Is SKAS still required for UML

Is SKAS still required for UML

[Jan Engelhardt]

Hi,


the UML page at http://user-mode-linux.sourceforge.net/old/skas.html 
mentions that "The UML kernel is present in the address space of each of 
its processes, and, by default, is writeable".

I tried to put this to a test, and actually failed to modify the 
UML kernel/memory image from within it. I had a simple kernel module 
with 'int val = 2;' and upon loading this, done printk("Val is at %p\n", 
&val); to get to know the address. A userspace program inside the UML 
then tried to dereference that address and read the value, but that 
ended in a segfault. /proc/xxx/maps also does not show the UML kernel 
being mapped in any process inside the UML. Note that I was running in 
SKAS0 mode, both host and guest are all 64-bit. Is the NX bit of the 
64-bit platform securing things off, or has SKAS3 become sort of 
obsolete?


Jan

Re: Is SKAS still required for UML

[Jeff Dike]

On Wed, Aug 27, 2008 at 12:57:53PM -0400, Jan Engelhardt wrote:
> the UML page at http://user-mode-linux.sourceforge.net/old/skas.html 
> mentions that "The UML kernel is present in the address space of each of 
> its processes, and, by default, is writeable".

That stuff, as you can see by the "old" in the URL, is old.

Since skas0 became the default, that statement is no longer true.
> 
> I tried to put this to a test, and actually failed to modify the 
> UML kernel/memory image from within it. I had a simple kernel module 
> with 'int val = 2;' and upon loading this, done printk("Val is at %p\n", 
> &val); to get to know the address. A userspace program inside the UML 
> then tried to dereference that address and read the value, but that 
> ended in a segfault. /proc/xxx/maps also does not show the UML kernel 
> being mapped in any process inside the UML. Note that I was running in 
> SKAS0 mode, both host and guest are all 64-bit. Is the NX bit of the 
> 64-bit platform securing things off, or has SKAS3 become sort of 
> obsolete?

SKAS3 is still a significant performance boost.

      	       		     Jeff

-- 
Work email - jdike at linux dot intel dot com

Re: Is SKAS still required for UML

[Jan Engelhardt]

On Wednesday 2008-08-27 15:06, Jeff Dike wrote:
>On Wed, Aug 27, 2008 at 12:57:53PM -0400, Jan Engelhardt wrote:
>> 
>> I tried to put this to a test, and actually failed to modify the 
>> UML kernel/memory image from within it. I had a simple kernel module 
>> with 'int val = 2;' and upon loading this, done printk("Val is at %p\n", 
>> &val); to get to know the address. A userspace program inside the UML 
>> then tried to dereference that address and read the value, but that 
>> ended in a segfault. /proc/xxx/maps also does not show the UML kernel 
>> being mapped in any process inside the UML. Note that I was running in 
>> SKAS0 mode, both host and guest are all 64-bit. Is the NX bit of the 
>> 64-bit platform securing things off, or has SKAS3 become sort of 
>> obsolete?
>
>SKAS3 is still a significant performance boost.

In tests of mine, oggenc in SKAS0 mode ran at 95% of native speed, which 
is pretty good.

Re: Is SKAS still required for UML

[Jeff Dike]

On Wed, Aug 27, 2008 at 04:00:24PM -0400, Jan Engelhardt wrote:
> In tests of mine, oggenc in SKAS0 mode ran at 95% of native speed, which 
> is pretty good.

Different workloads give different results.  Yours sounds
CPU-intensive.  Kernel builds here run at 50-60% of native with skas0,
low 80's with skas3, and ~86% with skas4.

    	      	     	 Jeff

-- 
Work email - jdike at linux dot intel dot com

Re: Is SKAS still required for UML

[Nix]

On 27 Aug 2008, Jeff Dike stated:
> SKAS3 is still a significant performance boost.

Nobody's forward-ported it to recent kernels that I can see, though :/

-- 
`Not even vi uses vi key bindings for its command line.' --- PdS

Re: Is SKAS still required for UML

[greg]

On Aug 27,  3:06pm, Jeff Dike wrote:
} Subject: Re: Is SKAS still required for UML

Hi Jeff, hope the week is ending well for you.

> > I tried to put this to a test, and actually failed to modify the 
> > UML kernel/memory image from within it. I had a simple kernel module 
> > with 'int val = 2;' and upon loading this, done printk("Val is at %p\n", 
> > &val); to get to know the address. A userspace program inside the UML 
> > then tried to dereference that address and read the value, but that 
> > ended in a segfault. /proc/xxx/maps also does not show the UML kernel 
> > being mapped in any process inside the UML. Note that I was running in 
> > SKAS0 mode, both host and guest are all 64-bit. Is the NX bit of the 
> > 64-bit platform securing things off, or has SKAS3 become sort of 
> > obsolete?

> SKAS3 is still a significant performance boost.

Any thoughts or timetable on SKAS4 going mainline?  I suspect the
SKAS3 patches are even remotely compatible with current mainline
kernels.

We've had good success with SKAS4 on 2.6.24.x but thats even getting
awful long in the tooth at this point.  I've been meaning to try the
last copy of the SKAS4 patch against 2.6.26.x but haven't found the
time yet to wade through the flock of rejects which I would expect.

With all the renames in the impending 2.6.27 tree I'm afraid the
prospects are even worse.  The time seems ripe to push it towards
mainline.

>       	       		     Jeff

Thanks again for all your efforts on UML.

Have a good weekend.

}-- End of excerpt from Jeff Dike

As always,
Dr. G.W. Wettstein, Ph.D.   Enjellic Systems Development, LLC.
4206 N. 19th Ave.           Specializing in information infra-structure
Fargo, ND  58102            development.
PH: 701-281-1686
FAX: 701-281-3949           EMAIL: greg@enjellic.com
------------------------------------------------------------------------------
"Atilla The Hun's Maxim: If you're going to rape, pillage and burn, be sure
 to do things in that order."
                                -- P.J. Plauger
                                   Programming On Purpose