current CVS breaks with madwifi

[Kel Modderman]

On Thursday 12 October 2006 10:49, Matt Brown wrote:
> Hi Jouni,
>
> The bug tracker is throwing 403 to all requests atm, so I'll post here
> instead.
>
> The current CVS version does not work with madwifi, in either plain text
> or encrypted mode. The error is:
>
> SIOCGIWRANGE: WE(compiled)=19 WE(source)=13 enc_capa=0xf
> ap: IEEE 802.11 Fetching hardware channel/rate support not supported.
> Flushing old station entries
> madwifi_sta_deauth: addr=ff:ff:ff:ff:ff:ff reason_code=3
> ioctl[IEEE80211_IOCTL_SETMLME]: Invalid argument

Reproduced here with recent madwifi and current hostapd CVS.

>
> The cause is the rearrangement of the interface setup routines that you
> committed as part of the multi-bss support. Specifically version 1.163
> of hostapd.c committed on 8th Sep.
>
> The driver_madwifi init routine sets the interface down.
>
> Previously ieee802_1x_init was called before hostapd_flush_old_stations
> as hostapd_setup_interface was executing, ieee802_1x_init brought the
> interface up and the flush call succeeded fine.
>
> Now, however, you call hostapd_flush_old_stations before calling
> hostapd_setup_bss, and 802.1x is not initialised until
> hostapd_setup_bss, so the end result is that hostap tries to deauth
> stations on a madwifi interface which is down, madwifi returns -EINVAL,
> and hostapd dies.
>
> I'm not sure what the best fix for this is, but I'm happy to help out
> with patches, more testing if required. Just let me know what I can do.

Neither am I, unfortunately.

Thanks, Kel.