Re: [RFC v2] arm,x86,fs/resctrl: Generic schema description Proof of Concept

[Chen, Yu C]

Hi Fenghua,

On 8/10/2026 11:25 AM, Fenghua Yu wrote:
> Hi, Chen,
>
> On 8/9/26 20:02, Chen, Yu C wrote:
> > Hi Fenghua,
> > On 8/10/2026 10:55 AM, Fenghua Yu wrote:
> >
> > [ ... ]
> >
> > > struct resctrl_ctrl {
> > > ...
> > >          struct list_head        emulated_by;
> > > };
> > >
> > > Does this mean a legacy resctrl control can be emulated by a few 
> > > resctrl controls (which is stored in the list)?
> > >
> > > e.g. legacy MB control can be emulated by a few controls? I thought 
> > > "MB:" line is emulated by only one native control. e.g. "MB:" line is 
> > > emulated by "MB_NODE:" line.
> >
> > If I understand correctly, for region-aware RDT, the legacy MB control 
> > may
> > be emulated collectively by region0_control, region1_control, ..., 
> > regionN_control.
> > In the original definition, MB represents the traffic that misses the 
> > L3 cache and
> > is refilled from all sources, which could include DRAM (region0), CXL 
> > (region1),
> > and others.
> I may misunderstand "emulation" concept here.
>
> I was thinking emulation mode is for backward compatibility, i.e. a 

Yes, this is also my understanding.

> native MB control is emulating "MB:" line: writing and reading both 
> lines are same. So legacy user app won't be broken on new control.

For legacy MB, it is currently percentage-based with a valid range of
[1, 100]. Userspace might still want to write using percentage-based
values to maintain backward compatibility.

The native MB control, on the other hand, may have a valid range of
[1, 255]. Therefore, a mapping is required to convert user-provided[1, 100]
 to [1, 255] before writing to the underlying hardware.

> For a list of resctrl controls (region0_control, region1_control, etc), 
> which one is for legacy "MB:" control so that reading/wring the control 
> is same as writing/reading "MB:"line?

My understanding is that all of them will be written to the same value,
which has been mapped from [1, 100] to [1, 255].
According to the change:
https://git.kernel.org/pub/scm/linux/kernel/git/reinette/linux.git/commit/?h=resctrl/controls_rfc_v2.1&id=2b4fa1beb03848555090a4e8a3d2bb80524e0478
list_for_each_entry(em_ctrl, &ctrl->emulated_by, entry) {
	//every domain for each emulating controller will write the same value
}

thanks,
Chenyu