8.2.1 WPA testing
Chris Ball
cjb at laptop.org
Sun Mar 1 19:08:26 EST 2009
Hi,
I've been working on trying to quantify the difference in WPA behavior
between 8.2.0 and 8.2.1. My test setup involves attempting to connect
to a WPA AP inside Sugar, performing suspend/resume, and trying again.
A success involves Sugar reporting "Connected", and a failure involves
Sugar asking for the WPA key again -- when this happened, I clicked
cancel, suspended, and began the next iteration.
Failures leave the string "Activation (eth0/wireless): disconnected
during association, asking for new key" in /var/log/messages. A second
type of failure is seen only on 8.2.0, namely "Unhandled network
capabilities 10000001"; this bug is fixed in 8.2.1, and was not included
in the totals below.
Machine 1, build 801:
50 iterations:
success: 23 (46% of attempts successful)
failure: 27
Machine 2, build 801 with 767 kernel and wireless firmware:
50 iterations:
success: 30 (60% of attempts successful)
failure: 20
Machine 3, build 767:
50 iterations:
success: 32 (64% of attempts successful)
failure: 18
Conclusion:
For this particular bug (being asked to re-enter the WPA passphrase
at connection time), with the particular WPA access points at 1cc, we
already had sporadic failures as of 8.2.0. 8.2.1 appears to make them
slightly worse, perhaps suggesting a widening of a race condition in
the newer driver or wireless firmware. We already know installing
WPA keys to be time-critical: see relevant recent commits ¹ and ².
Since the 8.2.0 behavior is seen on an 8.2.1 build with kernel and
firmware downgraded, we can rule out changes to Sugar/NetworkManager
as having caused the 8.2.1 regression.
Advice:
We should not claim reliable WPA support in 8.2.1. Since we have
other WPA fixes available in the 8.2.1 kernel (such as ³) the new
kernel is perhaps still a better 8.2.1 candidate than the old one.
In either case, we should document the workaround of "if you've
previously typed in the WPA passphrase and you're asked for it
again, you should click Cancel and attempt to reconnect", which
usually works. Also, note that this bug is not the only reported
failure around WPA, although it is the bug that seems to affect all
access points (rather than just individual ones).
There's a large delta between the 8.2.x kernel of 2.6.25 and the
current upstream 2.6.29 work; further effort at making WPA reliable
would probably be better-focused on the upstream kernel rather than
our 2.6.25 snapshot.
Further comments/thoughts welcome!
- Chris.
¹: http://dev.laptop.org/git?p=olpc-2.6;a=commit;h=f10b654367d7065d50e7d5e3649933453cea3102
²: http://dev.laptop.org/git?p=olpc-2.6;a=commit;h=b0a7455f0ba53774a30a7af6505816855488cc40
³: http://dev.laptop.org/git?p=olpc-2.6;a=commit;h=011dd3f2d6fbd75ae2aacdcf8f99ebeddf814227
--
Chris Ball <cjb at laptop.org>
More information about the Devel
mailing list