2018-07-16 16:16:41

by Rich Felker

[permalink] [raw]
Subject: cpu_no_speculation omissions?

At least the Centerton (late-generation Bonnell uarch) Atom family is
omitted from the cpu_no_speculation table added by commit fec9434a12f3
to arch/x86/kernel/cpu/common.c. Is this intentional? Would a patch
adding it and possibly other omissions be welcome?

Rich


2018-07-16 16:58:13

by Thomas Gleixner

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Mon, 16 Jul 2018, Rich Felker wrote:

> At least the Centerton (late-generation Bonnell uarch) Atom family is
> omitted from the cpu_no_speculation table added by commit fec9434a12f3
> to arch/x86/kernel/cpu/common.c. Is this intentional? Would a patch
> adding it and possibly other omissions be welcome?

Probably. Dave?

Thanks,

tglx

2018-07-16 17:30:57

by Dave Hansen

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On 07/16/2018 09:56 AM, Thomas Gleixner wrote:
> On Mon, 16 Jul 2018, Rich Felker wrote:
>> At least the Centerton (late-generation Bonnell uarch) Atom family is
>> omitted from the cpu_no_speculation table added by commit fec9434a12f3
>> to arch/x86/kernel/cpu/common.c. Is this intentional? Would a patch
>> adding it and possibly other omissions be welcome?
>
> Probably. Dave?

IIRC, Alan Cox was compiling a list on what is affected vs. not. He
would know way better than I.

Rich, did you have a specific set of family/model combinations that you
thought were missed?

2018-07-16 18:56:32

by Alan

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Mon, 2018-07-16 at 10:28 -0700, Dave Hansen wrote:
> On 07/16/2018 09:56 AM, Thomas Gleixner wrote:
> > On Mon, 16 Jul 2018, Rich Felker wrote:
> > > At least the Centerton (late-generation Bonnell uarch) Atom
> > > family is
> > > omitted from the cpu_no_speculation table added by commit
> > > fec9434a12f3
> > > to arch/x86/kernel/cpu/common.c. Is this intentional? Would a
> > > patch
> > > adding it and possibly other omissions be welcome?
> >
> > Probably. Dave?
>
> IIRC, Alan Cox was compiling a list on what is affected vs. not. He
> would know way better than I.

The pre Silvermont atom cores are in order. When I did the original
list I didn't bother with all the 32bit cores as we didn't have any
32bit mitigations then.

Alan


2018-07-16 19:22:04

by Thomas Gleixner

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Mon, 16 Jul 2018, Alan Cox wrote:
> On Mon, 2018-07-16 at 10:28 -0700, Dave Hansen wrote:
> > On 07/16/2018 09:56 AM, Thomas Gleixner wrote:
> > > On Mon, 16 Jul 2018, Rich Felker wrote:
> > > > At least the Centerton (late-generation Bonnell uarch) Atom
> > > > family is
> > > > omitted from the cpu_no_speculation table added by commit
> > > > fec9434a12f3
> > > > to arch/x86/kernel/cpu/common.c. Is this intentional? Would a
> > > > patch
> > > > adding it and possibly other omissions be welcome?
> > >
> > > Probably. Dave?
> >
> > IIRC, Alan Cox was compiling a list on what is affected vs. not. He
> > would know way better than I.
>
> The pre Silvermont atom cores are in order. When I did the original
> list I didn't bother with all the 32bit cores as we didn't have any
> 32bit mitigations then.

At least we should give the users that warm and fuzzy feeling that they are
not affected.

Thanks,

tglx

2018-07-16 19:56:05

by Rich Felker

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Mon, Jul 16, 2018 at 09:20:58PM +0200, Thomas Gleixner wrote:
> On Mon, 16 Jul 2018, Alan Cox wrote:
> > On Mon, 2018-07-16 at 10:28 -0700, Dave Hansen wrote:
> > > On 07/16/2018 09:56 AM, Thomas Gleixner wrote:
> > > > On Mon, 16 Jul 2018, Rich Felker wrote:
> > > > > At least the Centerton (late-generation Bonnell uarch) Atom
> > > > > family is
> > > > > omitted from the cpu_no_speculation table added by commit
> > > > > fec9434a12f3
> > > > > to arch/x86/kernel/cpu/common.c. Is this intentional? Would a
> > > > > patch
> > > > > adding it and possibly other omissions be welcome?
> > > >
> > > > Probably. Dave?
> > >
> > > IIRC, Alan Cox was compiling a list on what is affected vs. not. He
> > > would know way better than I.
> >
> > The pre Silvermont atom cores are in order. When I did the original
> > list I didn't bother with all the 32bit cores as we didn't have any
> > 32bit mitigations then.
>
> At least we should give the users that warm and fuzzy feeling that they are
> not affected.

It's not just fuzzies -- my box was actually affected by slowdown for
mitigation without adding "nopti" to boot command line, which is
rather unfortunate when you're running a severely-slower box already
for the sake of not being affected by these vulnerabilities.

Since Dave asked, mine is a Centerton as I originally mentioned, Atom
S1260, which seems to be the fastest in-order Intel chip with boards
still commercially available. I'm not sure if there are other
omissions that are actually interesting. I have several other Atoms
too, but I think all of those are covered by the existing table.

Rich

2018-07-16 20:40:00

by Thomas Gleixner

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Mon, 16 Jul 2018, Rich Felker wrote:
> On Mon, Jul 16, 2018 at 09:20:58PM +0200, Thomas Gleixner wrote:
> > On Mon, 16 Jul 2018, Alan Cox wrote:
> > > On Mon, 2018-07-16 at 10:28 -0700, Dave Hansen wrote:
> > > > On 07/16/2018 09:56 AM, Thomas Gleixner wrote:
> > > > > On Mon, 16 Jul 2018, Rich Felker wrote:
> > > > > > At least the Centerton (late-generation Bonnell uarch) Atom
> > > > > > family is
> > > > > > omitted from the cpu_no_speculation table added by commit
> > > > > > fec9434a12f3
> > > > > > to arch/x86/kernel/cpu/common.c. Is this intentional? Would a
> > > > > > patch
> > > > > > adding it and possibly other omissions be welcome?
> > > > >
> > > > > Probably. Dave?
> > > >
> > > > IIRC, Alan Cox was compiling a list on what is affected vs. not. He
> > > > would know way better than I.
> > >
> > > The pre Silvermont atom cores are in order. When I did the original
> > > list I didn't bother with all the 32bit cores as we didn't have any
> > > 32bit mitigations then.
> >
> > At least we should give the users that warm and fuzzy feeling that they are
> > not affected.
>
> It's not just fuzzies -- my box was actually affected by slowdown for

I was talking about 32bit. Yours seem to be 64bit.

Thanks,

tglx

2018-07-26 13:33:23

by Jiri Kosina

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Mon, 16 Jul 2018, Alan Cox wrote:

> > > > At least the Centerton (late-generation Bonnell uarch) Atom family
> > > > is omitted from the cpu_no_speculation table added by commit
> > > > fec9434a12f3 to arch/x86/kernel/cpu/common.c. Is this intentional?
> > > > Would a patch adding it and possibly other omissions be welcome?
> > >
> > > Probably. Dave?
> >
> > IIRC, Alan Cox was compiling a list on what is affected vs. not. He
> > would know way better than I.
>
> The pre Silvermont atom cores are in order. When I did the original
> list I didn't bother with all the 32bit cores as we didn't have any
> 32bit mitigations then.

Now that 32bit PTI is in tip, I believe refresh of that list would be
justified. CCing Joerg.

--
Jiri Kosina
SUSE Labs

2018-07-26 14:20:19

by Jörg Rödel

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Thu, Jul 26, 2018 at 03:29:53PM +0200, Jiri Kosina wrote:
> On Mon, 16 Jul 2018, Alan Cox wrote:
> > The pre Silvermont atom cores are in order. When I did the original
> > list I didn't bother with all the 32bit cores as we didn't have any
> > 32bit mitigations then.
>
> Now that 32bit PTI is in tip, I believe refresh of that list would be
> justified. CCing Joerg.

Agreed, PTI for x86-32 is in -tip now and will likely make it into
v4.19. In-order cores (which should not be not affected by Meltdown)
should be on the list, otherwise they will get a huge performance hit.


Regards,

Joerg

2018-07-26 15:58:09

by Rich Felker

[permalink] [raw]
Subject: Re: cpu_no_speculation omissions?

On Thu, Jul 26, 2018 at 04:18:42PM +0200, Joerg Roedel wrote:
> On Thu, Jul 26, 2018 at 03:29:53PM +0200, Jiri Kosina wrote:
> > On Mon, 16 Jul 2018, Alan Cox wrote:
> > > The pre Silvermont atom cores are in order. When I did the original
> > > list I didn't bother with all the 32bit cores as we didn't have any
> > > 32bit mitigations then.
> >
> > Now that 32bit PTI is in tip, I believe refresh of that list would be
> > justified. CCing Joerg.
>
> Agreed, PTI for x86-32 is in -tip now and will likely make it into
> v4.19. In-order cores (which should not be not affected by Meltdown)
> should be on the list, otherwise they will get a huge performance hit.

Does anyone have a complete list of pre-speculation Atom models? Mine
is one of the few 64-bit ones, but there are lots of 32-bit Atoms that
PTI would hit hard.

Rich