July 9, 2018

The FreeBSD Foundation kindly sponsored my trip to BSDCan 2018 in Ottawa, Canada that took place June 6th through the 9th. I was able to attend both the FreeBSD Developer Summit as well as the conference proper, and I met many people along the way. I am incredibly grateful for having had this opportunity.

Before even arriving in Ottawa I met Sean Bruno at LGA, my final stop before arriving in Ottawa. I had also run into Romain Tartière on my flight, but I did not learn who he was until a later time. I met Justin Hibbits, Emmanuel Vadot, Baptiste Daroussin, Jonathan Looney, Li-Wen Hsu, and Larus Maxwell of Netapp early on the first day of the dev summit, which opened with a core@ update and Release Engineering presentation. I found both of these talks to be interesting, as I recently became a part of the Project this last year. The insight into situations that core.9 dealt with, and unfinished business that core.10 would be dealing with provided me with more background for the upcoming voting session. Release Engineering’s presentation was rather good, but I found their clarification of support lifecycles to be the most helpful. I spent the afternoon sessions hacking on a lualoader patch to match an upcoming feature for the forthloader, and working on some concepts for making bsdgrep a little more efficient. 

Time at the hacker lounge was spent looking at replacing regex(3) with Onigmo, and meeting the other conference attendees. I initially met Ed Maste and Gordon Tetlow, while sitting at a table just inside HNN 013 and we discussed bsdgrep and lualoader for a little bit. I later met Rod Grimes and we discussed the intricacies of merging, a few procedures that we’re currently doing wrong and want to correct, along with some bootloader issues he’s been experiencing lately.

Day two of the dev summit had a secteam update and a different version of HaveNeedWant for 12 led by John Baldwin- consisting of discussing mostly what we’re trying to land in the next two months for 12.0, followed by some discussion of things we should start targeting for 13.0. It was definitely an experience- topics being listed left and right and understanding that we’ve certainly still got plenty on our plate to accomplish by 13,  both reminders that the work is surely never finished. The highlight of the second day’s afternoon was the boot code working group. We discussed transition plans for lualoader, which I have spent plenty of time getting into shape. We also discussed a future direction and recapped the different limitations that we have in different stages. This was also the point where I met Warner Losh, whom I’ve conversed with and collaborated with on different points of lualoader and other UEFI-related shenanigans. At the hackathon on the second night, I met Kristof Provost initially. I later met Bryan Drewery, Ryan Libby, and Mark Johnston and observed as they were diagnosing a VFS assertion that Justin Hibbits was encountering on his POWER9/Talos box. The conference proper was also a great experience. Overall, I went to the following talks:

  • The Tragedy of systemd by Benno Rice
  • Profiling the FreeBSD kernel boot by Colin Percival
  • A Public-Key Trust Infrastructure for FreeBSD by Eric McCorkle
  • Improving netdump Hardware Support and Performance with iflib by Sam Gwydir
  • Adding Verification to FreeBSD Loader by Simon Gerraty
  • FreeBSD ARM32/ARM64: Porting to a New Board by Emmanuel Vadot
  • Introducing FreeBSD in a New Environment: the Good, the Bad, and the Ugly by Baptiste Daroussin,
  • Making Your Own Console Server Using OpenBSD by Kurt Mosiejczuk ​
  • The Evolution of FreeBSD Governance by Kirk McKusick

Many of the talks I attended gave a glimpse of future work- especially in the case of public-key trust infrastructure, and Simon’s talk non veriexec/kernel verification. Colin’s talk on profiling the kernel boot certainly made me more aware that the things we do have effects on the boot, and that we really should be putting more effort into quantifying this effect and making sure we’re operating as efficiently as we can. The hallway track was perhaps the least-expected part of the conference that I enjoyed. Unfortunately, I missed a couple sessions due to I conversing with people out in the lobby. During the hallway track, I met a group working on TrueOS including Kris Moore and Nick Wolff. I confirmed that they should be OK to start looking at lualoader for TrueOS after receiving a complaint about loader.conf overlays setting boot_mute and friends in the process. I also met John Baldwin, Brad Davis, Ravi Pokala, and Alexander Sideropoulos of Netapp in the hallway track.

Attending BSDCan was an amazing experience. I met with many people whom I’ve only collaborated via e-mail and IRC, and also had a chance to discuss future plans and observe the group discussion in this BSDCan’s equivalent of a HaveNeedWant. I am incredibly grateful that the FreeBSD Foundation was willing to cover my travel expenses and allow me to attend, and I look forward to attending future conferences.

– Contributed by Kyle Evans