Date: Fri, 29 May 2026 17:15:34 -0700 From: Adrian Chadd <adrian@freebsd.org> To: freebsd-wireless <freebsd-wireless@freebsd.org> Subject: proposed upcoming work - channel representation stuff Message-ID: <CAJ-Vmom4=5R-xjgHYwbvOexv22u2Z3MU1m6y252WscJe9cr3GQ@mail.gmail.com>
index | next in thread | raw e-mail
hi! We've long known that the way channels are represented in net80211 can't scale. It's one entry per frequency per width per mode (a,b,g,n,ac) per HT/VHT channel combination. It is pretty bad. I've been tinkering with an idea for refactoring the channel pointer use into something that will be easier to further refactor/rewrite. I don't yet have a firm plan on what the end goal will be with regulatory and channel configuration, but I think this will be good refactoring work regardless. The initial proposal write-up is here - https://wiki.freebsd.org/AdrianChadd/Net80211ChannelRepresentation2026 - with some exploratory work in https://reviews.freebsd.org/D48172 . I'm running with that right now on a couple of laptops and so far so good, but obviously it'll need a lot more reviews and testing before it can land in -HEAD. If you're interested then please take a look at the proposal and diff and let me know what you think. It's a pretty big change and I'm not sure I'd want to see it MFCed back to stable/15. This however will have an impact on backporting later stack and driver work. Thanks! -adrianhome | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-Vmom4=5R-xjgHYwbvOexv22u2Z3MU1m6y252WscJe9cr3GQ>