On Wed, 25 Mar 2020 at 17:55, Laszlo Ersek <lersek@redhat.com> wrote:
>
> On 03/25/20 06:17, Ni, Ray wrote:
> >>
> >> The properties table should not be used. It has been superseded by the memory attributes table, per spec.
> >>
> >> In edk2, the properties table is controlled by the PCD, regardless of the memory attributes table.
> >>
> >> In edk2, the memory attributes table is always produced, regardless of the properties table.
> >>
> >> Please see the discussion under:
> >>
> >> [edk2] [patch 0/7] Add UEFI2.6 MemoryAttributesTable support.
> >>
https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Fmid.mail-archive.com%2F1454069539-4056-1-git-send-email-jiewen.yao%40intel.com&data=02%7C01%7Cbret.barkelew%40microsoft.com%7C46138683ba554847c60008d7d0de1753%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637207524599988545&sdata=7M4nzzz%2B8UZ2kb4LHyP%2BfBOWFrv5xFuAT19pdAn3kEE%3D&reserved=0
> >>
> >
...
>
> > Do you think we could remove properties table?
>
> Yes, that's exactly what Ard requested, as soon as Jiewen posted the MAT
> series. Back then, Jiewen said that some production OSes were still
> using the properties table, and would need time to migrate to MAT.
>
> The agreement -- four years ago! -- seemed to be that the UEFI spec
> should drop the properties table definition in some time, and then edk2
> could remove the reference implementation too.
>
> See the attached discussion.
>
> Given that the properties table had been deprecated in the UEFI spec
> even in Feb 2016, I think it's now high time to remove it altogether
> (both spec and edk2).
>
> > The existence of both is confusing.
>
> Yes, very much.
>
Agreed. We should at least log this as a task, and if anyone has the
bandwidth to do the work, plan it for the next stable tag.