Rusty Russell, Interview

20. 11. 2009 | Robert Krátký | Rozhovory | 9360×

1) You're known to have worked on many very different things. Networking, modules, virtualization, and whatnot. Is that a case of changing interests? How do you decide?

How do I choose, that's the hardest question. There's so much that you could be doing, so many great projects out there that there are two options. One is to spread yourself so thinly you don't make any difference to a huge number of projects. The other one is to keep going back to the projects you're comfortable with. And both are bad – you need to force yourself sometimes to go out of your comfort zone. And great things can happen. Like with Wesnoth. I'm not a game hacker, I don't use C++ anymore, but I said, you know, this could be fun. That was a really good experience. I like hacking on things. The other problem is always finding the time. I tried to keep down the number of unofficial projects I'm on to about three or four at a time.

2) What's next on your list?

CCAN is my current thing, and libantithread these are the two projects I'm working on. CCAN is the comprehensive C archive network, like C-PAN, only for C, and libantihread is basically a portable way of doing shared memory.

3) Is there an area of the kernel you want stay away from?

Yes, there is, actually. Filesystems. I don't touch filesystems. Two reasons. One is that in any other part of the kernel if you mess up it might crash, with filesystems you can create lasting damage if you corrupt someone's filesystem. That's a part of it. The second is that there are some people in filesystems that are particularly difficult to work with. And it's easy to just say 'well, I've got plenty of other things I can do' so I'm not going to write RustyFS any time soon.

4) What was the motivation for hacking on lguest? There were other virtualization solutions available…

There were two reasons to start hacking on it. The first one was that we were debating about paravirt_ops – this interface of a guest to talk to different hypervisors. And in order to really understand the needs of the different hypervisors we had then, we had VMs and VMI, and we had native hardware. And to really understand what the generic needs were, how to design the API, I thought that I had to write another one. As simple as possible one, to make sure we wouldn't make it too complicated. That was one provision.

The other is that I felt that the way Xen did virtualization was not the best way of doing it. So, lguest used a modular model where the host is just a normal kernel. There's no hypervisor kernel. And the guests are just normal processes. Fortunately, I never had to turn lguest into a serious hypervisor because KVM came along, using exactly the same model. Independently. And that meant that I didn't have to write millions lines of code. I just let Avi Kivity go off and do all that.

5) I was wondering if you ever considered working on KVM. Moving to Israel…

I have actually done some KVM patches, but not for a while. I like working on KVM. I think it's a great project. But they have plenty of hackers working on the core KVM.

5) You were involved in the negotiations about FTA with the US and the result was not exactly encouraging. Has there been any new development in that area, any court cases that would interpret the law?

Interestingly, we haven't seen any significant development on the things we were really worried about. That DMCA style. That said… two years ago my New Year's resolution was not to do any more intellectual property work, so I have been staying out of it. But it has been very quiet. But sometimes these things take years before you get a court case which will decide what the law actually says.

6) You said in an interview at that time that a possible way to make people care about these laws is to get them to use free software so that the laws would directly affect them. Do you see any progress in that regard? Have you been able to raise awareness about these issues?

I think people are more aware of IP law. And one of the reasons is simply because we've seen an uptake in important noises about intellectual property violations. I think it has made more people aware of these kind of issues. Free software is continuing to grow, pretty much independently because, basically, people are – while officially not distributing a lot of these things that might be questionable, they're widely available – and people are still using them. And because it's in that grey area it has had a suppressive effect, but it hasn't been clearly outlawed.

Should the law come down on the side of these things that are completely outrageous like you can't use them any more then we might see, at this stage, some real changes. But there are lot of fascinating things going on in the IP space. Like the open media people who are seeking to catalogue and work around the law, and wait for expiry of some of these patents. There's a lot of interesting stuff around the media side which is where the stuff really comes to head that's going on right now so people are definitely aware of the stuff but it's developing in a lot of different directions.

7) Australian Broadband minister Stephen Conroy proposed filtering BitTorrent and other internet traffic. Do you think these plans can actually get some traction in Australia?

I met Stephen Conroy. He's an idiot. But that doesn't necessarily his ideas won't go anywhere. I think we have a big job to do to make sure that people understand the benefits of technology, rather than fearing it and having these kind of allergic reactions. People often don't understand the peer-to-peer nature of internet itself. The fundamental difficulty with the idea that you should ban some aspects of it. I haven't been keeping a close eye on the situation. But it's fairly obvious to most technical people that he can do an awful lot of damage and throw us back to the stone age, and still have no real effect on any of the issues that he claims to be concerned about.

8) You mentioned some of your talks at conferences are actually recruiting drives aimed at exciting interest in newcomers. What would be your advice to potential kernel hackers looking for ways to contribute?

I don't think I'd start on the kernel. There's a lot of hackers who want to be involved with the kernel. That's great and there is some really good code in the kernel and you can learn some really good habits, but I actually wish more kernel coders would go out and work on other projects. And many of them do. Because it's much easier to have a really big impact on almost any other project than the kernel. In the kernel, there's a lot of people doing a lot of code, the amount of code churn is incredible, but on these other projects you can basically have a radical effect on the project for much less effort than you get in the kernel.

So, I would not just go down for the kernel. I mean, being a kernel hacker is great, but I would also hack on something else to save time. I think the skills will carry across and sometimes, the kernel can just be frustrating. Having another project to work on instead and flip to that can be a really good way of not getting completely frustrated. I think that would be my advice.

As far as kernel hacking itself goes there are just so many areas of the code that you could pretty much pick a random file in the kernel, read it, and you can find something that is worth tidying up. And you can repeat that forever. And as you read more and more code, you will find more and more things to fix. And your fixes will probably get bigger and more impressive and next thing you know you'll be writing major sections of the kernel. But it comes down to the willingness to read code, as well as to write code.

9) I was wondering why you gave up the patch-monkey.

Yeah, well, I gave up the patch monkey mainly because I proved it worked and Adrian Bunk offered to take it over from me. I didn't think that it was something what needed a great deal of skill any more, once we've sorted out the details of how it would work. And Adrian Bunk is a much more active trivial patch monkey than I ever was. So handing it over to him was a really great thing to do.

       

Komentáře