Re: [RFC PATCH 2/9] memory: scrub: sysfs: Add Documentation entries for set of scrub attributes

From: Jonathan Cameron
Date: Wed Oct 11 2023 - 12:36:06 EST


On Fri, 6 Oct 2023 13:06:53 +0000
"Sridharan, Vilas" <Vilas.Sridharan@xxxxxxx> wrote:

> [AMD Official Use Only - General]
>
> I do not believe AMD has implemented RASF/RAS2 at all.
>
> We are looking at it, but our initial impression is that it is insufficiently flexible for general use. (Not just for this feature, but for others in the future.)
>
> -Vilas

Hi Vilas,

So obvious question is - worth fixing?

I'm not particularly keen to see 10+ different ways of meeting this requirement.

Probably not too bad if that's 10+ drivers implementing the same userspace ABI, but
definitely don't want 10 drivers and 10 ABIs.

Jonathan

>
> -----Original Message-----
> From: Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx>
> Sent: Friday, October 6, 2023 9:02 AM
> To: David Rientjes <rientjes@xxxxxxxxxx>
> Cc: Jiaqi Yan <jiaqiyan@xxxxxxxxxx>; Luck, Tony <tony.luck@xxxxxxxxx>; Grimm, Jon <Jon.Grimm@xxxxxxx>; dave.hansen@xxxxxxxxxxxxxxx; Sridharan, Vilas <Vilas.Sridharan@xxxxxxx>; linuxarm@xxxxxxxxxx; shiju.jose@xxxxxxxxxx; linux-acpi@xxxxxxxxxxxxxxx; linux-mm@xxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; rafael@xxxxxxxxxx; lenb@xxxxxxxxxx; naoya.horiguchi@xxxxxxx; james.morse@xxxxxxx; david@xxxxxxxxxx; jthoughton@xxxxxxxxxx; somasundaram.a@xxxxxxx; erdemaktas@xxxxxxxxxx; pgonda@xxxxxxxxxx; duenwen@xxxxxxxxxx; mike.malvestuto@xxxxxxxxx; gthelen@xxxxxxxxxx; tanxiaofei@xxxxxxxxxx; prime.zeng@xxxxxxxxxxxxx
> Subject: Re: [RFC PATCH 2/9] memory: scrub: sysfs: Add Documentation entries for set of scrub attributes
>
> Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.
>
>
> On Wed, 4 Oct 2023 20:18:12 -0700 (PDT)
> David Rientjes <rientjes@xxxxxxxxxx> wrote:
>
> > On Wed, 27 Sep 2023, Jiaqi Yan wrote:
> >
> > > > > 1. I am not aware of any chip/platform hardware that implemented
> > > > > the hw ps part defined in ACPI RASF/RAS2 spec. So I am curious
> > > > > what the RAS experts from different hardware vendors think about
> > > > > this. For example, Tony and Dave from Intel, Jon and Vilas from
> > > > > AMD. Is there any hardware platform (if allowed to disclose)
> > > > > that implemented ACPI RASF/RAS2? If so, will vendors continue to
> > > > > support the control of patrol scrubber using the ACPI spec? If
> > > > > not (as Tony said in [1], will the vendor consider starting some future platform?
> > > > >
> > > > > If we are unlikely to get the vendor support, creating this ACPI
> > > > > specific sysfs API (and the driver implementations) in Linux
> > > > > seems to have limited meaning.
> > > >
> > > > There is a bit of a chicken and egg problem here. Until there is
> > > > reasonable support in kernel (or it looks like there will be),
> > > > BIOS teams push back on a requirement to add the tables.
> > > > I'd encourage no one to bother with RASF - RAS2 is much less
> > > > ambiguous.
> > >
> > > Here mainly to re-ping folks from Intel (Tony and Dave) and AMD
> > > (Jon and Vilas) for your opinion on RAS2.
> > >
> >
> > We'll need to know from vendors, ideally at minimum from both Intel
> > and AMD, whether RAS2 is the long-term vision here. Nothing is set in
> > stone, of course, but deciding whether RAS2 is the standard that we
> > should be rallying around will help to guide future development
> > including in the kernel.
> >
> > If RAS2 is insufficient for future use cases or we would need to
> > support multiple implementations in the kernel for configuring the
> > patrol scrubber depending on vendor, that's great feedback to have.
> >
> > I'd much rather focus on implementing something in the kernel that we
> > have some clarity about the vendors supporting, especially when it
> > comes with user visible interfaces, as opposed to something that may
> > not be used long term. I think that's a fair ask and that vendor
> > feedback is required here?
>
> Agreed and happy to have feedback from Intel and AMD + all the other CPU vendors who make use of ACPI + all the OEMs who add stuff well beyond what Intel and AMD tell them to :) I'll just note a lot of the ACPI support in the kernel covers stuff not used on mainstream x86 platforms because they are doing something custom and we didn't want 2 + X custom implementations...
>
> Some other interfaces for scrub control (beyond existing embedded ones) will surface in the next few months where RAS2 is not appropriate.
>
> Jonathan
>
>