6 ms·
Their weakest point is the hypervisor, Xen, which while a better choice than Linux/KVM, is still extremely bloated and has a poor security history. Thankfully,
by snvzz 9y ago
Their weakest point is the hypervisor, Xen, which while a better choice than Linux/KVM, is still extremely bloated and has a poor security history.
Thankfully, better designs such as seL4's VMM do exist, although it might need a little more work [1] until usable for the purpose.
[1] https://sel4.systems/Info/Roadmap/ https://sel4.systems/Info/Roadmap/
- dijit 9y agoCould you clarify "Better choice"? I've been using KVM/Xen/VMware for some time and always enjoyed it. And since Amazon and Google especially are going all in on KVM I'm surprised to hear the Xen is a better choice.
- hennsen 9y agoAmazon is going KVM?
- hennsen 9y agoAh - https://www.theregister.co.uk/2017/11/09/aws_deletes_new_hypervisor_kvm/ https://www.theregister.co.uk/2017/11/09/aws_deletes_new_hyp... Sorry for not googling before asking...
- snvzz 9y ago>Could you clarify "Better choice"? KVM is, like VMware, a Type 2 hypervisor. [1] Xen is a proper Type 1 hypervisor. [1] https://microkerneldude.wordpress.com/2010/10/14/much-ado-about-type-2/ https://microkerneldude.wordpress.com/2010/10/14/much-ado-ab...
- aleden 9y agoIt should be noted that KVM supports many different archs, and it lives inside the mainline Linux kernel while VMware's kernel modules are out-of-tree. I think this fact is an important difference (also that qemu-system-* are open-source, while vmware is not).
- theossuary 9y agoWhy is a type 1 hypervisor instantly considered more secure though? I'd assume using Linux, instead of rolling your own code to interface with hardware, would make you more secure?
- snvzz 9y agoIn the Linux vs Xen example, the TCB is much bigger with Linux. The idea is to keep the TCB as small as possible, with an emphasis on restricting the code size that's actually running privileged.
- monocasa 9y agosel4's virtualization support make it a type 2 hypervisor. Akaros too, which IMO has the right model for virtualization with it's 'VM threads' concept. All 'type 2' really means is that the kernel directly supports running threads in ring 3 in addition to ring 0. I guess it's your use of 'proper' that bugged me.
- mmrezaie 9y agoXen's hypervisor's size is very small. Qubes is about security and trustability of the whole system. In operating systems for measuring the trustability of the system, one very important measure is the lines of the code. Xen has a smaller footprint in the hypervisor part. Additionally, Xen has a robust model isolation for the drivers. That's why they went for Xen not KVM. But boy I wish to see more seL4. It was sad to see Gnu Hurd/seL4 didn't make it.
- walterbell 9y agoQubes mailing list thread about hypervisor choices: https://groups.google.com/forum/m/#!topic/qubes-devel/jEe4pQ2zSKo https://groups.google.com/forum/m/#!topic/qubes-devel/jEe4pQ... > It seems one major residing problem with KVM is the Linux kernel (which is large and vulnerable). A port of KVM to a thinner base layer would obviate those issues.
- nickpsecurity 9y agoOne of the trends I told Joanna about (i.e secure L4 kernels) led to folks developing exactly that. It was called KVM-L4. Here you go. http://os.inf.tu-dresden.de/papers_ps/liebergeld-diplom.pdf http://os.inf.tu-dresden.de/papers_ps/liebergeld-diplom.pdf Complexity was still yoo high. Most in high-assurance security were trying stuff like Nova microhypervisor as a result. KVM on separation kernels might be worth further investigation for these platforms that will stay on KVM regardless.
- xyzzyz 9y agoThe problem with Xen is that no major industry player is backing it, especially with Amazon going KVM now. (disclaimer: working at Google on virtualization security)
- pjmlp 9y agoWhat about Cisco?
- 9y ago
- monocasa 9y agoNo, I'd say that the weakest point is the IPC marshalling necessary to connect all of the containers together into a cohesive system. That's what I'd attack first.
- JoachimSchipper 9y agoA good place to look, but do note that that's the code written by the Qubes OS people - presumably, it's written with security in mind. Of course, Xen has had more eyeballs, so...
- monocasa 9y agoChrome's IPC was written with security in mind too, but most of the sandbox escape exploits have been around IPC marshalling. Unlike the nitty gritty of how the sandbox works, the IPC changes often with new releases. And quite frankly it isn't as fun, cool, or interesting as VMMs or other sandboxing techniques, so a lot of the time it isn't given the close eye that it should.
- mtgx 9y agoThe Genode team proposed some integration with Qubes a while ago, but not sure if the discussion went anywhere from that: https://secure-os.org/pipermail/desktops/2015-November/000008.html https://secure-os.org/pipermail/desktops/2015-November/00000...
- X86BSD 9y agoYou also have access to BHyve on FreeBSD for a good hypervisor.
- floatboth 9y agoJust like KVM, bhyve includes a whole unix kernel in the TCB. Sure it's a better one :) but still. Tiny hypervisors like NOVA http://hypervisor.org http://hypervisor.org, seL4-based are the ideal solution, but sadly no one seems to be pushing to make them usable and production-ready :(
- jjawssd 9y agoIs it possible to run secure code on any Intel microprocessor which supports the x86 instruction set? I do not think so.
- morganvachon 9y agoPre-2006 systems, maybe. I have a PIII based laptop that I'm reasonably sure doesn't contain any malicious microcode or BIOS shenanigans. It certainly doesn't have IME or equivalent. However, that was the CPU that started Intel's serial number controversy, and while I do have a BIOS setting to turn it off, is it really off? It's a pretty deep rabbit hole if you really want to go down it. You can make a case for not trusting any CPU that you didn't design and fab yourself, and even then you have to watch out for your own mistakes and bugs that can be used against you.