Re: [PATCH v2] KVM: arm64: Add KVM_CAP to control WFx trapping

From: Quentin Perret
Date: Thu Mar 28 2024 - 06:30:29 EST


Hi Colton,

On Monday 25 Mar 2024 at 20:12:04 (+0000), Colton Lewis wrote:
> Thanks for the feedback.
>
> Quentin Perret <qperret@xxxxxxxxxx> writes:
>
> > On Friday 22 Mar 2024 at 14:24:35 (+0000), Quentin Perret wrote:
> > > On Tuesday 19 Mar 2024 at 16:43:41 (+0000), Colton Lewis wrote:
> > > > Add a KVM_CAP to control WFx (WFI or WFE) trapping based on scheduler
> > > > runqueue depth. This is so they can be passed through if the runqueue
> > > > is shallow or the CPU has support for direct interrupt injection. They
> > > > may be always trapped by setting this value to 0. Technically this
> > > > means traps will be cleared when the runqueue depth is 0, but that
> > > > implies nothing is running anyway so there is no reason to care. The
> > > > default value is 1 to preserve previous behavior before adding this
> > > > option.
>
> > > I recently discovered that this was enabled by default, but it's not
> > > obvious to me everyone will want this enabled, so I'm in favour of
> > > figuring out a way to turn it off (in fact we might want to make this
> > > feature opt in as the status quo used to be to always trap).
>
> Setting the introduced threshold to zero will cause it to trap whenever
> something is running. Is there a problem with doing it that way?

No problem per se, I was simply hoping we could set the default to zero
to revert to the old behaviour. I don't think removing WFx traps was a
universally desirable behaviour, so it prob should have been opt-in from
the start.

> I'd also be interested to get more input before changing the current
> default behavior.

Ack, that is my personal opinion.

> > > There are a few potential issues I see with having this enabled:
>
> > > - a lone vcpu thread on a CPU will completely screw up the host
> > > scheduler's load tracking metrics if the vCPU actually spends a
> > > significant amount of time in WFI (the PELT signal will no longer
> > > be a good proxy for "how much CPU time does this task need");
>
> > > - the scheduler's decision will impact massively the behaviour of the
> > > vcpu task itself. Co-scheduling a task with a vcpu task (or not) will
> > > impact massively the perceived behaviour of the vcpu task in a way
> > > that is entirely unpredictable to the scheduler;
>
> > > - while the above problems might be OK for some users, I don't think
> > > this will always be true, e.g. when running on big.LITTLE systems the
> > > above sounds nightmare-ish;
>
> > > - the guest spending long periods of time in WFI prevents the host from
> > > being able to enter deeper idle states, which will impact power very
> > > negatively;
>
> > > And probably a whole bunch of other things.
>
> > > > Think about his option as a threshold. The instruction will be trapped
> > > > if the runqueue depth is higher than the threshold.
>
> > > So talking about the exact interface, I'm not sure exposing this to
> > > userspace is really appropriate. The current rq depth is next to
> > > impossible for userspace to control well.
>
> Using runqueue depth is going off of a suggestion from Oliver [1], who I've
> also talked to internally at Google a few times about this.
>
> But hearing your comment makes me lean more towards having some
> enumeration of behaviors like TRAP_ALWAYS, TRAP_NEVER,
> TRAP_IF_MULTIPLE_TASKS.

Do you guys really expect to set this TRAP_IF_MULTIPLE_TASKS? Again, the
rq depth is quite hard to reason about from userspace, so not sure
anybody will really want that? A simpler on/off thing might be simpler.

> > > My gut feeling tells me we might want to gate all of this on
> > > PREEMPT_FULL instead, since PREEMPT_FULL is pretty much a way to say
> > > "I'm willing to give up scheduler tracking accuracy to gain throughput
> > > when I've got a task running alone on a CPU". Thoughts?
>
> > And obviously I meant s/PREEMPT_FULL/NOHZ_FULL, but hopefully that was
> > clear :-)
>
> Sounds good to me but I've not touched anything scheduling related before.

Do you guys use NOHZ_FULL in prod? If not that idea might very well be a
non-starter, because switching to NOHZ_FULL would be a big ask. So,
yeah, I'm curious :)

Thanks,
Quentin