One of the main problems of the current Hurd implementation is very poor performance. While we have a bunch of ideas what could cause the performance problems, these are mostly just guesses. Better understanding what really causes bad performance is necessary to improve the situation.

For that, we need tools for performance measurements. While all kinds of more or less specific profiling tools could be conceived, the most promising and generic approach seems to be a framework for logging certain events in the running system (both in the microkernel and in the Hurd servers). This would allow checking how much time is spent in certain modules, how often certain situations occur, how things interact, etc. It could also prove helpful in debugging some issues that are otherwise hard to find because of complex interactions.

The most popular kernel instrumentation framework is Sun's dtrace, originally written for Solaris, but also adopted by some other systems. However, the GPL-incompatible license means it can't be used in Linux, and thus Linux developers created their own frameworks instead: first SystemTap, and now LTTng.

In 2008, Andrei Barbu did initial work on kernel probes for the Hurd. However, not all of his patches got merged, because some turned out not to be fully functional. Also, he didn't get around to work on userspace probes, nor on a nice frontend for writing test scripts employing the probes.

The goal of this project is to make the instrumentation framework more usable and complete, and to better integrate it in the Hurd. For that, the student will have to work on some real profiling and/or debugging tasks, and fix any shortcomings he encounters in the framework.

This is a pretty involved task. Previous experience with low-level programming is a must; and it also requires a good grasp on interactions in complex systems.

To work on this project, the student will have to get familiar with GNU Mach. (The microkernel employed by the Hurd.) Some understanding of other aspects of the Hurd will also be required, depending on the exact nature of the profiling/debugging performed.

Exercise: Use the existing probes to perform some simple measurement.