by Alan Cox

[permalink] [raw]

Subject: Re: 32-bit PTI with THP = userspace corruption

On Tue, 11 Sep 2018 14:12:22 +0200
Joerg Roedel <[email protected]> wrote:

> On Tue, Sep 11, 2018 at 02:58:10PM +0300, Meelis Roos wrote:
> > The machines where I have PAE off are the ones that have less memory.
> > PAE is off just for performance reasons, not lack of PAE. PAE should be
> > present on all of my affected machines anyway and current distributions
> > seem to mostly assume 686 and PAE anyway for 32-bit systems.
>
> Right, most distributions don't even provide a non-PAE kernel for their
> users anymore.
>
> How big is the performance impact of using PAE over legacy paging?

On what system. In the days of the original 36bit PAE Xeons it was around
10% when we measured it at Red Hat, but that was long ago and as you go
newer it really ought to be vanishingly small.

There are pretty much no machines that don't support PAE and are still
even vaguely able to boot a modern Linux kernel. The oddity is the
Pentium-M but most distros shipped a hack to use PAE on the Pentium M
anyway as it seems to work fine.

Alan

2018-10-21 12:38:27

On Mon 2018-10-22 09:56:42, Joerg Roedel wrote:
> On Sun, Oct 21, 2018 at 02:37:45PM +0200, Pavel Machek wrote:
> > On Tue 2018-09-18 14:00:30, Alan Cox wrote:
> > > There are pretty much no machines that don't support PAE and are still
> > > even vaguely able to boot a modern Linux kernel. The oddity is the
> > > Pentium-M but most distros shipped a hack to use PAE on the Pentium M
> > > anyway as it seems to work fine.
> >
> > I do have some AMD Geode here, in form of subnotebook. Definitely
> > newer then Pentium Ms, but no PAE...
>
> Are the AMD Geode chips affected by Meltdown?

Probably not.

I'm not saying this has meltdown/spectre etc. I'm just saying there
are relatively new machines without PAE.

Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Attachments:

(No filename) (891.00 B)
signature.asc (188.00 B)
Digital signature Download all attachments

2018-10-23 09:14:03

by Pavel Machek

[permalink] [raw]

Subject: Not-so-old machines without PAE was Re: 32-bit PTI with THP = userspace corruption

On Mon 2018-10-22 19:48:17, Alan Cox wrote:
> On Mon, 22 Oct 2018 09:56:42 +0200
> Joerg Roedel <[email protected]> wrote:
>
> > On Sun, Oct 21, 2018 at 02:37:45PM +0200, Pavel Machek wrote:
> > > On Tue 2018-09-18 14:00:30, Alan Cox wrote:
> > > > There are pretty much no machines that don't support PAE and are still
> > > > even vaguely able to boot a modern Linux kernel. The oddity is the
> > > > Pentium-M but most distros shipped a hack to use PAE on the Pentium M
> > > > anyway as it seems to work fine.
> > >
> > > I do have some AMD Geode here, in form of subnotebook. Definitely
> > > newer then Pentium Ms, but no PAE...
> >
> > Are the AMD Geode chips affected by Meltdown?
>
> Geode for AMD was just a marketing name.
>
> The AMD athlon labelled as 'Geode' will behave like any other Athlon but
> I've not seen anyone successfully implement Meltdown on the Athlon so it's
> probably ok.
>
> The earlier NatSemi ones are not AFAIK vulnerable to either. The later
> ones might do Spectre (they have branch prediction which is disabled on
> the earlier ones) but quite possibly not enough to be attacked usefully -
> and you can turn it off anyway if you care.
>
> And I doubt your subnotebook can usefully run modern Linux since the
> memory limit on most Geode was about 64MB.

Well, let me see. The machine is not too useful because of dead
battery, but I don't believe RAM would be a problem. It has 512MB or
more, IIRC. Missing PAE is, and missing instructions are. And horrible
keyboard and bad driver support from Linux. And... Ouch and fact that
I use its power supply to power something else.

It looks similar to this:
https://hexus.net/tech/news/laptop/13994-kohjinsha-launches-impressive-subnotebook-twist/
(but has no touchscreen).

Bios is "built 03/02/2007".

CPU is "AuthenticAMD" family 5 model 10, "Geode(TM) Integrated Processor
by AMD PCS". 500 MHz, flags "fpu de pse tsc msr cx8 sep pge cmov
clflush mmx mmxext 3dnowext 3dnow". 512MB RAM.

This is not a suitable compile server, but still could work as a
subnotebook given right software...

Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Attachments:

(No filename) (2.24 kB)
signature.asc (188.00 B)
Digital signature Download all attachments