KVM - The Linux Kernel-Based Virtual Machine
News, Blogs and Resources on the Linux (KVM) Kernel-Based Virtual Machine

KVM-83 Released: support for AMD IOMMU, qemu-svn merge

Wednesday, January 14, 2009 - 09:44 Haydn Solomon

Yesterday , kvm maintainers released kvm-83 featuring support for amd iommu for device assignment. Bear in mind however that you will need at least Linux kernel 2.6.29rc1 or higher to use these features as it requires core kernel support. There was also another qemu merge featuring virtio network fixes and a new –drive serial option.

 

 

IOMMU

AMD's IOMMU technology is a mmu that connects DMA capable I/O bus to main memory. It's job is similar to the MMU in that it translates device-visible addresses to physical addresses. Some units provide memory protection for misbehaving devices. Intel 's version of AMD's IOMMU is called Virtualization Technology for Directed I/O also known as VT-d.

For more on AMD's IOMMU, see the following link: I/O Virtualization and AMD's IOMMU

 

New serial parameter

The new serial parameter used with the –drive option was added to support the fact that windows calculates hardware uniqueness based on a hard drive serial number among other things. This parameter allows you to specify this at the command line. You could specify a serial number by starting kvm like so

qemu-system-x86_64….   –drive file=file.img,serial=xxxxx,….

Keep in mind that the serial specified is limited to 21 characters.

Changelog

The official changelog announced by the maintainers is shown below.

Changes from kvm-82:
- amd iommu support (with Linux 2.6.29+ hosts) (Joerg Roedel)
- merge qemu-svn
- qcow2 correctness fixes
- '-drive serial=...' option
- virtio-net migration fixes
- fix '--with-patched-kernel' build on Ubuntu (Nolan Leake)
- improved tsc handling (Marcelo Tosatti)
- should reduce timing errors on older Linux guests
- remove cpuid code duplication due to upstream qemu work (Amit Shah)
- guest S3 sleep (Gleb Natapov)
- load assigned devices option rom (Kechao Liu)
- '-no-kvm-pit-reinjection' option to improve timing on RHEL 3 era guests
(Marcelo Tosatti)
- remove duplicate firmware config interface probe in bios (Sebastian Herbszt)
- backward compatibility for marker_synchronize_register() (Eduardo Habkost)
- fix compile warning in tpr patching (Mark McLoughlin)
- don't install kvm headers to /usr/local
- fix smbios memory size in certain cases (Bill Rieske)
- simplify mmu shadow walking function
- stop using vmap() on guest pages (annoys ksm) (Izik Eidus)
- fix warning on MSR_VM_HSAVE_PA in dmesg
- fix i8254 timer count read (Marcelo Tosatti)
- ppc e500 core support (Hollis Blanchard)
- fix vmx real mode guest state validity checks
- make pop instruction emulation a little more generic
- implement 'ret far' instruction emulation
- properly lock i8354 timer
- remove ppc debug support (Hollis Blanchard)
- fix race freeing assigned device interrupt (Sheng Yang)
- fix pci interrupt in freeing in device assignment (Sheng Yang)
- allow disabling msi for assigned devices (Sheng Yang)
- emulate amd microcode level msr (Alexander Graf)
- helps esx boot on kvm
- make page unshadowing a little faster
- clear pending pit interrupts when unmasking the timer irq
- fixes xen-on-kvm
- mmu notifiers release method (Marcelo Tosatti)
- fixes large pages support
- expose ia64 registers to userspace (Jes Sorensen)
- fix type in has_svm() (Joe Perches)
- drop unnecessary memory clearing when allocating shadow page
(Marcelo Tosatti)
- rename "metaphysical" page attribute to "direct"
- fix irq routing with -no-kernel-irqchip

You can download kvm-83 here.

See Also

Comments

Proxmox VE 1.1 released - also including KVM-83

Friday, January 16, 2009 - 12:06 Martin Maurer (not verified)

we also included KVM-83 in our Proxmox VE.
see http://pve.proxmox.com

br, Martin

RE: Proxmox VE 1.1 released - also including KVM-83

Friday, January 16, 2009 - 15:37 Kenni Lund

[Off topic]
Proxmox is a very cool project, but it's just too bad that the developers won't support software RAID, which makes Proxmox useless in many server environments, including mine.

I would any day put my trust to the Linux kernel software RAID, which has been around for 6-8 years, instead of a hardware RAID with closed source firmware. I've had two unrecoverable harddisk crashes on HW RAID due to buggy firmware - I'm not going there again.

Proxmox 2.0

Friday, January 16, 2009 - 17:26 Anonymous (not verified)

True, but the new version 2.0 will have Software Raid and things like drbd.

ESX killer for all I care ;-)

Re: Proxmox 2.0

Friday, January 16, 2009 - 18:15 Kenni Lund

Cool, do you have a link or something confirming this? I was under the impression that Proxmox would never support software RAID (like written in several posts in their forum).

Here's the

Saturday, January 17, 2009 - 02:41 Anonymous (not verified)

Here's the Roadmap
http://pve.proxmox.com/wiki/Roadmap

But you where right, they are not listing Software RAID at all. That's such a "standard" I was fooling myself into thinking it would be in there.

I mean, if drbd is in, software Raid would be a no brainer...

No matter really, it's debian based, we can do it ourselves already.

Re: Here's the

Saturday, January 17, 2009 - 13:24 Kenni Lund

Yeah, but if you search their forums, you'll find several posts from their developer team, saying that "it simply will not work and doesn't make sense on Proxmox VE".

But when someone asks why it will not work, they most likely answer with some irrelevant and arrogant answer "because they are the developers and should know it". It seems like they have decided never to support software RAID in Proxmox VE and will keep it like this no matter what other people say.

The funny thing is that they already support software RAID in some of the other products, but they also think that software RAID is too complex for Proxmox VE.

http://www.proxmox.com/forum/showthread.php?t=838&highlight=software+raid

Software RAID for Proxmox VE

Saturday, January 17, 2009 - 15:50 Martin Maurer (not verified)

I am one of the responsible persons here (responsible for the resources) and it´s just a decision not to support Software RAID in the installer because of complexity, development and support resources and other reason in further versions.

And yes, we know how to configure software raid and we know the advantages and disadvantages.

If anyone want to use software raid - why not: if you know what you are doing you can configure it by yourself.

but for our target group we decided to go without. if you look on our competitors, also no support for software raid - so we can´t be so wrong here.

BUT for Proxmox VE 2.x, everything will be more flexible and we focus on that instead of implementing now something which will be not used in the future, just think of the amazing possibilities of btrfs ...

so its quite interesting why people call us "arrogant", we just want to give the easiest to use and best open source virtualization platform to the community with the best community support - and if you look in our forums and maillings list you will see that we have response times which you hardly get in commercial environments :-)

and just a last word: some people just complain, have limited knowledge and do not help to improve - FYI, no one donated to support software raid, not one single cent, so I assume these people are not REALLY interested - or am I wrong? If yes, why they don´t grab the source code and developed the missing feature? or pay other to do that?

Thanks for using Proxmox VE and watch out 2.x - will be the final winner!

Martin

Hi Martin I'm sorry if you

Saturday, January 17, 2009 - 19:48 Kenni Lund

Hi Martin

I'm sorry if you found my comment to be offensive. I fully understand your business decision that you will not include support for software RAID. When a project is primarily driven by commercial resources, the commercial interests will of course decide the way of the product. You have a very nice and impressive product, no doubt about that.

As you say, if you need software RAID, you can set it up yourself. I wouldn't have any issues doing this, and my first source would be to check out the official forums and mailing lists, to see if other people have been in the same situation and if they have solved it.

I did this and after reading some threads about software RAID in your forum, I found several replies from Proxmox staff members, which I would certainly call arrogant answers. I think that the thread I linked in my last comment is a fine example of this. If these answers from the Proxmox staff members had made it clear that this was a limitation due to commercial resources, instead of just saying "it makes no sense to use software RAID on Proxmox VE", "it will never work", "we are the developers, we know better than you", etc, then perhaps you would receive more positive feedback in public forums.

About making donations to implement new features like software RAID: When the general feedback from the Proxmox staff members is that it simply isn't possible and that it doesn't make sense to use software RAID with Proxmox PE, why should anyone then donate money to get it implemented, if it isn't possible?

If you accept donations for development of specific features, you should tell your users that, instead of saying that it makes no sense to implement the specific feature. pfSense, which also is a open source project driven by a company selling support for it, have a smart way of doing this: They gladly accept requests for new features bundled with a "bounty". The bounty is a small or large amount of money which the request maker will pay for the feature. If other users also would like to get the same feature, they can raise the bounty by donating the amount of money they would pay to get the feature. If the feature is highly wanted, the bounty will hopefully grow quite large. Instead of donating money to a company "because the company make cool software", people are actually donating money to solve their own problems, eg. they have a reason to donate money, other than just getting a nice warm feeling.

At the same time, you will also know for sure if a feature is highly wanted, instead of just assuming that since your competitors doesn't support software RAID, there isn't any market for it.

Link to the pfSense bounty-forum:
http://forum.pfsense.org/index.php?board=34.0

I'm looking forward to see the future developments of Proxmox VE :)

Best Regards
Kenni

Hi Kenny, Thanks for your

Monday, January 19, 2009 - 05:59 Martin Maurer (not verified)

Hi Kenny,

Thanks for your posting - I just reviewed the bounty system of pfSense and this is a really good system and I will think of something similar for Proxmox VE.

Br, Martin

Hi, is it possible to donate

Wednesday, January 21, 2009 - 18:04 Anonymous (not verified)

Hi,

is it possible to donate for skinnable themes for Proxmox? ;-)

BTW I still lag with Polish translation, damn me ;-)

Januszzz

yes.

Thursday, January 22, 2009 - 04:02 Martin Maurer (not verified)

yes.

Mmm it's good.But will it

Sunday, February 15, 2009 - 13:31 Anonymous (not verified)

Mmm it's good.But will it work with videocards?(IOMMU)

Post new comment

The content of this field is kept private and will not be shown publicly.
Type the characters you see in this picture. (verify using audio)
Type the characters you see in the picture above; if you can't read them, submit the form and a new image will be generated. Not case sensitive.