Question on HostAP, IAPP link layer update and bridge station cache

[Jouni Malinen]

On Mon, Mar 24, 2003 at 01:38:25PM +0100, Jon-Olov Vatn wrote:

> 7) When the entry in station cache (of the bridge) in AP2 ages out, or
>    when the STA decides to "actively" transmit a packet, the bridge
>    will start to forward the Echo Requests to the STA.
>    This may take several seconds.
> 
> Have anyone else also noticed this behaviour?

Yes and that's why there is (a preliminary version of) IAPP in hostapd..

> Note that this problem would not occur if the entry for MACofSTA in
> AP2 had already aged out at the time of the handover, however, it
> would be good if one did not have to rely on the aging timeout in the
> AP bridge. I should also mention that I uncommented some lines in
> "iapp.c" (for initializing "iapp_udp_sock") in order to avoid some run
> time errors (perror("ioctl(SIOCGIFADDR)").  I do not believe that this
> affects the scenario above. (Anyhow I have attached my modified iapp.c
> file in case anyone is interested.)

And by doing this, you effectively disabled IAPP..

> Now to my question. If one would like to fix this problem, I can see
> two different approaches, and I wonder what way to go:
> * "Interaction between hostAP and briding code"
> * "Inserting link layer update frame in wlan0 input queue"

I think it would be useful to implement more interaction between Host AP
and bridging code. However, before doing this, I would add third option
to your list:
* use IAPP to update the previous AP

Current version of hostapd has quite limited version of this, but it
should work in the case you described, i.e., having the APs sharing same
shared physical medium so that they can receive broadcast packets from
eachother. This requires IP addresses (so that you do not need to
comment our the initialization). If the APs cannot receive broadcast
frames from eachother, more complete IAPP implementation would be
needed.

-- 
Jouni Malinen                                            PGP id EFC895FA