Thomas Gleixner

Jump to: navigation, search

Interview Questions: Thomas Gleixner

Backgrounding: Homepage:

Hi Thomas,

I thought I'd begin our interview with a small set of general questions.  I'll get more 
detailed and specific as we proceed. I apologize if some (if not all) of these questions seem 
trivial to start.

NOTE:  The final interview will be subject to your review/approval before it's published.

Q1:  Can you please introduce yourself?
TG: I studied theoretical physics and electronics, worked as hardware designer and software developer in the embedded (industrial automation) space before I caught the Linux virus and decided to run a small business focussed on Linux in industrial environments.
My first touch with Linux was back in the 0.x days and I still have a working floppy with 0.98 around. I followed the development out of curiosity and sent small patches from time to time. Later I got seriously involved into the MTD subsystem and contributed to the RTAI project. I was never fond of the dual kernel approach and contacted Prof. Douglas Niehaus at the Kansas University, who maintains the KUSP (former KURT) project. I owe him big time for teaching me deep insight into realtime Linux related problems and the area of high resolution timers. The KUSP based LibeRTOS project, which provided single kernel user space realtime features for Linux 2.4, was just in the transition state to 2.6, when the realtime patches from MontaVista were published to LKML, which led to the realtime preemption patch of Ingo Molnar. After looking at the code, I joined his efforts and contributed valuable pieces, which I had already moved from 2.4 to 2.6. Since then I'm working mostly fulltime on the realtime patches and the high resolution timer patches, which were integrated into mainline as of version 2.6.21.
While the first time of working on that was spent on my own private time minus the hours I was able to bill into related consulting contracts, I got funding for my work on the base of a cooperation of linutronix and Timesys Corp. since January 2006.
I live and work in Germany near the Switzerland border with my wife and my five kids.
Q2:  Can you give us a description of your history with Embedded Linux?  i.e. first platform 
you played with and why? current interests or projects?  (This can be as brief or as detailed 
as you see fit.)
TG: My first embedded Linux platform was a custom designed i386 (yes a real i386) board with 2MB RAM. It was just a pet project to figure out, whether it might work or not.
The first real project was an ARM embedded board equipped with NAND Flash. The NAND Flash support in the kernel was a prototype implementation and as I fixed it I inherited the maintainence of the orphaned driver.
I'm still maintaining the NAND flash driver and work on JFFS2 from time to time, but my main focus has shifted to high resolution timers and realtime. The realtime project is driven by enterprise interests, but I try to keep an eye on the embedded - non x86 - support since the very beginning.
Q3: What do you feel is the greatest benefit of developing embedded projects with Linux?
Similarly, what do you dislike the most?
TG: The greatest benefit is the huge fund of applications which can be recycled for embedded projects. It's also fun to see the same Linux, which drives super computers, running on a space constraint device.
The most annoying things of embedded Linux are:
  • the lack of understanding of semiconductor vendors, embedded distro vendors and hardware manufacturers.
  • their ignorance and refusal to work with the community
  • the horrible quality of their Board Support Packages