Re: [Xen-devel] [PATCH 13/60] xen/sched: move some per-vcpu items to struct sched_unit

[Juergen Gross <jgross@xxxxxxxx>]

On 02.07.19 12:01, Jan Beulich wrote:
> On 02.07.2019 11:40, Dario Faggioli wrote:
> > On Tue, 2019-07-02 at 08:29 +0000, Jan Beulich wrote:
> > > On 02.07.2019 10:21, Dario Faggioli wrote:
> > > > On Tue, 2019-07-02 at 07:54 +0000, Jan Beulich wrote:
> > > > > And again - if someone pins every vCPU to a single pCPU, that
> > > > > last
> > > > > such pinning operation will be what takes long term effect. Aiui
> > > > > all
> > > > > vCPU-s in the unit will then be pinned to that one pCPU, i.e.
> > > > > they'll either all compete for the one pCPU's time, or only one
> > > > > of
> > > > > them will ever get scheduled.
> > > > I'm not sure I'm getting this. On an, say, SMT system, with 4
> > > > threads
> > > > per core, a unit is 4 vCPUs and a pCPU is 4 threads.
> > > No, the meaning of pCPU is a single thread of a single core. I.e.
> > > what is represented by a single cpumask_t bit.
> > Fine, let's continue to call that a pCPU. Then, when core-scheduling is
> > enabled, there's no <<multiple vCPUs of a unit being pinned to the same
> > pCPU and all competing for jut its CPU time>>.
> >
> > There's units of 4 vCPUs being pinned to 4 pCPUs (the 4 pCPUs of a
> > core, not 4 random, nor arbitrary, ones).
> >
> > That is the point, AFAIUI.
> Well, okay, quite possible. But then for the example topology
> you gave, what is going to be the effect of
>
> xl vcpu-pin 0 0 0
dom0 vcpu0 will be pinned to cpu0.
dom0 vcpu1 will be pinned to cpu1.
dom0 vcpu2 will be pinned to cpu2.
dom0 vcpu3 will be pinned to cpu3.

> ? In particular for Dom0 and in particular for CPU 0, there may
> be reasons to pin a particular vCPU to it.
In reality only pinning vcpu0 to cpu0 should be needed. And this is done
e.g. in dcdbas_smi_request().


Juergen

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel