[Prism54-devel] Attempting to assoc with hostap-driven AP

[vda]

> My mistake. I should keep patch-2.6.N-prism54-cvsYYYYMMDD.bz2
> instead of just 'latest' in order to be able to track patch version.

[ CC'ed hostap folks, they were very helpful in the past ]

I investigated the matter further.
I have 2 boxen sitting within 1 meter from each other.
Both of them have one SMC 2802w.
One of them also have two DWL-520 cards,
one of DWLs is configured as an AP.
I will refer to it as 'an AP box'.

Both run 2.6.2 patched with prism54 cvs20040204
and hostap 0.1.2.

Here's what seem to work:

* I can configure one SMC as a master and other as an STA.
  Using small finger-sized antennas I see throughput of
  2.4 megabytes/s (netcat flood testing). Results are
  the same with and without WEP.

* I can configure SMC in an AP box as a STA and associate
  it with hostap-driven DWL-520 in the same box, even without
  any antenna. I presume they use mainboard circuitry
  as rather peculiar means of RF transport ;)
  iwconfig is steady at:
  eth1 IEEE 802.11b/g  ESSID:"ear"  Nickname:"ear"
       Mode:Managed  Channel:11  Access Point: 00:05:5D:FA:58:45
  and /proc/net/hostap/wlan0/00:04:e2:64:23:c2 is showing:
  last_rx: silence=156 signal=198 rate=10
       Note rather strong s/n ^^^
  Unfortunately I did not stress test this setup, kernel keeps
  sending traffic via loopback device intead of radio.
  This is how assoc event looks like in syslog:
  eth1: firmware uploaded done, now triggering reset...
  wlan0: 00:04:e2:64:23:c2 auth_cb - STA authenticated
  wlan0: assoc from 00:04:e2:64:23:c2 with extra data (10 bytes)
     [<32><08><0c><12><18><24><30><48><60><6c>]
  wlan0: 00:04:e2:64:23:c2 assoc_cb - STA associated

Now, what does not work:

When I try to associate using SMC in the other box as a STA, 
it does so but is unstable, as per my first mail to the thread.
'watch iwconfig' is intermittently showing:
eth1 IEEE 802.11b/g  ESSID:"ear"  Nickname:"ear"
     Mode:Managed  Channel:11  Access Point: 00:05:5D:FA:58:45
and
     Mode:Managed  Channel:xx  Access Point: 00:00:00:00:00:00
/proc/net/hostap/wlan0/00:04:e2:64:15:e5:
last_rx: silence=156 signal=175 rate=10
               s/n is lower ^^^

This is what I see in syslog in AP box:
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
wlan0: STA 00:04:e2:64:15:e5 TX rate lowered to 55
wlan0: 00:04:e2:64:15:e5 auth_cb - STA authenticated
wlan0: assoc from 00:04:e2:64:15:e5 with extra data (10 bytes) 
[<32><08><0c><12><18><24><30><48><60><6c>]
wlan0: 00:04:e2:64:15:e5 assoc_cb - frame was not ACKed
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
wlan0: STA 00:04:e2:64:15:e5 TX rate lowered to 55
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
wlan0: STA 00:04:e2:64:15:e5 TX rate lowered to 20
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
wlan0: STA 00:04:e2:64:15:e5 TX rate lowered to 10
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
last message repeated 2 times
wlan0: 00:04:e2:64:15:e5 auth_cb - authentication failed
wlan0: 00:04:e2:64:15:e5 auth_cb - STA authenticated
wlan0: assoc from 00:04:e2:64:15:e5 with extra data (10 bytes) 
[<32><08><0c><12><18><24><30><48><60><6c>]
wlan0: 00:04:e2:64:15:e5 assoc_cb - frame was not ACKed
wlan0: STA 00:04:e2:64:15:e5 TX rate lowered to 55
wlan0: 00:04:e2:64:15:e5 auth_cb - frame was not ACKed
wlan0: 00:04:e2:64:15:e5 auth_cb - STA authenticated
wlan0: assoc from 00:04:e2:64:15:e5 with extra data (10 bytes) 
[<32><08><0c><12><18><24><30><48><60><6c>]
wlan0: 00:04:e2:64:15:e5 assoc_cb - frame was not ACKed

Syslog in the other box does not reveal anything interesting.

So, maybe signal level is just too low? I doubt it,
here's why.

* SMCs were able to do 2.4 megs/s with two tiny antennas,
yet SMC and DWL fail at the same distance even when I
use large antenna and place it right near open box case.
In the past, I did that many times when I tested DWLs
and I had good signal and stable assoc.

All my old, non-prism54 STAs have similar or lower signal level,
sometimes very low (they are far away), but hostap driver
happily lets them associate. They are working right now:

/proc/net/hostap/wlan0/00:04:e2:64:15:e5
last_rx: silence=156 signal=175 rate=10  <-- troublesome SMC
/proc/net/hostap/wlan0/00:04:e2:64:23:c2
last_rx: silence=156 signal=198 rate=10  <-- SMC which is ok
/proc/net/hostap/wlan0/00:05:5d:fa:1a:e3
last_rx: silence=156 signal=168 rate=110 <-- DWL
/proc/net/hostap/wlan0/00:05:5d:fa:56:20
last_rx: silence=160 signal=170 rate=110 <-- DWL
/proc/net/hostap/wlan0/00:05:5d:fa:56:36
last_rx: silence=156 signal=178 rate=110 <-- DWL
/proc/net/hostap/wlan0/00:05:5d:fa:58:2b
last_rx: silence=164 signal=173 rate=110 <-- DWL

Maybe SMCs just are far worse wrt working with RF signal?
Then, why did they work so good in "2 SMCs" testing?
--
vda