Please note: This project is no longer active. The website is kept online for historic purposes only.
If you´re looking for a Linux driver for your Atheros WLAN device, you should continue here .

Ticket #1908 (new defect)

Opened 14 years ago

Last modified 14 years ago

Channel selection on boot 'not working' - MIPS-EL (OpenWRT)

Reported by: nenad@panline.net Assigned to: benoit
Priority: minor Milestone:
Component: madwifi: driver Version: trunk
Keywords: mips transmit bug openwrt Cc:
Patch is attached: 0 Pending:

Description

I have tried madwifi-trunk-r3562 on MIPS platform (OpenWRT on ASUS WL-500gp), kernel 2.4.32 and experienced problems. On same hardware, madwifi-ng-r2628 works very good. Madwifi-trunk-r3562 is setup as AP.

Wireless interface setup script is attached. This script works for r2628.

Problems are:

  • after booting, channel should be set to 165 (and domain set to 11a), but it stays on channel 1. Only after ifconfig ath0 down; iwconfig ath0 channel 165; ifconfig ath0 up channel is changed to 165. After that, iwpriv ath0 doth_chanswitch 149 20 also works.
  • now, when on channel 165, client associates, but there is no transfer. Periodically, some of the packet pass (plain icmp ping) but with large RTT, ~5000ms. Kernel on AP side reports:
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    <6>NETDEV WATCHDOG: wifi0: transmit timed out
    

Strange thing is that r3562 works fine as 802.11b AP on i386 platform (PCengines WRAP). Both MIPS and i386 are running on kernel 2.4.32

Attachments

ifup-madwifi.sh (0.6 kB) - added by nenad@panline.net on 04/24/08 03:07:16.
bring up ath wireless interface
athdebug.txt (13.9 kB) - added by nenad@panline.net on 05/06/08 13:37:08.
athdebug dump (r3551)

Change History

04/24/08 03:07:16 changed by nenad@panline.net

  • attachment ifup-madwifi.sh added.

bring up ath wireless interface

(follow-up: ↓ 2 ) 04/24/08 06:34:14 changed by mrenzmann

Blind guess: r3552 introduces changes that are related to an issue related to NETDEV WATCHDOG. Could you please test r3551 and r3552, to possibly confirm whether your problem has been introduced by the changes in r3552?

(in reply to: ↑ 1 ; follow-up: ↓ 3 ) 04/25/08 01:56:47 changed by nenad@panline.net

I have tested r3551, with similar results:

<3>wifi0: ath_bstuck_tasklet: Stuck beacon; resetting (beacon miss count: 11)
<6>NETDEV WATCHDOG: wifi0: transmit timed out
<6>NETDEV WATCHDOG: wifi0: transmit timed out
<6>NETDEV WATCHDOG: wifi0: transmit timed out

after 5-10 secs of transfer.

Initial channel setup is not working, as in r3562. Both r3551 and r3562 has changed MAC address of card (00:0B:6B:xx:xx:xx has become 06:0B:6B:xx:xx:xx).

(in reply to: ↑ 2 ) 04/25/08 06:16:15 changed by mrenzmann

  • owner set to benoit.

Replying to nenad@panline.net:

I have tested r3551, with similar results:

Thanks. This might be interesting for benoit, he was (is?) working on that.

Both r3551 and r3562 has changed MAC address of card (00:0B:6B:xx:xx:xx has become 06:0B:6B:xx:xx:xx).

This is expected behaviour, see #1882 or #1716, for example.

04/25/08 09:14:20 changed by benoit

I am working on it. I'd like 2 things:

  • have a log with athdebug +watchdog +calibrate enabled
  • bisect to identify which changes introduces the bug (between r2628 and r3551). This will help understanding the bug (I'm not an expert, just trying to understand).

05/06/08 13:36:37 changed by nenad@panline.net

Sorry for delay. I'm sending athdebug dump, done with r3551.

05/06/08 13:37:08 changed by nenad@panline.net

  • attachment athdebug.txt added.

athdebug dump (r3551)

05/06/08 18:11:13 changed by nenad@panline.net

Here are the results of bisection:

  • last good SVN is r3547
  • r3548 works for some 20secs (approx 2MB transmitted)
  • r3549 does not work

05/06/08 18:15:02 changed by mentor

I suspect this is a duplicate of #1922; please would you check?

05/06/08 21:32:32 changed by nenad@panline.net

Yes, transmit on MIPS with r3616 works fine.

Anyway, problem with initial channel/mode selection still remains.

05/07/08 00:24:05 changed by mentor

  • priority changed from critical to minor.

Downgrading to minor for channel issues then.

05/07/08 00:24:42 changed by mentor

  • summary changed from Trunk - r3562 transmit not working on MIPS-EL (OpenWRT) to Channel selection on boot 'not working' - MIPS-EL (OpenWRT).