Microsoft Submits Linux Kernel Patches to Make Linux Run as Root Partition on Hyper-V

"Microsoft has submitted a series of patches to Linux kernel developers," reports ZDNet, "requesting that Linux run as the root partition on the Hyper-V, its hypervisor software for running Windows and non-Windows instances on hardware." Microsoft "wants to create a complete virtualization stack with Linux and Microsoft Hypervisor", according to Microsoft principle software engineer Wei Liu. Liu has proposed an RFC or request for comment that for now merely implements what are only the "absolutely necessary components to get things running... There will be a subsequent patch series to provide a device node (/dev/mshv) such that userspace programs can create and run virtual machines. We've also ported Cloud Hypervisor over and have been able to boot a Linux guest with Virtio devices since late July." Cloud Hypervisor is an experimental open-source hypervisor implementation from Intel written in the Rust programming language. It's a virtual-machine monitor that runs on top of KVM, the Kernel-based Virtual Machine hypervisor in the Linux kernel that's designed for cloud workloads...

Liu points out three more changes beyond amendments to the Hyper-V Top-Level Functional Specification. For example, Microsoft wants Linux to set up existing Hyper-V facilities differently. It also wants Linux kernel developers to change the kernel's behavior when accessing hardware memory in a way that affects driver access to the GPU and CPU that's being managed by an operating system memory manager. It's this issue that Microsoft engineers are least confident about and are asking for Linux developer support, according to Liu....

As Microsoft's executive VP of the cloud and enterprise group, Scott Guthrie, told ZDNet last year, Microsoft's shift to Linux and open source started over a decade ago when it open-sourced ASP.NET. "We recognized open source is something that every developer can benefit from. It's not nice, it's essential. It's not just code, it's community," explained Guthrie.

Sounds familiar

[quonset]

I seem to remember reading this recently [slashdot.org].

Re:

[Entrope]

Maybe Microsoft submitted a new set of patches. Slashdot Shirley wouldn't run a duplicate story.

Re:

[NFN_NLN]

Who is Slashdot Shirley?

And how to you know what she would post.

Re:

[Entrope]

I guess you've never seen the movie Airplane!? It has a repeated gag where someone says "surely [something]", and other characters respond "... and don't call me Shirley!"

Re:

[gwjgwj]

Who is General Failure and why is he reading my disk?

Re:

[ElizabethGreene]

Complaints can be directed to his CO, Kernel Fault.

Re:

[q_e_t]

That's a bit topsy-turvy. Is Private Network virtually in charge?

Re:

[ItsJustAPseudonym]

binspam at its finest.

Re: Extend...

[boudie2]

Extinguish! It's what they do.

Re:

[FaxeTheCat]

How?

Re: Extend...

[boudie2]

History lesson https://en.m.wikipedia.org/wik... [wikipedia.org]

Re:

[FaxeTheCat]

How can they extinguish anything by adding features to an open source, GPL licensed software?

Re: Extend...

[boudie2]

That's why microsoft has a 1500 person legal team. To figure things like that out.

Re:

[FaxeTheCat]

The software is under GPL. The size of the legal team does not matter. GPL has proven to be bulletproof.
So how was it they were going to extinguish?

Re: Extend...

[boudie2]

You don't think out of 1500 scumbag lawyers there's a few that think they could undermine the GPL? They're up to something, if I were devious enough to know what I'd be working in Redmond, not posting to slashdot.

Re:

[FaxeTheCat]

" They're up to something," So you have nothing. As I expected.

Re:

[boudie2]

Ah, but I have something you don't have, common sense.

Re:

[FaxeTheCat]

I stick to the facts: Nobody has yet successfully challenged GPL.

Re: Extend...

[boudie2]

Well then, I guess there's nothing to worry about. Glad we settled that.

Re:

[business_kid]

M$ never helped any OSS project crack any of it's many, many proprietary formats. At the bottom of this is M$ desire to profit on the back of hard graft by unpaid (in many cases) devs and their desire to cherry pick OSS for their ends?

Let them go OSS also, put up their source repository including things like DirectX, or give them the middle finger. My â0.02Â

Re:

[business_kid]

Er, my two cents worth.

Exclude?

[SodaStream]

If only there was a way to not include GPL'd modules in a GPL'd kernel...

Re:

[ArchieBunker]

If that were even remotely true they would be copying BSD code.

Noooo!

[fustakrakich]

Let them make modules!

principles?

[algaeman]

Shouldn't a Microsoft principle systems engineer restrict their patches to the repository code of conduct?

Re:

[OneHundredAndTen]

Shouldn't a Microsoft principle systems engineer restrict their patches to the repository code of conduct?

'Principle' is a word that does not appear in MS's vocabulary. Surely you mean 'principal'?

Are these hooks to run Hyper-V, or the real thing?

[ctilsie242]

Are these just hooks into Hyper-V to allow it to run, or is MS adding Hyper-V as another hypervisor choice, like Xen, KVM, or others, for use on a native level? If Hyper-V is used in this fashion, it can be useful, especially to isolate what is running from the hardware, and to make it easy to add VMs and move them between Linux and Windows environments seamlessly.

In any case, I hope this isn't a bad thing overall. MS wants stuff to run on Azure, so it helps them in this department.

Comment removed

[account_deleted]

Comment removed based on user account deletion

what about hyper-v adding ceph? not over iscis

[Joe_Dragon]

what about hyper-v adding ceph? not over iscis

Could this be a GPL circumvention system?

[tap]

The classic way way to the write a Linux device driver, and not have it be GPL, is to make a userspace driver. A GPL module is written that provides hooks that allow access to the kernel. Usually it's some stupid thing that allows arbitrary access to kernel memory or binding to any kernel symbol or something that totally destroys the concept of kernel mode separation from user mode. Then the real device driver is written to use the hooks. The argument is made that the two parts, the GPL part that is ju

Re: Could this be a GPL circumvention system?

[aRTeeNLCH]

I think I understand what you're saying technically and the worry you express. On the other hand, would that construction not make it loads easier to reverse engineer drivers with Linux at the lowest layer, perhaps with slight modifications for more logging for the signalling of the higher level driver...?