Tracing / BPF

The Linux Plumbers 2017 Tracing track is focusing on the various tracing infrastructures in Linux today and how various people and companies use them.

Topic suggestions

  • Building your own tools around existing kernel infrastructure
  • Wishlist items for kernel tracing
  • Issues that have been discovered using tracing
  • New features in bpf/perf/ftrace
  • Visualization frameworks
  • Distributed tracing
  • Always-on analytics and monitoring (including TCP/networking/storage)
  • Tracing at large scale
  • Hardware Tracing (including latest work in perf/CoreSight and trace decoders)
  • Integration of trace buffering and aggregation tools

Possible interested attendees

  • Arnaldo Carvalho de Melo (declined)
  • Ingo Molnar (declined)
  • Steve Rostedt (accepted)
  • Brendan Gregg (accepted)
  • Martin Lau (accepted)
  • Daniel Borkmann (accepted)
  • Brenden Blanco
  • Mathieu Desnoyers (accepted)
  • David Ahern (tentative)
  • Sasha Goldshtein (declined)
  • Suchakra Sharma (attending)
  • Allan McAleavy
  • Yonghong Song (accepted)
  • Wang Nan (declined)
  • He Kuang
  • Ming Lei (declined)
  • Daniel Wagner
  • Josh Poimboeuf (tentative)
  • Mark Drayton
  • Teng Qin (accepted)
  • Marek Vavrusa
  • Andi Kleen (declined)
  • Mathieu Poirier (accepted - unless clashes with “power management and energy awareness”)
  • Jim Mauro
  • Abderrahmane [Abder] Benbachir (attending)
  • Namhyung Kim (attending)
  • Masami Hiramatsu (attending)
  • Geneviève Bastien (attending)
  • Daniel Bristot de Oliveira
  • Staffan Tjernstrom
  • Manoj Rao

Talk proposals

Please feel free to add your name and proposal here.

  • Next BPF tools, eg, chaingraphs (Brendan)
  • CTF for maps (Brenden)
  • Faster uprobes/kerninst (Brenden/Yonghong)
  • Hardware Tracing with PT (VM Analysis and kernel-assisted trace re-construction for JITed code) (Suchakra)
  • Trace Visualization: eBPF to CTF (Common Trace Format), perf to CTF and trace analysis with Trace Compass (Suchakra/Geneviève)
  • LLVM for BPF (Yonghong)
  • Hypertracing: Tracing through virtualization layers (Abder)
    • Demo of a mechanism to trace VMs on any nested level using hypercalls and ftrace infrastructure
    • Presentation on performance overhead of hypertracing and visualizations
  • user + kernel tracing w/ uftrace (Namhyung)
  • addr→sym resolution by kernel in stack traces (Alexei)
  • Challenges related to doing real world debugging for non-kernel engineers (Josef)
  • Debugging BPF programs on production systems remotely, even when you are a kernel engineer (JohnF)

Tentative Schedule

No schedule as of yet.

Note: The final schedule will be posted on the linuxplumberscong.org website on the Schedule page. Presentation slides will available on the Plumbers page by following the links to the abstracts.

Discussion notes

Notes from the sessions will be recorded using Etherpads (links will be provided). It is very important that each session has good notes. All too often, if notes were not written down, the session might as well not have happened.

Contact

Proposal added by Josef Bacik jbacik@fb.com and I will run this microconf along with Alexei Starovoitov ast@fb.com and Brendan Gregg bgregg@netflix.com.

 
2017/tracing.txt · Last modified: 2017/09/10 06:11 by 54.240.198.33
 
Except where otherwise noted, content on this wiki is licensed under the following license:CC Attribution-Noncommercial-Share Alike 3.0 Unported
Recent changes RSS feed Donate Powered by PHP Valid XHTML 1.0 Valid CSS Driven by DokuWiki