Bug 378000

The answers to the questions:

a) I don't know, I don't even know who decides that.
b) NM gets it's device list from HAL, it doesn't poke around the pci and usb buses to find devices that might exist but for which the driver/firmware isn't installed.
c) look (b).

There's absolutely nothing NM can do here, please open two different bugs instead of this one: 1) "iwl3945-ucode not installed by default" and 2) "HAL should offer to help install drivers".

So - we should work out whom to re-assign this to then, rather than closing invalid. Who owns HAL ? is that Kay.

Kay - any idea ? or is this a simple broken pattern problem ?

Looks like something to solve at the package selection level. The ucode should probably contain a match, the same way like some kmp packages are selected based on currently found hardware.

all done already

The .spec file has:
Supplements: modalias(pci:v00008086d0000422[27]sv*sd*bc*sc*i*)

which is handled by the installer/libzypp.

was the card already in the machine as you installed the system?

Sure; it's a built-in card. Of course; it's possible this is some 'purists' thing - I installed from an Alpha 1(?) DVD - is it the case that we avoid installing proprietary(?) microcode in these cases & thus break usability (or even testing of default patterns ;-). ?

if the package is on this DVD, the solver should have picked up the Supplements tag actually.

I had my wifi working on my Thinkpad T60 (using the iwl3945) under alpha 3 however my wifi did not work under v11 beta 1 (I used the B 1 DVD + added factory sources on SUN  APRIL 2008 when I installed)

v11 
# initial lsmod results post install
jpearson@linux:~/.kde/share/config>  lsmod | grep 3945
iwl3945               106332  0
firmware_class         25984  3 microcode,pcmcia,iwl3945
mac80211              190356  1 iwl3945


Here's what I did to get wifi iwl3945 working
# 1
I downloaded compat-wireless-2.6.tar.bz2  from
http://linuxwireless.org/en/users/Download#DownloadlatestLinuxwirelessdrivers

Note 1 :compat-wireless-2.6.tar.bz2 
This package is updated daily. It reflects the latest on wireless-testing.git tree.

# 2 
Did a make, checkinstall and smart install


Results 
Wifi works perfectly after installing compat-wireless-2.6 drivers 
 when I select it from KNetworkManager

Note 2 
Note the new line below on my lsmod output
which was absent on the initial install
cfg80211               43784  2 iwl3945,mac80211


# Final lsmod results
jpearson@linux:~/.kde/share/config> lsmod | grep 3945
iwl3945               108504  0
firmware_class         25984  3 microcode,iwl3945,pcmcia
mac80211              238820  1 iwl3945
cfg80211               43784  2 iwl3945,mac80211

what is the state of this bug with beta1/beta2? Does the given modalias match your hardware?

Not quite sure what to state really; if I do:

$ cat /sys/devices/pci0000\:00/0000\:00\:1c.1/0000\:03\:00.0/modalias 
pci:v00008086d00004227sv00008086sd00001011bc02sc80i00

$ lspci | grep Wireless
03:00.0 Network controller: Intel Corporation PRO/Wireless 3945ABG Network Connection (rev 02)

So - I assume yes, the modalias matches my hardware.

DVD1 contains the package in Beta2:
./DVD1/suse/noarch/iwl3945-ucode-2.14.1.5-45.noarch.rpm

So, this should work with Beta2 - can you retest?

My wifi connection works fine on my Intel Corporation PRO/Wireless 3945ABG card (using
iwl3945) on Beta2 once I boot and I select my named (and configured) WPA connection via
KNetworkerManager wifi buton on the kde panel This is good news...  (This was *** not *** the case using Beta1)

However despite the fact that my wifi is up and works OK konqueror in the Network Status section of
the "sysinfo:/ page  shows

Unkown network status

I put a photo at 
http://pagesperso-orange.fr/j.pearson/T60.B2.png


wifi still does not come up automatically on boot (despite the fact that is configured to
do so). I gather that this (See message "Couldn't get connection
secrets: dbus-settings.c.108 - Connection's plugin did not return a secrets hash") is a
known problem which is caused by the way netmanager interacts with opensuse boot
configuration settings. 

Here is what I see on my machine with regards to that...

# netmanager log post boot
May  3 14:26:25 linux NetworkManager: <info>  Trying to start the system settings daemon...
May  3 14:26:29 linux NetworkManager: <info>  Bringing up device eth0
May  3 14:26:29 linux NetworkManager: <info>  Deactivating device eth0.
May  3 14:26:29 linux NetworkManager: <info>  Bringing up device wlan0
May  3 14:26:29 linux NetworkManager: <info>  Deactivating device wlan0.
May  3 14:26:30 linux NetworkManager: <info>  (wlan0) supplicant interface is now in state
2 (from 1). May  3 14:26:32 linux NetworkManager: <info>  Activation (wlan0) starting
connection 'System Mercury (wlan0)' 
May  3 14:26:32 linux NetworkManager: <info>
Activation (wlan0) Stage 1 of 5 (Device Prepare) scheduled... 
May  3 14:26:32 linux
NetworkManager: <info>  Activation (wlan0) Stage 1 of 5 (Device Prepare) started... 
May  3 14:26:32 linux NetworkManager: <info>  Activation (wlan0) Stage 2 of 5 (Device
Configure) scheduled... 
May  3 14:26:32 linux NetworkManager: <info>  Activation (wlan0) Stage 1 of 5
(Device Prepare) complete. 
May  3 14:26:32 linux NetworkManager: <info>  Activation (wlan0)
Stage 2 of 5 (Device Configure) starting... 
May  3 14:26:32 linux NetworkManager: <info>
Activation (wlan0/wireless): access point 'System Mercury (wlan0)' has security, but
secrets are required. 
May  3 14:26:32 linux NetworkManager: <info>  Activation (wlan0)
Stage 2 of 5 (Device Configure) complete. 
May  3 14:26:32 linux NetworkManager: <WARN>
get_secrets_cb(): Couldn't get connection secrets: dbus-settings.c.108 - Connection's
plugin did not return a secrets hash.. 
May  3 14:26:32 linux NetworkManager: <info>
Activation (wlan0) failed for access point (Mercury) 
May  3 14:26:32 linux NetworkManager:
<info>  Marking connection 'System Mercury (wlan0)' invalid. 
May  3 14:26:32 linux
NetworkManager: <info>  Activation (wlan0) failed. 
May  3 14:26:32 linux NetworkManager:
<info>  Deactivating device wlan0.

sounds like this bug (the install problem) went away at least & can be closed.