Venom Security Vulnerability

[KMyers]

Hello All,

It seems like there is a major bug in several server visualization platforms including QEMU and Xen that could allow an attacker to take over a complete node.

http://www.zdnet.com/article/venom-security-flaw-millions-of-virtual-machines-datacenters/?tag=nl.e589&s_cid=e589&ttag=e589&ftag=TREc64629f

Any thoughts? 

 

[Gang Starr]

Any thoughts? 

Yes, we shall be thankful it was detected and we shall pray and wait for a fix .

 

[KMyers]

Edit : More details - http://venom.crowdstrike.com/

 

[DomainBop]

Subject     [sECURITY] [DSA 3259-1] qemu security update


Date     Today 18:07


-----BEGIN PGP SIGNED MESSAGE-----


Hash: SHA1


- -------------------------------------------------------------------------


Debian Security Advisory DSA-3259-1                   


http://www.debian.org/security/                       Moritz Muehlenhoff


May 13, 2015                           http://www.debian.org/security/faq


- -------------------------------------------------------------------------


Package        : qemu


CVE ID         : CVE-2014-9718 CVE-2015-1779 CVE-2015-2756 CVE-2015-3456


Several vulnerabilities were discovered in the qemu virtualisation


solution:


CVE-2014-9718


    It was discovered that the IDE controller emulation is susceptible


    to denial of service.


CVE-2015-1779


    Daniel P. Berrange discovered a denial of service vulnerability in


    the VNC web socket decoder.


CVE-2015-2756


    Jan Beulich discovered that unmediated PCI command register could


    result in denial of service.


CVE-2015-3456


    Jason Geffner discovered a buffer overflow in the emulated floppy


    disk drive, resulting in the potential execution of arbitrary code.


For the oldstable distribution (wheezy), these problems have been fixed


in version 1.1.2+dfsg-6a+deb7u7 of the qemu source package and in version


1.1.2+dfsg-6+deb7u7 of the qemu-kvm source package. Only CVE-2015-3456


affects oldstable.


For the stable distribution (jessie), these problems have been fixed in


version 1:2.1+dfsg-12.


For the unstable distribution (sid), these problems will be fixed soon.


We recommend that you upgrade your qemu packages.


Further information about Debian Security Advisories, how to apply


these updates to your system and frequently asked questions can be


found at: https://www.debian.org/security/

 

[KuJoe]

Has anybody seen an update for CentOS 5 yet? Looks like CentOS 6 is taken care of but nothing for CentOS 5 in any of the mirrors I've checked.

 

[zzrok]

The only thing on the announcement mailing list so far for Centos 5 is xen.

 

[MannDude]

These vulnerabilities keep getting more and more creative names.

I predict we'll see a "Nuclear Aids" vulnerability in the next two years.

 

[KuJoe]

These vulnerabilities keep getting more and more creative names.

I predict we'll see a "Nuclear Aids" vulnerability in the next two years.

Imagine how much sooner the exploit would have been reported to the right channels if they didn't have to figure out a fancy name and  wait for a logo to be created, website built, domain purchased, and hosting provisioned.

 

[rds100]

@KuJoe Just checked our CentOS mirror - there were these new/updated packages for CentOS 5:

5.11/updates/i386/RPMS/xen-3.0.3-146.el5_11.i386.rpm
5.11/updates/i386/RPMS/xen-3.0.3-146.el5_11.i686.rpm
5.11/updates/i386/RPMS/xen-devel-3.0.3-146.el5_11.i386.rpm
5.11/updates/i386/RPMS/xen-devel-3.0.3-146.el5_11.i686.rpm
5.11/updates/i386/RPMS/xen-libs-3.0.3-146.el5_11.i386.rpm
5.11/updates/i386/RPMS/xen-libs-3.0.3-146.el5_11.i686.rpm
 

But nothing for qemu or kvm ont CentOS 5.

 

[joepie91]

Imagine how much sooner the exploit would have been reported to the right channels if they didn't have to figure out a fancy name and  wait for a logo to be created, website built, domain purchased, and hosting provisioned.

Doubtful. Vendors typically get advance notice.

 

[Francisco]

Since I just woke up, has anyone confirmed if checkpointing a KVM would work or does that actually save a copy of the binary into memory too?

From what http://libvirt.org/guide/html/Application_Development_Guide-Lifecycle-Save.html is saying, it's only saving the VM's memory and not the entire binary, which means it should technically be able to boot the VPS back up on the new binary and continue onwards.

Since there's no POC out there's no way to know, but the above makes sense to me.

Any feedback?

EDIT - Since they said live-migrating a VM from a node, updating the node, and migrating ti back is perfectly OK and will have them running on an updated binary, a 'virsh save' should be fine. I still wish there was a way to test.

Francisco

 

[Francisco]

https://github.com/juergh/lqs2mem outlines what a 'virsh save' looks like and it seems to be what I was thinking.

Since the exploit isn't in the BIOS or .rom files, it'd be OK.

Francisco

 

[rds100]

Updated kvm packages for CentOS 5 were released too.

5.11/updates/x86_64/RPMS/kmod-kvm-83-272.el5.centos.x86_64.rpm
5.11/updates/x86_64/RPMS/kmod-kvm-debug-83-272.el5.centos.x86_64.rpm
5.11/updates/x86_64/RPMS/kvm-83-272.el5.centos.x86_64.rpm
5.11/updates/x86_64/RPMS/kvm-qemu-img-83-272.el5.centos.x86_64.rpm
5.11/updates/x86_64/RPMS/kvm-tools-83-272.el5.centos.x86_64.rpm
 

 

[HalfEatenPie]

I believe Proxmox has yet to receive updates on this.  

Anyone else confirm? 

 

[dcdan]

Time to switch to OpenVZ :lol:

Generally whenever there is an OpenVZ-related exploit first thing you hear is "everyone should switch to KVM". For some reason it does not work the other way.

 

[HalfEatenPie]

Time to switch to OpenVZ :lol:

Generally whenever there is an OpenVZ-related exploit first thing you hear is "everyone should switch to KVM". For some reason it does not work the other way.

Everyone should switch to VMWare.

Kek.

 

[DomainBop]

Everyone should switch to VMWare.

rebuttal to the pie's suggestion -->  http://www.vmware.com/security/advisories/

 

[dabtech]

I believe Proxmox has yet to receive updates on this.  

Anyone else confirm? 

http://forum.proxmox.com/threads/22127-Venom-vulnerablity?p=111920#post111920

In pvetest, hopefully it'll hit the stable repo's soon.

 

[Hxxx]

This is why exactly the "cloud" (see the ")  is not ready for PHI. Doesn't matter how much certified is the provider. Bugs like these make all the certifications go down the toilet.

 

[KuJoe]

This is why exactly the "cloud" (see the ")  is not ready for PHI. Doesn't matter how much certified is the provider. Bugs like these make all the certifications go down the toilet.

In reality, how many medical facilities are buying VPSs for their infrastructure or allowing outside parties on their servers? I doubt many if any at all.