Summary: rc6 testing is going well, kernel.org is down and dave needs a vacation, so rc7 is going to slip a few weeks. Details of the rc7 plan below. [Original News item posted 24Sep2011; moved to the Wiki 17Mar2012]
As the kernel development process has slowed due the ongoing recovery
from the kernel.org, bottlenecking changes to cerowrt and
debloat-testing… and I can’t offer anything but empathy to those at
kernel.org getting things back online….
… and I just arrived in Paris to work out of the lincs.fr lab there
for the winter, where I hope to be doing some deep analysis of the
data collected so far, and working on the harder algorithmic and
Now seems like the perfect time to finally take some vacation.
so (sept 30-oct 10) I will be taking some serious time off in Paris
and Giverny to recharge my batteries, and also find a place near the
lab to live, and get integrated with matters here…
So, I’ve slipped cerowrt’s dates back a couple weeks. Some pruning of
the current list is in order if we are ever to get something
Cerowrt-1.0-rc6 is proving stable, however during that process we’ve
found a few major issues that are on the roadmap to fix in rc7. At the
kernel level, many are trivial (yet important - see bugs 266,265)
fixes that should automagically arrive after the Linux development
process gets back closer to normal. We’re now struggling mostly with
issues higher on the stack (see bugs 277, 113, 268, etc), and an
onslought of new packages and feature requests from various parties
(274, 279) - all of which are too much, combined, to deal with, in my
exhausted, jetlagged state.
All: Please feel free to add features, bugs, and suggestions to the
rc7 release and rc7 release plan after reviewing the RC6 release
All: Also feel free to jump in and fix some of this stuff - much of it
now is in mainline code, not openwrt or kernel code. Example: bug 224
is so simple yet crosses so many boundries as to make my head spin -
and it’s all over a SINGLE BIT. The right answer is patches to glib
and uclibc, or so I think. Another simple set of patches is fixing
dscp marking under ipv6 for things like openssh…. a great mystery
regarding TOS setting under IPv6 has been resolved in bug 249…
And do keep testing RC6! Downloads available at:
My efforts this coming week will be in getting bloatlab #1 up and
fully operational. Preliminary documentaion on that lab is at:
and iperf and netperf servers are enabled on jupiter and io.
I’m very happy with how far we have come with the bufferbloat effort
in the 9 months since the inception of bufferbloat.net - in particular
the recent lwn.net article (which will be available to non-subscribers
soon) and the responses to it were heartwarming as to how much people
‘got’ what we have been trying describe, diagnose, and fix.
And I’m really glad everyone has been so helpful, and also really glad
to finally be planning a little time away from the bloat.