I am excited to see a AR910x specific patch appear in ath9k from an Atheros employee.  It looks to me like the real-time-clock wasn't being initialized properly for AHB bus.  This could explain a lot of performance issues... here is hoping!  I have not tested, but it should be in the compat-wireless 2009-02-18 release and I'll test it tomorrow.

http://git.kernel.org/?p=linux/kernel/g … 8ac3fcd86d

(Last edited by RoundSparrow on 18 Feb 2009, 00:25)

Follow-up, with compat-wireless 2009-02-19 and later... I think my ping-climbing problem is now resolved, but I still have erratic retrain of rate and still performing 3x to 4x slower than the native firmware driver (madwifi based) that these routers come with.

The primary thing that I think should be resolved is why these chips do not behave the same as other chips in client mode.  I am trying to get Atheros employees to try out client mode and see what they can do https://lists.ath9k.org/pipermail/ath9k … 01245.html

great work,

A bunch of new hostapd wme_* parameters are being discussed for ath9k.  I thought these were madwifi only, but it appears not!  So where are all these documented?

See: http://forum.openwrt.org/viewtopic.php?pid=83027#p83027

Another new exciting ath9k patch to test that should improve the stability issues.

https://dev.openwrt.org/changeset/14745/trunk

I tested against a Planex MZK-W04NU (ar9132/ar9103) in AP mode running OpenWrt SVN 14745.  Client is a laptop running Ubuntu 8.10 with Intel iwl3945 driver and chipset.  Tests about 6 feet apart, this is G only mode.

1.  Very stable ping times to the router from the client.  out of 39 packets, 0% loss min/avg/max/mdev 1.051/1.602/6.147/0.931
2.  with wpa2-aes private, I was able to copy from wireless client to local LAN server with speeds of 2.1MB/second.

This is far better than what I've been seeing lately on the Trendnet.  Not sure if it is the Planex doing better other the ar9130/ar9102 Planex or what.  I'm also testing at a different physical location, not much interference here.  I will retest the Trendnet on a newer release in a day or two and report back if I get different results.

(Last edited by RoundSparrow on 4 Mar 2009, 22:06)

Hello friends,
did the wifi 5ghz 300 mbps work now, on TEW-632 with openwrt ?
Ced

Just curious if both the madwifi and ath9k drive can run at the same time? Probably a stupid question, but I'm looking at trying it with 2 cards in the RouterStation, one atheros G card and one atheros N card. Would a conflict arise in hostapd if I put setting in for N? Thanks for any and all replies.

Hi guys,

I'm planning to move from oleg's firmware to openwrt on a asus wl500gP v1. I also want to use a different wireless adapter wich supports N. Preferably with one antenna. I found out that TL-WN722n usb adapter is atheos based, has antenna connector, supports N and also it is a high gain one. It has atheros ar5008x chip (looks like it is the first atheros usb model).

Do you guys think that it is gong to work openwrt running on a wl500gp v1?

dp79

just wanted to share my experience that a stable client mode (STA) on a RouterStation (ar7161) platform using R54n wireless (ar9220) miniPCI
with OpenWrt trunk r18058 (ath9k etc. 2.6.30.8) I could only achieve by first bringing the card up as an AP, to bring hostapd and the card down, and then to initialise it as STA using wpa_supplicant.
Straight into STA mode produces good pings towards the AP and back (~2 ms) but erratic behaviour when pinged from the AP and back (100ms) although with a strong signal and no noise.
Loads of disconnects (Reason: 3).  After a while the system may crash.
Doing the STA with wpa_supplicant after briefly having brought it up before as an AP with hostapd results in rock stable performance.
There must be some setting(s) done in AP mode to set the hardware and drivers up that are also necessary for STA mode but have been forgotten.