[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ipw2100-devel
Subject:    Re: [Ipw2100-devel] Idea + patch: start probing after "ifconfig up" instead of driver loading
From:       Stefan Rompf <srompf () isg ! de>
Date:       2004-06-18 16:59:15
Message-ID: 200406181859.16137.srompf () isg ! de
[Download RAW message or body]

Am Mittwoch, 16. Juni 2004 17:20 schrieb James Ketrenos:

> >OTOH, ipw2100 behaviour is quite unique in that case. Berlios atmel
> > driver, madwifi, wlan-ng, and, from a glance at the source, prism54
> > enable the transmitter only as long as the interface if configured up.
>
> AFAIK the other cards all have "association" == "link state". 

Link state is reflected to userspace as IFF_RUNNING via 
netif_carrier_on/off(), and last time I've checked many wireless cards did 
not touch that flag - there has been some discussion and a conclusion between 
the HostAP people though.

> This was 
> one of the key motivators to change ipw2100 away from the model of only
> turning on the radio when ifup--users of the other cards were confused
> by the behavior of the ipw2100; in addition, the ipw2100 didn't work
> right with the networking scripts of various distributions.

So it seems that even the existing drivers - whether in kernel or external 
modules - behave inconsistent. Therefore I'd suggest that I'll address the 
topic on netdev. Currently, there is some discussion about rethinking 
wireless extensions, and the question when to enable the transmitter fits 
nicely.

> >Many ethernet cards without MII phy (one may argue that they are obsolete
> > by now, but anyway) behave different.
>
> How do they behave?  It was my understanding that you plug in the cable
> and when/if link is detected (exposed by ethtool get_link) then scripts
> kick in and bring up the interface.  No link, no interface.

No. When the interface is configured up and the link is detected or lost, in 
2.6 a netlink message is sent by the most popular drivers and the IFF_RUNNING 
flag is set accordingly. I'd consider scripts that check for link once and 
refuse to start the interface if none is found broken.

> I would prefer you to not have to do that; would the virtual RF switch
> meet your needs?  You could then do:
>
> % echo off > /proc/net/ipw2100/eth1/state

Thanks, that would a be start, until an agreement on netdev has been made.

Stefan


-------------------------------------------------------
This SF.Net email is sponsored by The 2004 JavaOne(SM) Conference
Learn from the experts at JavaOne(SM), Sun's Worldwide Java Developer
Conference, June 28 - July 1 at the Moscone Center in San Francisco, CA
REGISTER AND SAVE! http://java.sun.com/javaone/sf Priority Code NWMGYKND
_______________________________________________
ipw2100-devel mailing list
ipw2100-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ipw2100-devel
[prev in list] [next in list] [prev in thread] [next in thread]