<br><div class="gmail_quote">On Thu, Oct 22, 2009 at 6:41 PM, Jouni Malinen <span dir="ltr"><<a href="mailto:j@w1.fi" target="_blank">j@w1.fi</a>></span> wrote:<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, Oct 22, 2009 at 06:23:08PM +0200, Maxence Dalmais wrote:<br>
<br>
> I am unable to understand what happens, but just after the binding, my whole<br>
> computer freeze (ie, I have to reboot).<br>
> I don't know if the problem come from Hostap, the driver, or the kernel.<br>
<br>
</div>hostapd should not be able to freeze the computer (hang/crash the kernel<br>
most likely), i.e., I would expect this to be an issue in the kernel<br>
(ath5k or mac80211).<br></blockquote><div><br>The problem is the same with compat-wireless version (2.6.32 and the last git version) , with hostap comming from the last git update.<br><br>With the the last stable release, neither crash nor problems in hostapd logs appears, but vlan is not plainly functionnable. ( Rx and Tx workind differently ...)<br>
<br>My system in an Ubuntu 9.04 with a 2.6.28 kernel.<br><br> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div><br>
> The assignment to the VLAN seems to be OK, but the crash come just after.<br>
> It looks like a packet arrived just after and make the server crash.<br>
<br>
</div>I have not tested this with ath5k, but the last time I tested, dynamic<br>
VLAN setup was working without problems (I think that test was with<br>
mac80211_hwsim as the driver).<br></blockquote><div> </div><div> Is mac80211_hwsim not only to create virtual device for client ?<br> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div><br>
> Another probleme that I notice is that the virtual interface created are<br>
> named "wlan0.X_rename" . Whereas the hostap.vlan file specify "wlan0.X" (for<br>
> example wlan0.5_rename instead of wlan0.5)<br>
> The i802_set_sta_vlan function use the "wlan0.5" notation.<br>
> Do you think that could be a part of the problem ? if yes, how to solve this<br>
> ?<br>
<br>
</div>I would not expect this to be the cause for the hang. The interface is<br>
not created as wlan0.X_rename; it is created as wlan0.X. Something else<br>
in the system is renaming the interface afterwards. I don't know the<br>
details of your system configuration, so I cannot be sure what is doing<br>
this, but anyway, I would assume it is udev and its persistent net<br>
rules. It should be possible to disable that somehow (google udev<br>
persistent net rules for more info).<br>
<div><br></div></blockquote><div><br><br>I'm now certain that the problem don't come from here.<br>I change my hostap.vlan to have vlanX, and that now Ok.<br> <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div>
> Can somebody give me more information to help me to solve my problem ?<br>
<br>
</div>Have you seen any error messages from the kernel? I would suggest<br>
running the test in text mode (i.e., with kernel messages showing up on<br>
the console). That may be enough to get the needed information to figure<br>
out which component is crashing.<br>
<font color="#888888"><br></font></blockquote><div><br>Since my kernel crash, it is pretty hard for me to get a full kernel trace.<br><br>I have currently solved my probleme by using the madwifi driver, and having adapted the patch referenced in <a href="http://madwifi-project.org/ticket/446">http://madwifi-project.org/ticket/446</a> .<br>
<br><br>Maxence.<br></div></div>