1 |
On 19 Nov 2012 at 11:37, Maxim Kammerer wrote: |
2 |
|
3 |
> On Mon, Nov 19, 2012 at 2:25 AM, Matthew Thode |
4 |
> <prometheanfire@g.o> wrote: |
5 |
> > Originally virtualization was slow on grsec/pax with either uderef or |
6 |
> > kernexec enabled. |
7 |
> |
8 |
> My impression was that UDEREF/KERNEXEC were slow in guest. Is it |
9 |
> wrong, or did these settings affect host as well? |
10 |
|
11 |
there was a bug in the per-cpu pgd feature (that those two features rely on |
12 |
on amd64) that, when enabled on the host, would cause a big guest slowdown |
13 |
(regardless of what the guest was). |
14 |
|
15 |
that these two features have a performance impact on their own is a separate |
16 |
issue and something i can't help without proper hw support (think SMEP/SMAP). |
17 |
|
18 |
> > Pipacs overcame this limitation in 3.5.4-r1 and |
19 |
> > overcame a memory commit issue kvm was having in 3.5.4-r2. He overcame |
20 |
> > it using nested page tables on newer CPUs, which means older CPUs will |
21 |
> > likely still be slow. |
22 |
> |
23 |
> So one needs at least 3.5.4-r2 in both hardened guest and host, and |
24 |
> nested page tables support in CPU? |
25 |
|
26 |
for this bug only the host matters and use more like 3.6 please since we no |
27 |
longer support 3.5 (and in a few weeks that'll become 3.7 ;) or our 2.6.32/3.2 |
28 |
stable series. |
29 |
|
30 |
nested page tables help with the inherent performance impact of per-cpu pgd |
31 |
(that is, if you enable it in your guest kernels as well), independently of |
32 |
the performance bug i fixed some months ago. |