ACX100 Wireless

From SlackWiki
Jump to: navigation, search

Introduction

This is a Howto on how I was able to get the ACX100 open source driver to work with my card and Slackware. There are several other good resources for this driver on the 'net, but I thought it would be good to create one more that included the specific steps I took in Slackware to get my wireless internet running.

For the time being, this is mostly just a placeholder, I will update this when I can find some spare time. Check out http://acx100.sourceforge.net for more info.


ACX100 Readme

This tarball is targeted at 2.6 inclusion only.
Not designed to work or even compile with 2.4.
Contact:
netdev@vger.kernel.org
acx100-devel@lists.sourceforge.net
acx100-users@lists.sourceforge.net
Bug reports:
Visit http://www.catb.org/~esr/faqs/smart-questions.html
Please describe your wireless setup, manufacturer and type
(acx100/acx100usb/acx111?) of your hardware. Which firmware image(s)
are you using? If problem is reproducible, #define ACX_DEBUG 2
in acx_config.h and modprobe driver with debug=0xffff.
Post resulting kernel log (bzipped). It is large but very useful
for bug hunting. Try older versions of the driver.
Firmware images:
You should not supply firmware_dir= parameter anymore. Driver will try
to load the following images via hotplug (not from /usr/share/acx
directory as older driver did, hotplug firmware directory location
varies for different distros, try /lib/firmware or
/usr/lib/hotplug/firmware):
PCI driver:
'tiacxNNNcMM' (NNN=100/111, MM=radio module ID (in uppercase hex)):
combined firmware for specified chipset and radio.
Failing that, it will try to load images named 'tiacxNNN'
(main firmware for specified chipset) and 'tiacxNNNrMM' (corresponding
radio module). For example, my firmware is in file named 'tiacx111c16'.
Alternatively, I may remove it and use pair of files 'tiacx111' and
'tiacx111r16' instead.
USB driver:
image is named 'tiacxNNNusbcMM'
Build instructions:
* Create drivers/net/wireless/acx subdirectory inside
 your kernel tree.
* BTW, if your kernel has drivers/net/wireless/tiacx directory,
 you already may have acx driver (some different version).
 Decide which one do you want.
* Unpack tarball into drivers/net/wireless/acx directory.
* Add a line to drivers/net/wireless/Makefile:
     obj-m += acx/
* Build your modules as usual (perhaps "make modules modules_install").
This will create acx module.
Remove "acx-obj-y += usb.o" line in Makefile
and "#define CONFIG_ACX_USB 1" line in acx_config.h
if you want PCI-only driver. Ditto for USB-only one.
USB snooping:
If you are an USB driver developer and need to see USB traffic,
http://benoit.papillault.free.fr/usbsnoop/ may be useful.
Another very good way to snoop the USB frames is under Linux if your
driver happens to run under ndiswrapper (which is often the case),
either by savage printk's, or by using the USB snooping facilities from
Pete Zaitcev.


From: Per Bjornsson <perbj@stanford.edu>
To: ACX100 user mailing list <acx100-users@lists.sourceforge.net>
Date: Fri, 08 Jul 2005 13:44:26 -0700
Hi,
Just a note for those who, like me, prefer not to bother recompiling
their whole kernel: It's not at all difficult to compile Denis's
snapshots out of tree, despite what the readme file says! At least in
somewhat recent 2.6.x kernels, there's a special syntax for compiling
modules in a particular directory (as noted in the kernel docs in
Documentation/kbuild/modules.txt):
     * Unpack Denis's tarball in a new directory (note that the
       tarballs currently don't contain a root directory, just the
       files, so you want to do the untarring in an empty directory)
     * If you're building for the currently running kernel, build the
       modules with the command
       make -C /lib/modules/`uname -r`/build M=`pwd`
     * Install the modules (must be root for this step, so use
       'su' if that's your preferred method of doing root stuff) with
       make -C /lib/modules/`uname -r`/build M=`pwd` modules_install
I figured that the snapshots might get more testing if more people know
that they don't have to muck around with the whole kernel source in
order to build this. I tested this on Fedora Core 4; sane distributions
should be using this setup for accessing the kernel build files (if they
listened to Linus's suggestions anyways). The one trick is that you
might need a special "kernel development files" package; on Fedora Core,
the files are included in the regular kernel package for FC2 and 3 but
you need the 'kernel-devel' package for FC4.
And finally, the kernel won't figure out that the module has been
installed until you run 'depmod -ae' as root. You can check that
something useful got picked up with 'modinfo'. (Both of these commands
are likely installed in /sbin so you might need the full path unless
you're fully logged in as root.)
The latest snapshot (acx-20050708.tar.bz2) generally seems to work well
for me, except that it's somewhat noisy and reports a lot of DUPs in the
system log. However, it appears to associate more consistently with my
access point that the old versions (when I had those modprobed on boot
and brought up by the system network scripts, the card would often just
sit and scan around and never actually associate until I manually
restarted the connection) - admittedly I've only tried a couple of times
yet so I don't know just how consistent this is. My hardware is a
Netgear WG311v2, using the firmware from the latest Netgear Windows
driver. (Can't check the details right now since I'm not at that
computer, sorry - it got too late when I was mucking with this and other
stuff yesterday for me to get a message sent...)
Cheers,
Per
--
Per Bjornsson <perbj@stanford.edu>
Ph.D. Candidate, Department of Applied Physics, Stanford University