Package: installation-reports

NB. This installation was intended to test the ath5k wireless driver with
d-i, and does not use a "standard" weekly or daily image.

Boot method: netboot
Image version:
Date: Wed Apr 2 20:05:01 BST 2008

Machine: IBM ThinkPad R51e

04:02.0 Ethernet controller: Atheros Communications, Inc. AR5212/AR5213
Multiprocol MAC/baseband processor (rev 01)

Base System Installation Checklist:
[O] = OK, [E] = Error (please elaborate below), [ ] = didn't try it

Initial boot: [O]
Detect network card: [O]
Configure network: [E]
Detect CD: [ ]
Load installer modules: [ ]
Detect hard drives: [ ]
Partition hard drives: [ ]
Install base system: [ ]
Clock/timezone setup: [ ]
User/password setup: [ ]
Install tasks: [ ]
Install boot loader: [ ]
Overall install: [ ]

This installation was intended to test the ath5k wireless driver with d-i:

* netcfg correctly identified the wireless device and did not show any
spurious control devices

* "Automatic" configuration of the device failed

* Entering an ESSID manually resulted in a DHCP failure

* I continued the installation using my wired ethernet adaptor to a point
where I could install wireless-tools-udeb

* "iwconfig wlan0" and "iwgetid" showed that the wlan0 had been configured
to point at the correct access point by BSSID address

* Running:
- iwconfig wlan0 ap off
- ip link set wlan0 up
- iwconfig wlan0 essid my_essid

.. resulted in dhclient working, although there was an interesting
conflict with the existing wired configuration:

DHCPREQUEST on wlan0 to port 67
DHCPDISCOVER on wlan0 to port 67 interval 7
DHCPREQUEST on wlan0 to port 67
ip: RTNETLINK answers: File exists
dhclient.c(2129): null pointer

This process and message looped indefinately until ^C.

* Retracing my steps in the netcfg component did not result in a working
wireless configuration


So, it appears that netcfg does not call 'ap off' (or ipiw equivalent) and
does not set the link up.

< fjp> Hmm. iwconfig wlan0 ap off is basically only to reset the AP for
new scan/association
< fjp> Probably makes sense to do that by default in netcfg for any retry.

I would agree, but would add that running that (or its iwlib equivalent)
before the first attempt might assist drivers that initialise in a strange
way - remember that in my test above, the wireless device was actually
configured with the correct BSSID, but required an reset via "ap off"
before it worked.

However, this and other quirks may be specific to ath5k, a somewhat
experimental driver.


Chris Lamb, UK
GPG: 0x634F9A20

Version: GnuPG v1.4.6 (GNU/Linux)