As part of my internship, I had a travel allowance for attending a conference or workshop so I chose to attend one of the largest Linux conferences: LinuxCon Europe, which took place in Düsseldorf, Germany.
I landed in Düsseldorf on Friday, along with my friends. We were a pretty large group of Romanian people, with 4 OPW interns (I believe we were the group with the largest number of females in the whole conference :) ). The weekend was reserved for exploring Düsseldorf and all it has to offer. I am not going to lie – german beer is as good as they say!
On Monday, the conference began. I attended the starting keynotes and thought the new project Linux Foundation announced, Dronecode, is really cool! Building a community around an open source platform for drones will bring them closer to the end user and better prepare them for specific tasks, as dictated by the community needs.
After the keynotes I went to the first talk: Enhancing Real-Time Capabilities with the PRU, by Ron Birkett from Texas Instruments. I am interested in embedded systems and currently working on a microkernel for such systems so this talk look like it could give some insight into how real-time characteristics can be achieved. This PRU (Programmable Realtime Unit) in question is a piece of hardware composed of RISC cores, memory and some other components. Its aim is to eliminate all non-determinism sources in a system. As a side note, it’s worth saying real-time characteristics are not really related to executing a task extremely fast, but by being able to guarantee a deadline for the execution of that task. One source of non-determinism are caches. The needed block of memory might or might not be in the cache when it is needed. Because of this, you can’t guarantee a certain task will get executed in a given amount of time. The PRU keeps its own memory logic for dealing with this. Similarly, interrupts are a source of non-determinism as you can’t predict when they come. That’s why the PRU doesn’t use interrupts, but polling. These are only two of the tricks a PRU does. As one of my colleagues puts it, this is a very orthodox way of obtaining real-time characteristics and a source of inspiration for our future endeavours into real-time operating systems.
The next presentation I wanted to attend was 12 Lessons Learnt in Boot Time Reduction but, apparently, many people wanted to see it so the room got full before I could get it. That wasn’t much of a problem as I wanted to go around the booths as well.
One of my objectives for LinuxCon was to find internship opportunities for next summer. Unfortunately, this didn’t go as I expected. Part of the booths were just for exposing stuff. From the people at company booths I learned they were interested in hiring but looking for full-time employees and knew little about internship programs in the respective companies. On the bright side, I did get some contacts and going to enquire about internships. I found this situation rather peculiar but, on the other hand, it is an event targeted towards professionals and not so much towards newbies.
Later in the day there was the Kernel Internship Report presentation, where OPW interns presented their work. I had exactly 5 minutes to speak about a three months internship so I had to carefully choose the most relevant parts. In the end, I think all interns did a great job presenting their results.
Both Thursday and Wednesday were full, with lots of talks to attend and people to talk to. It was nice getting to hear talking people whose names I’ve seen many times on the kernel mailing list or in tech articles. After LinuxCon, I feel inspired and motivated to continue working on open source software. And I plan on attending next year as well. Looking forward to the event in Dublin!