This writeup concludes our series of in-depth postings about the LPC 2014 microconferences. Thank you for reading so far! Looking forward to an exciting and productive conference in Düsseldorf.
Tracing is heavily used throughout the Linux ecosystem and provides an essential method for extracting information about the underlying code that is running on a system. Although tracing is simple in concept, effective usage and implementation can be quite involved (see this LWN.net article) because of the variety of environments that tracing must handle.
The ultimate goal of tracing is to get as much information as possible with as little overhead as possible. These two requirements are clearly in conflict and different approaches to balance this conflict have been taken by ftrace, perf and LTTng. One of the topics to be discussed at this microconference will be how to better integrate these three differing tracing implementations, for example, using Common Trace Format (CTF) to allow each to read the others’ tracing output. There are also other tools, such as systemtap, which complement tracing, providing more control features, and the kernel may benefit by working more inclusively with them.
Another topic for this microconference will be how tracing could make use of JITted code, given that, surprising though it might seem, the Linux kernel now contains a JIT for the Berkeley Packet Filter (BPF, as explained in this LWN.net article). At the same time, kernel-bloat considerations will generate a discussion of how to reduce the Linux kernel’s current per-tracepoint memory overhead.
This microconference is a follow-on to a Tracing Summit. The Summit will focus more on traditional style presentations, while this microconference will focus on finding and implementing solutions to Linux’s current tracing problems.
(Thanks to Paul McKenney for help with this post)