_back_ebbs_test.c
> > > > cpu_event_pinned_vs_ebb_test.c
> > > > cycles_test.c
> > > > task_event_pinned_vs_ebb_test.c
> > >
> > > In which case perhaps drop my first patch for now, and instead simply replicate 
> > > your tcg algorithm in c exactly -- using none of the helpers that currently 
> > > exist.
> > >
> > > We can improve the code, and the use of pmc_get_event from 
> > > hreg_compute_hregs_value second.
> >
> >
> > While attempting to do that I figured what was off with this series and ended up
> > fixing it.
> >
> > It's now working with the event-based branch interrupt tests and Avocado seems 
> > happy
> > as well. It took some small adjustments/fixes in patches 1/2 and an extra patch 
> > of mine
> > tuning the existing logic after the reorg.
> >
> >
> > I'll clean it up and re-send it next week/year.
>
> Shouldn't we merge this series first ? It is really improving emulation
> and keeps the check-avocado tests under the timeout limit (which I find
> important).


As it is this series is breaking EBB tests if we apply the EBB interrupt patches
on top of it. If your concern is strictly with the Avocado tests then we'd 
better
off pushing my TCG version which fixes Avocado and don't break anything else.

Besides, I'll sent the v2 of this series today, tomorrow at the latest. We're 
better
off pushing the fixed version.


Thanks,

Daniel



> Thanks,
>
> C.