hostap in master mode and firmware 1.6.3

[Jouni Malinen]

On Tue, Jan 28, 2003 at 05:27:09PM +0100, Martin Whitlock wrote:

> First of all I think that you should use PRI 1.1.1 together with STA 1.6.3.
> But still, it doesn't seem to work in HostAP mode. No stations can
> associate. I have verified with a spectrum analyzer that some kind of beacon
> is being sent periodically, and the transmitted frequency responds when I
> change channel with iwconfig. I suppose I should be able to capture the
> frames by putting another card in monitor mode, but I actually don't know
> exactly how to. Any hints?

Yes, getting a capture log should certainly help since I cannot test
this myself at the moment. You can use Host AP driver in another host to
do this capturing. Configure the sniffer on the same channel with the AP
and set it to monitor mode with 'iwconfig wlan0 mode monitor' (or
'iwpriv wlan0 monitor 2' if your wireless tools do not understand new
'monitor' mode option). You will also need to set wlan0 device UP on the
sniffer host. After this, you should be receiving the sniffed frames
with 802.11 headers from wlan0 interface.

You can make a capture log, e.g., using the latest version of libpcap
with Ethereal or latest (CVS?) version of tcpdump. Alternatively, you
could use wlansniff program in the Host AP package, but I prefer to get
capture logs in some libpcap-compatible raw mode that I can load into
Ethereal.

> wlan0: RX status=0x0000 (port=0, type=0, fcserr=0) silence=10 signal=43
> rate=10
> rxflow=0; jiffies=97746
>    FC=0x0080 (type=0:8) dur=0x0000 seq=0xb200 data_len=40
>    A1=ff:ff:ff:ff:ff:ff A2=00:60:b3:6f:8b:56 A3=00:60:b3:6f:8b:56
> A4=67:23:78:71
> :74:aa
>    dst=ff:ff:ff:ff:ff:ff src=00:50:ba:e7:75:67 len=40

This is a beacon frame from 00:60:b3:6f:8b:56. type=0 (management)
frames use addr1 as the destination, addr2 as the source, and addr3 as
the BSSID. You can ignore A4, dst, and src for management frames.

> wlan0: prism2_rx: dropped non-data frame (type=0x00, subtype=0x08)

Aah.. Now this is interesting. The management frame was not catched in
the RX path.. That would certainly break AP functionality. Now I would
like to test that even more myself ;-). Which card manufacturer is using
these new STA firmware images?

As a semi-blind guess, could you please try the attached patch (to the
current CVS version) with the new station firmware and report if there
are any changes in Host AP mode behavior?

> BTW, new features in STA 1.6.3 seems to be RIDs for Enhanced Security, i.e.
> hiding SSID in beacon and block responces to "ANY"-clients. "Frames to
> stations in promiscous mode will now be ACKed", but I havn't verified that
> yet. Otherwise it's mostly PS stuff.

Is that EnhSec RID also for Host AP mode now? It has been there long for
tertiary firmware AP.. Fixing ACKs is always welcome, they are kind of
needed in 802.11 ;-).

-- 
Jouni Malinen                                            PGP id EFC895FA
-------------- next part --------------
Index: driver/modules/hostap_hw.c
===================================================================
RCS file: /home/jm/cvsroot/hostap/driver/modules/hostap_hw.c,v
retrieving revision 1.71
diff -u -p -r1.71 hostap_hw.c
--- driver/modules/hostap_hw.c	11 Jan 2003 06:04:58 -0000	1.71
+++ driver/modules/hostap_hw.c	29 Jan 2003 04:29:37 -0000
@@ -2784,7 +2784,7 @@ static void hostap_rx_skb(local_info_t *
 	if (hostap_rx_frame_invalid(dev, macport, status, len))
 		goto rx_dropped;
 
-	if (msg_type == HFA384X_RX_MSGTYPE_MGMT) {
+	if (msg_type == HFA384X_RX_MSGTYPE_MGMT || type == WLAN_FC_TYPE_MGMT) {
 		if (hostap_rx_frame_mgmt(local, skb, type, stype))
 			goto rx_dropped;
 		else