Re: MA111 problem at boot on Fedora FC3

Peter

In my experience, this most likely works because on
some systems the command messages between the prism2_usb
driver and the actual card can fail (for no very good
reason), but often issuing a second command then succeeds.

There have been many people who find that issuing
a command like 'ifstate=enable' twice is enough to
do the trick, because often if the first one fails, 
the second attempt does get through. Sometimes people
have found that simply inserting a sleep of 1 second,
e.g. after firmware upload, is enough to improve reliability
of the card initialising - but its seems somewhat random
and system dependant unfortunately.

Now you are on FC3, and have the card coming up okay, we
will all be interested to see if you have the network lockups some minutes later that other people have had. Do you have the HAL deamon (hald) running on your system?

Karl

> 
> From: "Peter Ison" <[email protected]>
> Date: 2005/06/06 Mon PM 09:18:41 BST
> To: <[email protected]>
> Subject: [lwlan-user] MA111 problem at boot on Fedora FC3
> 
> I am using Fedora Core 3 (kernel 2.6.11) with the wlan pre26 driver on an AMD 1200 MHz processor desktop PC, and have been having boot problems with the MA111 since the 2.6.9 kernel (and the pre 23 driver). I use WEP shared key security, and a static IP address. I leave the MA111 connected all the time. 
> 
> Previously I posted a method which entailed a loop in rc.local, removing module prism2_usb and restarting wlan repeatedly until a connection was achieved. This was slow and services which required access to LAN or internet could not connect during boot because the wlan0 interface was not yet available.
> 
> I have found a better way, which is to patch the script /etc/wlan/shared by adding one line as follows:
> 
> In /etc/wlan/shared 
>         wlan_enable()
>         .......
>         nsdname=`wlan_nsdname $1`
>         if ! ${nsdname}_fwload $1 ; then
>                 ${ECHO} "Firmware failed to load for device $1"
>                 return 1
>         fi
>         # Enable the interface
>         # ADD THE FOLLOWING LINE
>         $WLANCTL $1 lnxreq_ifstate ifstate=disable
>         result=`$WLANCTL $1 lnxreq_ifstate ifstate=enable`
> 
> The patched file /etc/wlan/shared is called by /etc/hotplug/wlan.agent and also by /etc/init.d/wlan.
> 
> I am not sure why this works, except that before the patch the result of the "wlanctl-ng  wlan0 lnxreq_ifstate ifstate=enable" in the above script was "implementation failure" as shown below:
> Jun  6 21:04:09 myntech kernel: hfa384x_docmd: ctlx failure=REQ_FAILED
> Jun  6 21:04:09 myntech wlan.agent[19532]: Failed to enable the device, resultcode= implementation_failure .
> Jun  6 21:04:09 myntech wlan.agent[19532]: WLAN Could not bring up wlan0
> Jun  6 21:04:09 myntech kernel: hfa384x_drvr_start: cmd_initialize() failed, result=-5
> Jun  6 21:04:09 myntech kernel: prism2sta_ifstate: hfa384x_drvr_start() failed,result=-5
> 
> 
> After the patch the wlan0 interface is now stable, and trouble-free.
> 
> I notice that after un-plugging and re-plugging, sometimes the network routing table entries get lost. To get going again I must do the following:
> 
> service wlan restart
> service network restart
> 
> and the link and network are restored.
> 
> Peter Ison
> 
> 

-----------------------------------------
Email provided by http://www.ntlhome.com/