Intel CPUs' Kernel Page Table Isolation (KPTI) fix

What's a 30% CPU performance hit between friends?

----- Forwarded message from Vincent Deffontaines <[email protected]> -----

Date: Wed, 03 Jan 2018 11:43:40 +0100
From: Vincent Deffontaines <[email protected]>
To: [email protected]
Subject: Huge Intel CPU Bug Allegedly Causes Kernel Memory Vulnerability With
	Up To 30% Performance Hit

Greetings,

And happy new year.

This is hitting all of us soon.
In short: a huge (and long lasting) hardware bug, present in about
all Intel CPUs that have been sold for the last 10 years.
It is not fixable by microcode, and requires ugly patching from the
kernel layer . Other OSes such as Microsoft are concerned as well.

Nice overview of the thing :
https://hothardware.com/news/intel-cpu-bug-kernel-memory-isolation-linux-windows-macos

More detail :
http://pythonsweetness.tumblr.com/post/169166980422/the-mysterious-case-of-the-linux-page-table

Linus's patch :
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5aa90a84589282b87666f92b6c3c917c8080a9bf


Vincent Deffontaines


----- End forwarded message -----
----- Forwarded message from Henrique de Moraes Holschuh <[email protected]> -----

Date: Wed, 3 Jan 2018 12:05:12 -0200
From: Henrique de Moraes Holschuh <[email protected]>
To: [email protected]
Subject: Re: Huge Intel CPU Bug Allegedly Causes Kernel Memory Vulnerability
	With  Up To 30% Performance Hit

On Wed, 03 Jan 2018, Vincent Deffontaines wrote:
> It is not fixable by microcode, and requires ugly patching from the kernel
> layer. Other OSes such as Microsoft are concerned as well.

Nobody but Intel knows whether it is fixable in microcode or not, for a
given processor family.  And Intel has *NOT* published any public
information about this yet, AFAIK.

Take for example the LAPIC memory sinkhole issue[1], which is quite
dangerous (allows for persistent, stealth ring -2 malware that is
invisible to the BIOS, hypervisor, and operating system).  It was widely
discussed, it was officialy acknowledged, and it was touted everywhere
as "impossible to fix" in microcode because it was an architectural
thing.

Well, the LAPIC issue was not just fixable in microcode: it *was indeed
fixed* by a set of microcode updates.  The fixing was done silently(!),
though.  My best guess is that this particular set of updates could not
be issued to the wide public because it *does* change an architectural
behavior, and thus it is supposed to always be deployed along with a
BIOS update to ensure compatibility.  But it is available to vendors,
and at least one vendor of corporate desktops and servers *did* deploy
such firmware updates with the new microcode for systems with processors
as old as the Core2 duo...

Most likely, this new X86_BUG_CPU_INSECURE issue either cannot be fixed
in microcode on every affected processor model (some have more hardwired
paths than others) so you would still need a software-level fix anyway,
or it is just too painful performance-wise to do it in the microcode.

But this, too, is just speculation.  Until Intel document it somewhere
public, I am not assuming it is impossible to fix it with an microcode
update.

[1] http://www8.hp.com/us/en/intel-processor-memory-sinkhole.html

-- 
  Henrique Holschuh


----- End forwarded message -----
----- Forwarded message from Rohan McLeod <[email protected]> -----

Date: Thu, 4 Jan 2018 16:39:02 +1100
From: Rohan McLeod <[email protected]>
To: Luv Main <luv-main-mJLk/csuh/[email protected]>
Subject: Re: Intel CPUs

Russell Coker via luv-main wrote:
> https://www.fool.com/investing/2017/12/19/intels-ceo-just-sold-a-lot-of-stock.aspx
>
> https://www.theregister.co.uk/2018/01/02/intel_cpu_design_flaw/
>
> Lev just tweeted the above links.  This is annoying, it means rebooting all 
> systems with Intel CPUs for which security is important and also ongoing 
> performance loss for all modern systems with Intel CPUs.

Well it seems Intel has been very naughty

https://www.theregister.co.uk/2018/01/04/intels_spin_the_registers_annotations/

But then again perhaps,
"every-modern processor has-unfixable security flaws"

https://arstechnica.com/gadgets/2018/01/meltdown-and-spectre-every-modern-processor-has-unfixable-security-flaws/

regards Rohan McLeod

_______________________________________________
luv-main mailing list
luv-main-mJLk/csuh/[email protected]
https://lists.luv.asn.au/cgi-bin/mailman/listinfo/luv-main

----- End forwarded message -----