Hi Matthew, We have been working on a Kernel Hardening project. Please find slides at http://events.linuxfoundation.org/sites/events/files/slides/Kernel%20Protection-Nakajima.pdf . We presented this idea in KVM Forum 2016. The idea is to protect CPU/platform resources and kernel managed resources (IDT, kernel page tables etc.) during execution of a VM. This approach is extended to baremetal/host OS by switching the execution of host OS into guest mode and monitoring the host OS with a very thin hypervisor, probably kvm module extension. Currently we have a PoC, contained in kvm module, for switching the host OS into guest mode. We are open for collaboration and feedback. Thanks -Sainath > -----Original Message----- > From: kvm-owner@xxxxxxxxxxxxxxx [mailto:kvm-owner@xxxxxxxxxxxxxxx] On > Behalf Of Matthew Giassa > Sent: Tuesday, February 14, 2017 7:32 PM > To: Steve Rutherford <srutherford@xxxxxxxxxx> > Cc: Jidong Xiao <jidong.xiao@xxxxxxxxx>; kernel- > hardening@xxxxxxxxxxxxxxxxxx; KVM <kvm@xxxxxxxxxxxxxxx>; Rik van Riel > <riel@xxxxxxxxxx> > Subject: Re: Introduction + new project: "rootkit detection using > virtualization". > > On 2017-02-14 01:25 PM, Steve Rutherford wrote: > > On Tue, Feb 14, 2017 at 10:06 AM, Matthew Giassa <matthew@xxxxxxxxxx> > wrote: > >> Hi Jidong, > >> > >> You are correct on all the points noted above:My goal is to develop a > >> production-ready, non-academic implementation of such a tool. I'm in > >> it for the long haul. > > Is your goal for this to work on all architectures, or are you > > planning to focus on Intel-x86 (or AMD-x86, or ARM, or...)? > >> > >> On Fri, Feb 10, 2017 at 7:43 PM, Jidong Xiao <jidong.xiao@xxxxxxxxx> > wrote: > >>> Thanks Matthew. So if I understand correctly, even though many > >>> people have proposed similar solutions, none of them have actually > >>> contributed their code (of their solution) into Qemu/KVM. To make it > >>> "real" (i.e., as a part of Qemu/KVM code) is your goal, right? That sounds > interesting! > >>> > >>> -Jidong > >>> > >>> On Fri, Feb 10, 2017 at 8:21 PM, Matthew Giassa <matthew@xxxxxxxxxx> > wrote: > >>>> > >>>> On 2017-02-10 03:18 PM, Jidong Xiao wrote: > >>>>> > >>>>> Sorry, I have to resend this again, as the original two emails > >>>>> were blocked because of the url. > >>>>> > >>>>> "Rootkit detection using virtualization" has been widely studied > >>>>> for a decade. Is the approach you are going to use different from > >>>>> all of these existing ones: > >>>>> > >>>>> "Survey: Virtual Machine Introspection Based System Monitoring and > >>>>> Malware Detection Techniques" - by Haofu Liao at University of > Rochester. > >>>>> > >>>>> -Jidong > >>>> > >>>> > >>>> On 2017-02-10 05:37 PM, Rik van Riel wrote: > >>>>> > >>>>> One of the things that Matthew can do is build on the read-only > >>>>> memory protections in the kernel, and have the hypervisor enforce > >>>>> that the memory the kernel marks as read-only is never written > >>>>> from inside the virtual machine, until the next reboot. > >>>>> > >>>>> That seems like it might be a useful place to start, since it > >>>>> would immediately make the other read-only protections that people > >>>>> are working on much harder to get around, at least inside virtual > >>>>> machines. > >>>>> > >>>> > >>>> > >>>> My initial plan was to start with what Rik proposed, and focus on > >>>> additional memory protections. With respect to long-term plans, a > >>>> lot of my work/research so far has been focused on implementing a > >>>> system similar to that presented by Payne et al (ie: Lares). > >>>> > >>>> -Matthew Giassa > >>> > >>> > >> > >> > >> > >> -- > >> > ============================================================ > >> Matthew Giassa, MASc, BASc, EIT > >> Principal Developer; Security and Embedded Systems Specialist > >> linkedin: https://ca.linkedin.com/in/giassa > >> e-mail: matthew@xxxxxxxxxx > >> website: www.giassa.net > > My initial aim is x86/x64 targets, unless there are additional resources I can > tap into for expanding to ARM. If I can get a working prototype up and running > and into "staging", then expanding to ARM architecture would be viable. >
