Re: [PATCH i-g-t] lib/xe/oa: Re-instate TestOa metric sets in LNL/BMG XML's

["Dixit, Ashutosh" <ashutosh.dixit@intel.com>]

On Mon, 18 Nov 2024 09:20:00 -0800, Souza, Jose wrote:
>
> > diff --git a/lib/xe/oa-configs/codegen.py b/lib/xe/oa-configs/codegen.py
> > index 51498f803d..7251f5c7a6 100644
> > --- a/lib/xe/oa-configs/codegen.py
> > +++ b/lib/xe/oa-configs/codegen.py
> > @@ -352,6 +352,9 @@ class Gen:
> >          m = re.search(r'\$GtSlice([0-9]+)XeCore([0-9]+)$', name)
> >          if m:
> >              return 'intel_xe_perf_devinfo_subslice_available(&perf->devinfo, {0}, {1})'.format(m.group(1), m.group(2))
> > +        m = re.search(r'\$GtXeCore([0-99]+)$', name)
> > +        if m:
> > +            return 'intel_xe_perf_devinfo_subslice_available(&perf->devinfo, {0}, 0)'.format(m.group(1))
>
> Just note sure about this one, it have 8 GtXeCore so should be the
> subslice/DSS not the slices.  But you would also need need to ajust the
> first parameter to take into consideration subslices in the second and
> other slices or add a function that does this for you.

Yup, I just put it in "something" there to make this compile for now. I
want to add a FIXME here to fix later.

It can be fixed later if anyone is going to use it. We don't know how to
use this information yet, at least in IGT, so I am assuming this can be
fixed if needed, if it's going to be used, at the time we try to actually
make use of the TestOa metric in IGT.

Is there any use for this in Mesa?

Thanks.
--
Ashutosh