#776: regression: preempt_scan inhibits association to AP
---------------------------------------+------------------------------------
Reporter: [EMAIL PROTECTED] | Owner:
Type: defect | Status: new
Priority: major | Milestone: version 0.9.x -
progressive release candidate phase
Component: madwifi: driver | Version: trunk
Resolution: | Keywords:
Patch_attached: 1 |
---------------------------------------+------------------------------------
Comment (by [EMAIL PROTECTED]):
I sent a private mail to David Platt. Here is his answer (he gave me
permission do attach it to the ticket):
{{{
As far as I can tell, the problem exists because the madwifi driver
(or the firmware) seems to treat "scanning" and "association" as being
rather similar sorts of things. It seems to be the case that if you
tell the low-level code to abort a scan-in-progress, and if the
firmware is currently trying to associate with an AP, then the
association operation itself ends up being aborted, and the client
fails to associate with the AP.
This problem hits some users and not others. I believe that it's
extremely AP-and-timing-specific - whether it affects you or not
depends on things like the length of time that the AP requires to
process an association request, the beacon rate, other activity on the
AP, and perhaps the phase of the moon and the price of granola in
Dubuque.
I've looked through the madwifi driver code a bit to see if I could
figure out how to avoid the problem. I didn't get all that far...
I've concluded that the madwifi/80211 code structure which handles
scanning and association has a bunch of race conditions in its design.
The passive-scan-vs.-active-scan condition that I originally noticed,
and tried to resolve by aborting the existing scan to start a new one,
isn't the only one and is perhaps not the worst. I rather suspect
that the whole flow-of-control needs to be redesigned as a more
complex state machine, in order to ensure proper operation in the face
of timing variability.
At about the time I came to that conclusion, I bought a new laptop
which has an Intel ipw2200-type miniPCI wireless card. Since I'm no
longer using an Atheros-based NIC I haven't put any further effort
into trying to rework the madwifi driver - that's a task which would
take more time than I have available at this point.
The best I can suggest is to un-apply the cancellation patch for your
card, since your setup is apparently more sensitive to timing-related
association problems than some other peoples'.
}}}
--
Ticket URL: <http://madwifi.org/ticket/776>
MadWifi <http://madwifi.org/>
Multiband Atheros Driver for Wireless Fidelity-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Madwifi-tickets mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/madwifi-tickets
