Profile cover photo
Profile photo
Jesper Dangaard Brouer
648 followers -
Linux Kernel Developer in the network area, with a passion for scalability and performance
Linux Kernel Developer in the network area, with a passion for scalability and performance

648 followers
About
Jesper's posts

Post has attachment
Thomas Gleixner presents om Real-Time Linux
Photo

Post has shared content
+Jesper Dangaard Brouer​ on Linux Kernel: eXpress Data Path (XDP) for DDoS protection


Photo

Post has attachment
Sharing a trick on how you can read kernel variables runtime with perf probe. Useful trick for a life as a kernel developer...

Post has attachment
Reading live runtime kernel variables
Tips and tricks from a kernel developer Q: How to read variables a given kernel function get invoked with on a given workload? I recently had a workload ( page allocator micro-benchmark ) that showed a function that was semi-hot (policy_zonelist) in my perf...

Post has shared content
Thanks goes out to +Dave Taht for working so hard on saving the Internet for everybody!
Make-wifi-fast talk about speeding up #wifi by eliminating #bufferbloat is now summarized and available here:

https://lwn.net/SubscriberLink/705884/1bdb9c4aa048b0d5/

(hopefully jon won't mind me propagating the subscriber link. His write-up was much better than the talk itself, IMHO)

Post has shared content
Florian gives a great overview about nfttables and its current state of development: http://developers.redhat.com/blog/2016/10/28/what-comes-after-iptables-its-successor-of-course-nftables/ "[…] Nftables is a new packet classification framework that aims to replace the existing iptables, ip6tables, arptables and ebtables facilities. It aims to resolve a lot of limitations that exist in the venerable ip/ip6tables tools. The most notable capabilities that nftables offers over the old iptables are: […] There a few iptables matches and targets that still lack a native nft replacement. Most notably those are: […]"
Photo

Post has attachment
Video online of Network Performance Workshop, NetDev 1.2, Tokyo
https://www.youtube.com/watch?v=vsjxgOpv1n8

Post has shared content
My keynote at netdev 1.2 in Tokyo.

Post has shared content
LWN is cool
Would you like to write full-time for one of the smartest and most interesting audiences out there?  LWN is currently looking for a new editor/writer, and we would like to hear from you.  Please see the job description below and drop us a note if you'd like to be a part of LWN.

Post has shared content
On respect versus merit.

I've recently seen a few typical Open Source Collisions happen and being involved in it partially as well. As I'm a fairly pragmatic person I tend to shrug it off and focus on the work at hand, but there are always a few people around that can't understand that respect and merit are orthogonal.

Any new person who starts out doing Open Source should be met with the utmost respect. They have absolutely no merit to begin with, and others should encourage them and show the beginner mistakes in their work. The new people should treat experienced people as you would treat any good teacher: without any significant more respect than anyone else(!). Poke them, ask them, prod them for answers and explanations, but certainly do not go easy on your mentors - they are there not to sit on a throne and rule, but to guide everyone to do better.

Any experienced person should treat new contributors with respect, but treat their code for what it's worth. No need to get salty if it's bad. Just say "It's terrible" and leave it at that.

But that's where things go wrong. If you, as an experienced developer, fail to explain why a submission is wrong or misinformed, you're not giving someone the education or knowledge that you have, and you're guilty of depriving them of a chance to learn.

Now what I've noticed is that there seem to be many capable, experienced OSS contributors who lavish in merit and destroy their own respect, by ignoring this advice. These aren't business critical projects, but nonetheless it matters to a lot of people, so things get heated pretty quickly.

I've now seen two out of control spirals of disrespect end in people leaving. For no good reason than that the involved senior people entirely confuse merit and respect, and think that they are interchangeable.

It starts with reviews ending up saltier and shorter, especially for reviews from newer contributors. It ends with someone giving up, and sadly it's usually the newer people that give up, even though the potential that they will contribute more and better code in the future is often far more likely than that the merit-soaker is coming back to do actual coding.

So, takeaways for those that recognize the situation? If you don't code anymore in a project, don't become the grumpy reviewer. Let others take over. Stay constructive and technical, and teach instead of criticize. Never attack a person, ever.

Yes, there are indeed plenty of public OSS figures out there that violate these guidelines, and it's inexcusable, really. And totally not needed, either. I've most certainly have been on the wrong side as well, for sure, in the past. I hope I've made up for it, though, and intend to improve where I can.
Wait while more posts are being loaded