August 27, 2012
The last of the BSDCan trip reports have been received. Hiren Panchasara posted his on his blog. Adrian Chadd’s report is as follows:
The last couple of years has seen me take stewardship of the net80211 stack and Atheros wifi driver development and maintenance. The FreeBSD Foundation provided me with a travel grant to enable me to participate in the BSDCan 2012 developer summit and conference as the wireless stack maintainer. For this I am very grateful.
I split my time between attending events and talking with people about networking and wireless in general. I even had a few developers come up to me and ask for some assistance in debugging their particular wireless problem. Everyone seemed very happy that the wireless subsystem was being actively developed again by myself and the other volunteers, which unfortunately couldn’t make it to BSDCan.
I presented a wireless update during the developer summit, which was well received. Although the focus was FreeBSD, the progress being made in the other BSDs has been very impressive. OpenBSD in particular has made great strides in porting wireless drivers from Linux and other sources. It was a shame that there wasn’t a strong presence from OpenBSD this year as I’d have loved to discuss this with them.
It was good putting names to faces. David Italiano and I had been talking online about PMC and MIPS related things. Alexander Motin and I had some brief discussions about time-keeping with his new tick management system, and some of the challenges faced with ACPI and MIPS. Most importantly (for me) was to meet Stefan Bethke, who had been working on the ethernet switch support in FreeBSD. We committed the work to FreeBSD-current during the course of the conference so that the code was in the tree and available by the time he gave his BSDCan presentation. I worked with Justin Hibbits on trying to identify endian problems in one of the wireless drivers (since he develops on PPC platforms) and provided him with an Atheros NIC which does work on PPC. I hope to continue working with all of these people and more in the future.
The networking discussions during the developers summit was of particular interest. Wireless support provides many challenges to a network stack design, both in terms of locking and in terms of queue management. Wireless devices are also approaching and exceeding gigabit-per-second speeds and this puts them on par with the traditional, ethernet style networking. I provided some feedback during the mbuf and multi-queue redesign discussions which I hope will see FreeBSD continue to maintain high-quality and high-performing wireless support now and in the future. The usual suspects were there (Robert Watson, John Baldwin) and it was good to bounce ideas off of them.
I also found the vendor summit quite inspiring. The turnout was fantastic and many, many vendor representatives had a lot to say -both positive things about the current supported features as well as providing excellent suggestions for new features. Although the wireless side of things was woefully under-represented (I believe I was the only active wireless representative there) I hope to help change this in the future. George kept writing down blackboards full of ideas and for a while I wondered when the ideas and brainstorming would stop. All in all, a very positive and reassuring event.
I am very thankful for the FreeBSD Foundation to provide me with the means to participate in this year’s BSDCan. I look forward to participating next year.