SCIENTIFIC-LINUX-USERS Archives

October 2004

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"Alan J. Flavell" <[log in to unmask]>
Reply To:
Alan J. Flavell
Date:
Wed, 13 Oct 2004 21:15:43 +0100
Content-Type:
TEXT/PLAIN
Parts/Attachments:
TEXT/PLAIN (79 lines)
It appears that the user-friendly scripts on redhat-flavoured linuxes
expect wireless lan cards to have interface names of ethN for some
value of N=0..7.


I found two classes of solution for wireless lan cards: a native linux
driver, or "ndiswrapper".  (ndiswrapper - in case anyone hasn't met it
- provides an environment in which win32 wireless lan drivers can be
used, subject to some restrictions I gather.)

"ndiswrapper" solutions, by default, present an interface name of
wlan0; but can easily be made compatible with the ethN interface name
which redhat^W SL expects, and this is shown in the ndiswrapper
documentation: the clue is to add to /etc/modules.conf the lines

options ndiswrapper if_name=eth%d
alias eth1 ndiswrapper

This has worked already for me on a couple of different types of card:
an expensive Netgear type, and an inexpensive unbranded type based on
the texas instruments chipset.  Neither of which I really wanted to
buy for our group.

I'm now trying to work up a card that's based on the Rt2500 chipset:
the card itself has the type designation EW-7108PCg, and looks to be a
promising choice for a group purchase (low cost, good wireless range,
friendly Windows software, and the option of a native linux driver).


At the moment the ndiswrapper solution (version 0.11) seems to be
usually working, but has sometimes crashed the OS, with each of the
win32 driver versions I have tried.  The driver I'm using at the
moment calls itself version 2.02.08.0000 (if anyone's interested)
and seems mostly OK (on a few hours experience of usage).

So much for the ndiswrapper approach...


As for the native driver, downloaded as source from
http://www.ralinktech.com/supp-1.htm , this evidently can work, but it
creates an interface called ra0.

(Obviously I built the source code first...)

When I load the kernel module manually with "insmod rt2500.o",
there appears in the log (/var/log/messages) a line which says

 localhost /etc/hotplug/net.agent: invoke ifup ra0

which shows some promise.  But, as I say, the GUI (sytem tools ->
Internet configuration wizard) for defining a wireless LAN interface
doesn't seem to admit to the existence of this interface ra0.  So I
guess I'd need to resort to coding the network scripts manually(?)

 Well, to be honest I did try brutally editing the source, but it
 didn't help.  I found two instances of ra%d coded explicitly in the
 source code (no parameterisation!) and tried hacking them to say
 eth%d, but presumably I missed some other places in the code which
 would have to be adjusted to suit.  So I reverted to the original.

The thing seems to be somewhat of a hall of mirrors, because part of
the machinery (the hotplugging, I guess) is evidently acting "behind
the scenes" and moving things around under one's feet.

Has anyone got further than me with this, please?  I tried googling
the archives but don't yet seem to have hit the spot.  There's a guy
toying with both options here
http://www.fedoraforum.org/forum/archive/index.php/t-21460.html
and another here
http://www.linuxquestions.org/questions/history/225068

In both of the articles, people seem to be issuing ifconfig commands
directly; but, as I say, there appeared to be processes (hotplugger?)
watching what I was doing and moving the scenery around when I was
trying it.  I'm not sure whether to try and find how to stop that
happening, or how to co-operate with it :-}

thanks

ATOM RSS1 RSS2