Please ignore the previous one patch, it would cause an infinite loop.<br>Attached is the updated patch.<br><br>Regards,<br>YJ.<br><br><br><div class="gmail_quote">2008/5/22 Yen-Jung Chang <<a href="mailto:yenjung@gmail.com">yenjung@gmail.com</a>>:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Hi Jouni,<br><br>I run into the same issue with 0.4.8, so I back port the robust mechanism from 0.5.10 to <a href="http://0.4.8." target="_blank">0.4.8.</a><br>
<br>Would you please give me some advice about the attached patch?<br><br>Thanks,<br>
YJ.<br><br><br><div class="gmail_quote">2008/3/18 Jouni Malinen <<a href="mailto:j@w1.fi" target="_blank">j@w1.fi</a>>:<div><div></div><div class="Wj3C7c"><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div>On Thu, Mar 06, 2008 at 09:18:46AM +0200, Uri Simchoni wrote:<br>
<br>
> Working with madwifi driver, we've encountered some rare conditions in<br>
> which clients can't receive WPA-encrypted multicast packets. This has<br>
> become a more serious lately since Windows Vista uses broadcast packets<br>
> for DHCP by default.<br>
><br>
> After digging into it for a while, we realized that the group-key state<br>
> machine was stuck in the SETKEYS state, meaning that it negotiates a new<br>
> group key, but doesn't activate the key at the driver. Further<br>
> investigation showed that the GNoStations variable got negative.<br>
<br>
</div>Thanks for reporting this.<br>
<div><br>
> The supplied patch (against 0.4.10, sorry...) fixes these issues, and<br>
> also remove the reliance on GNoStations, because it looks like a<br>
> not-very-robust way to determine how many stations need to negotiate<br>
> group key (although it's based on the standard...)<br>
<br>
</div>I replaced the group key update management with a more robust mechanism<br>
about a month ago (the new version is included in 0.5.10). It uses a bit<br>
different solution when compared to your patch, but I would expect the<br>
end result to be quite similar since GNoStations variable was removed<br>
completely and only the stations that are really active and ready for<br>
group key update are include in the dynamic count.<br>
<br>
I haven't ported the patch to 0.4.x branch yet, but I would prefer to<br>
use the same solution in all branches to minimize amount of code to<br>
maintain in the future.<br>
<font color="#888888"><br>
--<br>
Jouni Malinen PGP id EFC895FA<br>
_______________________________________________<br>
HostAP mailing list<br>
<a href="mailto:HostAP@lists.shmoo.com" target="_blank">HostAP@lists.shmoo.com</a><br>
<a href="http://lists.shmoo.com/mailman/listinfo/hostap" target="_blank">http://lists.shmoo.com/mailman/listinfo/hostap</a><br>
</font></blockquote></div></div></div><br>
</blockquote></div><br>