Microsoft Developer Made the Most Changes To Linux 3.0 Code

sfcrazy sends this quote from the H: "The 343 changes made by Microsoft developer K. Y. Srinivasan put him at the top of a list, created by LWN.net, of developers who made the most changes in the current development cycle for Linux 3.0. Along with a number of other 'change sets,' Microsoft provided a total of 361 changes, putting it in seventh place on the list of companies and groups that contributed code to the Linux kernel. By comparison, independent developers provided 1,085 change sets to Linux 3.0, while Red Hat provided 1,000 and Intel 839."

Comments

The number itself is entertaining but ...

[shutdown -p now]

... it really is useless trivia. What's more important is what the contributions are, specifically. Per TFA:

This work by Microsoft was to clean up the “Microsoft Hyper-V (HV) driver” so that the Microsoft driver would be included in the mainline Linux kernel. Microsoft originally submitted this set of code changes back in July 2009, but there were a lot of problems with it, and the Linux kernel developers insisted that it be fixed. The Linux community had a long list of issues with Microsoft’s code, but the good news is that Microsoft worked to improve the quality of its code so that it could be accepted into the Linux kernel. Other developers helped Microsoft get their code up to par, too. ( Steve Friedl has some comments about its early technical issues.

and why:

Getting code into the mainline Linux kernel release, instead of just existing as a separate patch, is vitally important for an organization if they want people to use their software (if it needs to be part of the Linux kernel, as this did). A counter-example is that the Xen developers let KVM zoom ahead of them, because the Xen developers failed to set a high priority on getting full support for Xen into the mainline Linux kernel. As Thorsten Leemhuis at The H says, “There are many indications that the Xen developers should have put more effort into merging Xen support into the official kernel earlier. After all, while Xen was giving developers and distribution users a hard time with the old kernel, a new virtualisation star was rising on the open source horizon: KVM (Kernel-based Virtual Machine) In the beginning, KVM could not touch the functional scope and speed of Xen. But soon, open source developers, Linux distributors, and companies such as AMD, Intel and IBM became interested in KVM and contributed a number of improvements, so that KVM quickly caught up and even moved past Xen in some respects.” Xen may do well in the future, but this is still a cautionary tale.

Re:The number itself is entertaining but ...

[Sc4Freak]

So? A contribution is a contribution, even if it is for selfish reasons.

Re:The number itself is entertaining but ...

[shutdown -p now]

I did not imply otherwise. My point is that the contribution and its nature are of more importance than the associated random statistical fluke.

Re:The number itself is entertaining but ...

[shutdown -p now]

You want to expose host's hardware to the guest with as few layers in between as possible. Traditional emulation is rather slow, so instead you set up a fast channel that exposes exactly what is needed in a most efficient way, and write drivers for the guest which use that to work with hardware.

I believe this is also true for scheduling - if host and guest cooperate (which necessarily requires special code running in guest's kernel), they can do much better at it.

Re:The number itself is entertaining but ...

[wrook]

Even more importantly, contributing for selfish reasons creates a win-win situation. Contributing in a way that is detrimental to you, is detrimental for the community. It is important for people and organisations to realise that we want them to succeed in their enterprises.

I think a lot of people misunderstand the driving forces behind free and open source software. They see it as some kind of charity where the group "donating" code is losing out. Instead, groups should understand how they are going to benefit from contributing to a free software project before they do so. Benefit can come in the form of money, it can come in the form of eyeballs (attracting attention to an under serviced area), or it can simply come from the pleasure of contributing. These are all benefits.

Free and open source software allows more than one group to benefit from contributing to a project. You can't control how much benefit another group can get from a project, but the more you do to tie your success to the success of the project, the more you benefit you get from other people's contributions. Ideally, we want companies like MS to make money from the success of free software. The more they do so, the more they will understand the opportunities they are missing. The more they rely on our success, the more everyone benefits.

Re:The number itself is entertaining but ...

[jcoy42]

Since it was a Microsoft employee, I'm going to assume that it was really just one contribution and 360 patches to make it work.

Expect more patches soon.

Re:The number itself is entertaining but ...

[staalmannen]

The whole point (which Linus often stresses) is that open source is all about "scratching your own itch". This means that all contributions are self-serving. I am actually quite surprised that Apple has not tried to push patches to Linux for kernel GCD support yet. That would also have been a self-serving improvement since they want to push the standards of C in that direction.

Re:The number itself is entertaining but ...

[TheRaven64]

Why would Apple want to do that? Libdispatch runs on *BSD, Linux and Solaris, but on Linux and Solaris uses libkqueue to provide an emulation of the kqueue APIs and it only uses the kernel scheduling on Darwin and FreeBSD. If you write code using libdispatch, it works everywhere except Windows, but people using Linux get an inferior experience to people using FreeBSD or Darwin. That sounds pretty much idea from Apple's perspective.

Oh, and someone did implement kqueue on Linux a couple of years ago. It was rejected because the mess of timerfd, signalfd, and epoll() was considered better by the NIH mentality of the Linux kernel team. As someone who has used both, I'm always glad when I don't have to make my code work on Linux.

Re:The number itself is entertaining but ...

[TheRaven64]

Speaking as the author of the only book about the internal working of Xen:

It's very important for scheduling. The guest OS typically lets processes run for 10ms windows (or until they hit some blocking code). In a virtualised environment, 10ms does not necessarily translate to 10ms of CPU time. A guest OS should run its processes for 10ms of time that the guest is scheduled, not for 10ms of elapsed time. If the guest is not aware of when it is scheduled and when it is idle, then it can not schedule its processes effectively. In some cases, you have latency-sensitive processes in the guest. The hypervisor and the guest can then cooperate so that the guest is not preempted while these processes are scheduled. Cooperation between host and guest schedulers is a very active research topic at the moment, because it can have a huge impact on overall throughput.

Re:The number itself is entertaining but ...

[Entrope]

This. Some of the people on LKML pointed out that the guy's floods of ~180 patches at a time grossly violated the patch submission standards laid out in Documentation/SubmittingPatches ("Do not more than 15 patches at once to the vger mailing lists!!!"). I know it annoyed me, and it seemed like a huge amount of code churn for a driver in staging. I didn't realize until I saw this story what the driver was or who the author was.

Re:Community Myth

[Anonymous Coward]

I wish people would get over this myth that "could care less" means that you couldn't care less.

Not Microsoft...

[Anonymous Coward]

He has only been part of Microsoft since february 2011. Until then, he was part of Novel.

This... is stupid.

[Anonymous Coward]

Perhaps they couldn't care less (note the correct usage), what does that matter? So long as they contribute valid code, in compliance with licensing, that addresses a need, I don't care *who* contributes. Hell: Hans, from his jail cell, can contribute, for all of me. If it makes Linux better, and it's not some patent landmine, IJustDon'tGiveADamn.

As for user-driven innovation, yes, it is. For two reasons:

- Solo users still do contribute. Check the numbers.
- Solo users who manage to work for large companies does not mean they still don't adhere to the spirit.

So neener.

Re:changes != LoC

[scromp]

Describing LoC as a "not perfect" metric is an astonishing understatement.

Re:Community Myth

[rtb61]

Contributions to Linux take many forms. Code-centric people only view contributions to the Linux kernel as contributions to Linux. Far more aware people, take contributions to Linux being amongst the following,
contributions to GNU (something you obviously need to look up),
contributions to Linux compatible hardware drivers,
contributions to Linux documentation,
contributions to Linux based graphical users interfaces,

contributions to Linux compatible applications,
contributions to graphics design including icons, appearance, fonts, screens savers, layouts,
contributions to marketing and promotion,
contributions to Linux protecting patents,
contributions to service and support,
contributions to the Linuc user community,
and of course contributions to Linux based distribution without which Linux would not exist as an operating system rather than just a kernel.

Seriously only a real asshat would take all those contributions and treat them as nothing either that or a microtroll. It amazes me that after all these years how people still fail to understand how a community developed product like Linux comes into being, how all contributions small and large are highly regarded (the value being in the sharing) and how contributions of individuals are valued (even those employed by M$, M$ did you create code, those people employed by M$ did).

From your selfish self centred viewpoint, it appears that I must apologise for using Linux whilst not being a good enough coder to contribute to the kernel. So "I am sorry", my coding sucks and my others contributions to Linux are not good enough to appease you. Of course to the rest of the Linux community I say thank you for all the contributions made no matter how great or small.

Re:Yay

[BitZtream]

So you're saying the people who review kernel patches are so shitty they couldn't spot any of the things you're referring too?

You do realize you're insulting your own team more than the other team right, you're just too stupid to realize ... oh ... never mind.

Re:Community Myth

[causality]

Linux is still "a community effort of users putting their minds together". There may have been a shift from the community made up mostly of individuals to corporations but it is still, a community. Does it matter if a contributor is an individual, an individual contributing on behalf of a company or company contributing as long as the code is of good quality, is offered in ways that agrees with the norms of the community and does not violate any license used by the community? Microsoft is contributing code because they are either using linux or they have people they are supporting who are using it hence they are part of the community effort. It doesnt matter how little any individual or company cares about the GPL, all it matters is that they conduct themselves in a way that does not violate it.

I will tell you what really makes me personally feel like I am participating in a community. For most Open Source software I have used, if I have a question or a suggestion or simply some feedback, I can usually communicate directly with the maintainer or lead developer of the project. They are accessible. They are fellow human beings, not corporate conglomerates. There are no layers of sales reps or receptionists or PR personnel. Sometimes I send an e-mail just to say "thank you" for the simple reason that they owe me absolutely nothing, yet I benefit from the work they have chosen to make freely available.

It would be like calling up Microsoft and speaking directly to Ballmer about Windows. No regular Microsoft customer is ever going to do that. That's the difference between a community and a conglomerate. That, and with most Linux distributions users help each other as much as (if not more than) organizations provide formal support.

Re:changes != LoC

[maugle]

Describing LoC as a "not perfect" metric is an astonishing understatement.

I
disagree,
using
LoC
as
a
metric
has
made
my
measured
productivity
skyrocket!

Re:A Purge Needed

[93 Escort Wagon]

The code from ANYONE at Microsoft is venom ... and must be dead ... and must be deleated.

Neither Microsoft nor Apple can be trusted! They are both EVIL. KILL their CODE! ... Let their
bodies BURN.

A real good day will be when the bodies of Microsoft and Apple employees are burning in the streets.

-- //

Just a suggestion... cut back on the caffeine.

Re:Community Myth

[fnj]

No, it's called illiteracy.

Re:Yes let's just get down and dirty in the code

[shutdown -p now]

FOR THOSE USING WINDOWS, they can _also_ run Linux.

If they ran Linux they could host whatever they want. THAT'S why Microsoft did this. VMware Server is free, and it runs on Linux or Windows, and it hosts EVERYTHING.

That's cool. Hyper-V Server is also free, and does not require any OS to run (VMware also has a similar product - ESX). You don't need to run Windows.

Though I find it interesting that you object to running Linux in one proprietary piece of software (Hyper-V), but not the other (VMware). I'd understand your perspective it it was, at least, a matter of FOSS purity - but then you should, at least, argue for KVM or Xen.

This does not enhance Linux nor the experience of anyone who uses it.

It does enhance everyone's Linux experience if there are more Linux machines out there, don't you agree? If some previously Microsoft-only shop can now run e.g. LAMP instances on their Hyper-V servers, that's one more customer software and other companies would consider.

Re:Yes let's just get down and dirty in the code

[Ost99]

1) Hyper-V has worked fine with Linux hosts for a long time, this is just a code cleanup and bug fix for existing Hyper-V specific drivers in the kernel.
2) VMWare also needs guest drivers to get Linux to run as it should. These are not part of the Linux kernel. VMware releases their drivers separably, but you still need them to make full use of the server.

Microsoft decided it would be easier for their users if they got their drivers included in the kernel. More work for Microsoft and the kernel maintainers, less work for the users.

Re:Community Myth

[Sun]

And then I, as lead (and often only) developer for several FOSS projects, get an email with a question, suggestion or bug report to my personal email. When I reply with "please use the mailing list", people like you, who, to them, "community" means that the lead developer needs to answer their questions directly, complain, get upset, and sometimes get downright rude.

As a lead developer, I want a community to form. This means that I want to give all people in the community a chance to answer your question, not only myself personally.

Shachar

Re:Community Myth

[X0563511]

Saying one thing and meaning it's exact opposite is not an evolution of an "idiom" - it's just idiocy.

Unless you're telling me that in 300 years "4" might actually mean "27" or something along those lines?

Re:Yes let's just get down and dirty in the code

[Bert64]

Interoperability for MS is a short term goal..

When IE was new and competing with netscape, they worked to make it interoperable with netscape...
Once netscape was gone, that flew out of the window and they tried to maintain lock-in.
Now that firefox and chrome have become popular, ie is now trying to interoperate again.

The same can be said of msoffice, when they faced serious competition they supported opening wordperfect files and had a relatively open rtf spec, once the competition was overcome they turn the screws of lockin again.

I predict the same will happen with hyper-v if they get the chance... If vmware, xen and kvm fall by the wayside then it wont be long before hyper-v only runs windows, the linux support will stagnate and new versions of hyper-v will come out which aren't compatible with it.

Re:Yes let's just get down and dirty in the code

[Bert64]

Interestingly, windows boots a lot faster inside of a vm running on linux than it does on the hative hardware (seriously, give it a try)...
If going the other way round, linux runs somewhat slower inside of a vm running on windows... The performance penalty when running in a vm on linux is much smaller.