[Cerowrt-devel] resuming on the make-wifi-fast front

Dave Taht dave.taht at gmail.com
Sun Sep 19 12:24:03 EDT 2021


It looks like I might get a bit of new funding (already had some from
comcast research) to resume work on
make-wifi-fast stuff. I'm pretty happy with openwrt 21.02 so far, and
wifi is my first love.

In this case, my goal was to first test and verify correctness across
several newer routers, and I got stopped by personal/political
overload from doing much after 21.02-rc2.

I sometimes have had  the ambition to do another cerowrt (starwrt!),
based on several months of realization that everybody, not just
starlink,
was shipping ancient kernels now, with crippling and weird offloads.
But there's no way, not without 3 years of steady funding lined up,
bringing in new people, and somehow trying to pay those putting in the
most valuable work, I would put myself and us into that level of
commitment again.

There's all kinds of other stuff, in the queue there, like leveraging
some of prplOS TR-XXX stuff, apple's thread and unicast mdns
implementations, homenet...

I'm also VERY interested in getting in on the openwifi FPGA project.

So, in the case for the small grant I'm writing today, and for a
larger one later, I am seeking small, limited goals, with incremental
and testable gains, that can be easily cycled directly though the
openwrt project, for wifi, specifically. And also looking for folk, on
a paid or unpaid basis, willing to help, on any front, once again. On
my *personal* list are:

verifying correctness of the fq_codel implementations in 5+ different chipsets
implementing an algorithm to handle hardware retries better
sort on dequeue
ack-filtering
reducing max txop sizes (also in the beacon), under load

I fear that step 1 above might eat everything and all the time I might
be able to get!? I hear the archer c7v2 is being problematic, and
there are some problems with the mt76 elsewhere, also.

I am very interested in y'all views as to what would be useful to
tackle in wifi, in the next 3 months or so, and then, over the next
year. Big picture, small picure, whatever axe you might have to grind,
go for it!





-- 
Fixing Starlink's Latencies: https://www.youtube.com/watch?v=c9gLo6Xrwgw

Dave Täht CEO, TekLibre, LLC


More information about the Cerowrt-devel mailing list