heads up: kernel regression in CentOS 2.6.32 and OpenVZ 2.6.18/2.6.32 kernels creates local DoS (reb

[jarland]

Doesn't mean we work for free.. and no amount of caffeine or liquor is going to make me start believing that we should function as a free, public helpdesk <_<

Invoice me for every time I've asked Fran for advice and I'll pay that with a smile

People should be paid for what they do, Fran just does so much awesomeness that the list would probably crash WHMCS

 

[Magiobiwan]

And the PayPal fees would be outrageous!

 

[Francisco]

Is the 2.6.18 kernel line affected? Namely the latest stable version kernel.

Yus.

RHEL 2.6.32 is a complete cluster fuck since it has so much crap backported. I simply can't get a ksplice to generate properly. I had to change how some of the structs were handled

Right when I think I have it figured out? it blows up again.

Francisco

 

[kaniini]

RH still claims they can't reproduce this.  Whatever.

This is CVE-2013-2224

 

[kaniini]

RHEL 2.6.32 is a complete cluster fuck since it has so much crap backported.

I wasn't able to get hemlock.c to run successfully on Debian 6, which also uses 2.6.32.  This leads me to believe the vulnerability is specific to RHEL's kernel series.  I bet some of their modifications opened the possibility of this vulnerability.

 

[Francisco]

I wasn't able to get hemlock.c to run successfully on Debian 6, which also uses 2.6.32.  This leads me to believe the vulnerability is specific to RHEL's kernel series.  I bet some of their modifications opened the possibility of this vulnerability.

If you have a bunch of kernels to try, try something from the 2.6.39 range.

A long time ago I had a developer work on a fairly large addition to the .32 kernels and he went on a huge rant about how RHEL's .32 is pretty much .39.

Francisco

 

[kaniini]

I'll build a 2.6.39 kernel and try it out in a bit if I can spare some free time.

 

[Steven]

Redhat Bugzilla

https://bugzilla.redhat.com/show_bug.cgi?id=979936

 

[KuJoe]

OpenVZ.org released a new kernel today to address it.

 

[eva2000]

If you have a bunch of kernels to try, try something from the 2.6.39 range.


A long time ago I had a developer work on a fairly large addition to the .32 kernels and he went on a huge rant about how RHEL's .32 is pretty much .39.


Francisco

Has anyone tried CentOS with Oracle Linux 6 UEK 2.6.39 kernel (3.0.16 mainline) ?

 

[kaniini]

RedHat has confirmed it is an RHEL-specific regression caused by a badly done rebase.

 

[rds100]

@kaniini so they were able to reproduce it finaly?

 

[concerto49]

@kaniini so they were able to reproduce it finaly?

Or more like finally willing to acknowledge it.

 

[NodeBytes]

removed.