First of all - I had most problems due to a not working firmware. So try out different versions of Firmware. The one working for me can be found here. Find other firmware links and more documentation under the SpeedLink page.
Also - it is very difficult to actually find decent informations. I found some kind Help on the conexant Maillinglist

Please Note that this FAQ refers to setting up a Link in Germay - using a Annex-B module, however it should also be valid for other Annex-Modules.
Also - as you are installing a quite complex hardware on a complex system - I assume you know how to patch/compile/install kernels and configure a ppp-link - and won't go into details here.

So - let's go for it. Download the conexant usbatm patches for the 2.6.14.x kernels on the project page. Unpack these into the right kernel-source directory:

cd /usr/src/linux
tar jxf ~/usbatm-20050216.tar.bz2
drivers/usb/atm/cxacru.c
drivers/usb/atm/Kbuild
drivers/usb/atm/Kconfig
drivers/usb/atm/usbatm.c
drivers/usb/atm/usbatm.h

Kernels 2.6.14.5 and higher already have the conexant drivers - so don't apply the patch if you use a newer version.
Configure the kernel to have support for the Conexant Chipset.

Device Driver -> Networking Support -> Networking Options ->
 Asynchronous Transfer Mode (ATM) (EXPERIMENTAL)
 Classical IP over ATM (EXPERIMENTAL)
[*] Do NOT send ICMP if no neighbour (EXPERIMENTAL)
 LAN Emulation (LANE) support (EXPERIMENTAL)
 Multi-Protocol Over ATM (MPOA) support (EXPERIMENTAL)
 RFC1483/2684 Bridged protocols
[*] Per-VC IP filter kludge

Device Driver -> USB Support ->
 Support for USB
......
---   USB Host Controller Drivers
   EHCI HCD (USB 2.0) support
[ ]     Full speed ISO transactions (EXPERIMENTAL)
[ ]     Root Hub Transaction Translators (EXPERIMENTAL)
   OHCI HCD support
   UHCI HCD (most Intel and VIA) support
 Connexant Accessrunner USB support

B.e. you can use the following kernel-config for a 2.6.12.x kernel. Copy that file into your kernel-source root directory and perform a make oldconfig. Compile and install the kernel and the modules.
The Firmware is kind of the most tricky part. For this - download the firmware extracter, and compile it

cd /usr/src
mkdir fw-extract
cd fw-extract
wget http://www.elektrorevue.cz/clanky/05041/cxacru-fw-ng.tar.gz
tar xzf cxacru-fw-ng.tar.gz
make

You can also get a local copy of the firmware extractor attached to this FAQ, then download & extract the latest firmware

wget http://www.atlantis-land.com/firmware/1863_DriverWinMultiV099.085WAN.zip
unzip 1863_DriverWinMultiV099.085WAN.zip
mkdir -p /usr/local/lib/firmware
./cxacru-fw CnxEtU.sys /usr/local/lib/firmware/cxacru-fw.bin

as a backup option - check out the latest copy I found - find it in the download section of this site.
Alternatively - take the already extracted Firmware suitable for Linux linked to this site here.
After that - find out where your hotplug system looks for firmware. Mine checks under /usr/lib/hotplug/firmware - and that's where I'll copy the firmware.

cp cxacru-fw.bin /usr/lib/hotplug/firmware

This done reboot your system and check you have informations about your modem. The following message should show up during the boot process:

device class 'firmware': registering
bus usb: add driver cxacru
bound device '2-2:1.0' to driver 'cxacru'
usbcore: registered new driver cxacru
usb 1-1: new full speed USB device using uhci_hcd and address 3
cxacru 2-2:1.0: found firmware cxacru-fw.bin
cxacru 2-2:1.0: ADSL line: attemtping to activate
cxacru 2-2:1.0: ADSL line: down
cxacru 2-2:1.0: ADSL line: attemtping to activate
cxacru 2-2:1.0: ADSL line: channel analysis
cxacru 2-2:1.0: ADSL line: up (7168 kb/s down | 640 kb/s up)

Checking the proc-filesystem for the proper devices to exist - you should find:

cat /proc/net/atm/cxacru:0
ADSL USB MODEM (usb-0000:00:07.3-2)
MAC: 00:05:b4:12:57:68
AAL5: tx 38291 ( 0 err ), rx 58395 ( 0 err, 0 drop )
Line up

This shows - the modem is up and has a Link. Now - as I'm using PPPoE - I need to make sure the right modules are loaded for the br2684 module which enables a bridging connection with br2684ctl to run PPPoE over ATM.

modprobe br2684 

Make sure it is loaded at boot-time. On a Debian-System - add the line to the file /etc/modules

echo br2684 >> /etc/modules

Install the br2684ctl tools. Under debian:

apt-get install br2684ctl 

will do the trick. Find out what VPI/VCI your provider is using. In Germany - it's VPI=1 and VCI=32 at 1&1. To activate it - I wrote the following little script br2684_up.sh:

#!/bin/sh

PATH=/usr/local/sbin:/usr/sbin:/sbin:/usr/local/bin:/usr/bin:/bin
export PATH

 # Build a little while loop - to check if the conexant modem
 # has a ADSL Line up
 count=0
 while [ $count -lt 40 ]
 do
   sync=`cat /proc/net/atm/cxacru:0 | grep 'Line up'`
   [ "$sync" = "Line up" ] && break
   sleep 1
   count=$((1+$count))
 done

if [ "$sync" = "Line up" ]
 then
  br2684ctl -b -c 0 -a 1.32
  sleep 3
  ifconfig nas0 172.16.2.6 netmask 255.255.255.248
fi

The while loop waits at least 40seconds until the cxacru-driver has synced to the telco - after that it will set up the bridge-interface to the ATM-device - which can be used by the pppd daemon.
Using the following file in /etc/ppp/peers/dsl-connection

noipdefault
defaultroute
user "UserName@domain.de"
noauth
updetach
usepeerdns
plugin rp-pppoe.so
nas0

hide-password
# Override any connect script that may have been set in /etc/ppp/options.
persist
mtu 1492

# RFC 2516, paragraph 7 mandates that the following options MUST NOT be
# requested and MUST be rejected if requested by the peer:
# Address-and-Control-Field-Compression (ACFC)
noaccomp
# Asynchronous-Control-Character-Map (ACCM)
default-asyncmap

I can call the connection using: pppd call dsl-connection
Note that you'll have to configure quite more stuff, especially to make the connection come up at boot time. On my debian system - I have copied the br2684_up.sh script to /etc/ppp, configured the ppp-connection using the pppoeconf package - and adapted the /etc/network/interface file to reflect the correct pppd-call. Make sure the br2684ctl is loaded at boot before the ppp is started.

auto dsl-connection
iface dsl-connection inet ppp
     provider dsl-connection
     pre-up /etc/ppp/br2684_up.sh
# please do not modify the following line
     pre-up /sbin/ifconfig nas0 up # line maintained by pppoeconf

 Get the newest snapshot from the CVS repository under: http://at76c503a.berlios.de/cvs.html
Unpack the archive in /usr/src/ - assuming you have already compiled the kernel version 2.6.12.4-EpiaV (Adapted in the Makefile of the kernel-source directory)

tar xzvf at76c503a-cvsroot.tar.gz
mv at76c503a CVS
cvs -d `pwd`/CVS co at76c503a
cd at76c503a
make KERNEL_VERSION=2.6.12.4-EpiaV
make KERNEL_VERSION=2.6.12.4-EpiaV install
/sbin/depmod -ae -F System.map  2.6.12.4-EpiaV

Note that you don' t need to specify all kernel-related version stuff when already running the kernel you compiled the driver against.
I have found out that there are several Firmware Versions available for this WLan Module - and I still don't know where I got the latest version - but in case someone looks for it - check it out firmware version 1.103.2 #175.
Please note that I tested it with USB device ID: 03eb:7605 only.

Note that some people have reported version 1.103.2 not to work on their hardware. In that case - use firmware version 1.101.0 #84. This version seems to work almost everywhere.  

 Actually - the lm_sensors 2.7.x package performs a quite good identification of the existing hardware. On my system - the following modules need to be loaded using a 2.6.12 kernel - in the /etc/sysconfig/lm_sensors:

MODULE_0=i2c-viapro
MODULE_1=i2c-isa
MODULE_3=eeprom
MODULE_4=via686a

Make sure you use the ISA Bus, as the SMBus uses the lm80 Chip/Module which provides only Junk data. Also - the /etc/sensors.conf file looks like this:

chip "via686a-*"
    label "1.5V" "CPU core"
    label "2.5V" "+2.5V"
    label "3.3V" "I/O"
    label "5.0V" "+5.0V"
    label "12V" "+12V"
    ignore fan1  #"CPU Fan"
    ignore fan2  #"P/S Fan"
    label temp1 "CPU Temp"
    label temp2 "SYS Temp"
    ignore temp3 #"SBr Temp"
    set temp1_hyst 50
    set temp1_over 55
    set temp2_hyst 55
    set temp2_over 60
    set temp3_hyst 60
    set temp3_over 65

That provides me decent data on the System. Note that the mini-ITX-Lex system is passive cooled - so no FANs. After creating the sensors.conf file - run "sensors -s" and you should be done. 

 The Lex 860A EPIA 5000 board uses the 533MHz C3 CPU. By loading the longhaul module - you can throttle the CPU. Note that it became stable with kernel 2.6.11.11 only on my system.
So - all you need to do is enable the longhaul by loading the longhaul module:
modprobe longhaul
Check using the dmesg command to see if it loaded correctly.
If this works - make sure the longhaul module gets loaded at every boot. For this - just add "longhaul" and "cpufreq_powersave" into the /etc/modules file.
Using the longhaul stuff - make sure you use under a 2.6.12.4 kernel-module (as configured in config-2.6.12.4) with the userspace-governor eabled by default - then change to:

# cd /sys/devices/system/cpu/cpu0/cpufreq
cpufreq # ls
affected_cpus
scaling_available_frequencies
scaling_governor
cpuinfo_cur_freq
scaling_available_governors
scaling_max_freq
cpuinfo_max_freq
scaling_cur_freq
scaling_min_freq
cpuinfo_min_freq
scaling_driver
scaling_setspeed
cpufreq # modprobe cpufreq_powersave
cpufreq # cat scaling_available_governors
powersave userspace
echo powersave > scaling_governor
cpufreq # cat scaling_cur_freq
399000

That's it ...  

 First of all - you need to realize that the CV860A has a Via C3 Ezra CPU which is a 486 Based CPU, with some Pentium extension. So most things will work.
The Installation will make no problems - however - as soon as you reboot the System - it will crash badly - especially if you have a USB Wireless Lancard using the Amtel Chipset (this was my case).
I found out - that the usb subsystem was the cause of the problems - so deactivating it on boot appending: nousb to the lilo-command line - did the trick. What was required to do - was to deactivate the usb-uhci driver loading if the modules for the Wireless Interface was to be loaded (Which was the cause of the problem) and use the uhci module. Do this with changing the probeall USB line in the etc/modules.conf file to:

# Wireless Lan stuff
probeall usb-interface uhci
post-install usbvnetr /bin/fastvnet.sh

This will fix the Crash. 

 Here' a picture downloaded from the linitx-page


You can get it a http://www.linitx.com
This little box is a perfect home-server for Internet access, can also be used as a very nice firewall.
Technical details:

 Dimensions: 170x50x225mm
 VIA 533 MHz cpu - No Fan!
 CV860A Motherboard
 512MBYtes 168 pin DIMM SDRAM
 2.5inch IDE 40GBytes
 Onboard Compact Flash adaptor (unused)
 Onboard DiskOnChip socket
 3 10/100 Realtek 8139 LAN ports
 2 rear USB ports
 1 front USB port
 Mouse, Keyboard, VGA,2 Serial, Parallel sockets
 Onboard power supply
 External 12v power brick

 The symptoms are always the same:

• HD LED steadily one
• HD harddrive not accessible (connected on IDE 2)
• rest of system keeps running (network devices ok, graphics & sound ok)

The following error-message will usually occure:

hdc: DMA timeout error
hdc: dma timeout error: status=0xd0 { Busy }
ide: failed opcode was: unknown
hdc: DMA disabled
ide1: reset: success
hdc: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
hdc: drive_cmd: error=0x04 { DriveStatusError }
ide: failed opcode was: 0xb0

This is mostly due to the harddisk overheating. Having a Lex-System at home - the harddisk is mounted over a chip that gets rather hot - circled in red in the following Picture:

Sometimes the system won' t recover from such thing - thus hanging and not beeing able to actually recover to access-mode on the disk. If you're not on site to actually get that issue fixed on hot days - set the CPU speed to 400MHz, FSB to 66MHz and make sure the harddrive does not get too hot.
It seems - the smart-capability of the HD did deactivate the Harddisk to protect it from damage - due to overheating.
Here is what I run to actually stabilise it:

service smartd stop
chkconfig --del smartd
smartctl --smart=off --offlineauto=off --saveauto=on /dev/hdc
hdparm -d1 -Xudma2 /dev/hda

Another thing that was found on the Web - under http://www.directron.com/faqviaas.html

I am having trouble with a system using a VIA chipset including a 686b southbridge and a Creative Sound Blaster Live card.

When the 686b southbridge was first released, motherboard manufacturers and VIA discovered a problem when trying to transfer files between the primary and the secondary IDE channels on motherboards using ultra-DMA, when a Sound Blaster Live was plugged in and drivers activated. This issue was caused because too much noise is transferred across the PCI bus by the Sound Blaster Live driver set. In an attempt to fix this issue, some motherboard manufacturers modified their BIOS. In some instances, these modifications to the BIOS caused a data corruption error even when not using a Sound Blaster Live. VIA released a patch which resolves this issue, which is incoroporated in the 4in1 drivers from the 4.31 version onwards. Motherboard manufacturers were advised to change the modifications made to thier BIOS to elimate the data corruption issue. If you are experiencing data corruption or lock up when transferring files between two IDE drives:

• Make sure you have the latest BIOS from your motherboard manufacturer
• Make sure you have the latest 4in1 drivers
• Make sure when you set up your system that you install the 4in1 drivers both before and after you install the SBL to make sure that the drivers see your SBL and install the correct patch. The patch will only install if the SBL is installed.

Discussions about these issues can be found:
a. Via-Arena
b. Symaptico Epia Forum
c. Tryphon
d. Kernel Maillinglist  

 These kernel-configurations are using the plain unpatched kernels as found on kernel.org.
The system these are running on - is a Lex 860A Epia 5000 V.
Please note that if you have the Wireless modules - you need to compile the modules externally. Check the Related Faq Link at the bottom of this FAQ. 

 VIA VT8601A & VT82C686B EDEN on board CPU



Check the Manufacturer homepage details CV860A Board Specs