I’m really proud of this great machine. In addition i also ordered a set of travel bags and a special waterproof attachment for a navigation system. I can’t wait to use it for travelling to my new employer in Bautzen Germany. As special navigation systems (in contrary to those for normal cars) for motorbikes are very expensive I’m happy I found this special mounting kit.

Before i only had a motor scooter for travelling inside Bautzen, but of course this cannot compare to a real motorbike. Hopefully i’ll soon also can post some travellogs. So, let the good times roll….

Sadly its still a no go for linux users. It still has the problem the netbook cannot suspend anymore after the update. The only good working BIOS Version is still only 1.09 (which i can give you  on request).

009: My Wind U110 was delivered with this version and all that was wrong was that i noticec that BIOS reported false possible CPU C-States when on battery power (just C2 instead of C6 that cpu supports). But rebooting without AC adapter “solved” it.

10D: This version i downloaded as update from MSIs homepage and after some struggles to update it (finally FirmwareUpdateKit made it real easy … fuk –image /dev/sdb /tmp/biosfiles/*) i had it on my machine… and guess what… it was really disappointing. Eventhough msi fixed the C-State problem, the netbook now refused to suspend!!!

Eventhough all /sys/devices/*/power/wakeup were disabled it now wakes up immediatelly or hangs hard. Before (with version 009) it suspended fine (with just “s2ram -f”) and even had wlan (via rt2860sta module) working still fine afterwards. But now it was impossible to suspend anymore… which makes daily use of netbook really horrible imho.

So i contacted msi and after the first bothering tips if i really updated correctly they provided me the older BIOS version

10C: But this one behaved exactly like 10D in regard to suspend. So after some more mails with msi they sent me the “very” old 009 plus the version 109 and that one is interesting.

109: Here suspend works again and also C-States are reported correctly.. so all could be fine… BUT it wasn’t in all detail: Now WLAN/rt2860sta doesn’t work anymore after suspend and resume. And simply un- and reloading the module sadfully also didn’t help. There were just countless lines like this in the kernel log:

ERROR!!! BBP(viaMCU=1) read R0 fail

After some search in the net i found that this again seems to be BIOS bug happening when the wlan-module is powered off. But there is a ugly workaround for this problem on my machine.

1. make sure pciehp module is loaded with options pciehp_force=1 (which again points to a buggy bios) and also pciehp_poll_mode=1  (without this second parameter the system didn’t seem to be stable after suspend for me)

2. configure pm-utils to stop network on suspend and restart it on resume (otherwise the module cannot be unloaded) and also to unload and reload the module rt2860sta. I further changed the /etc/init.d/network script to also kill wpa_supplicant on stop action so really nobody anymore accesses the device.

If you need help for those steps feel free to contact me.

Nearly everything else works great. LAN support is a bit funny. The ethernetcard sits on the usb bus and only activates itself as soon as a cable got plugged in. Then a usb asix device pops up and works great (with the asix module).
Sound needed “options snd_hda_intel model=5stack-digout” in /etc/modprobe.d/sound to work under 11.1.
WLAN is a Ralink RT2860 which works with the module rt2860sta. For 11.1 there are kmps in the buildservice, but they never worked for me. I compiled the sources from the ralink homepage and noticed that the driver works well until udev renames the device from (originally) ra0 to wlan0. Afterwards it does not respond anymore.
The fix that worked best for me was to compile the rt2860sta version from kernel 2.6.30 (in staging tree) for openSUSE 11.1. To make it easier to use this i put it here (yes, i’m too lazy to build a kmp .

P.S.: Actually to make wlan work with the packages from above repository from alex you have to blacklist rt2800pci.

P.P.S.: I also recommend to update this machine to the latest BIOS version available from MSI (currently Version 1.0D). As only with this i get fully low-power C-States correctly. To install it i used the package UpdateFirmwareKit and a normal USB-Stick.
P.P.P.S: No, you better don’t update your BIOS to Version 1.0D. It seems this version totally breaks suspend on this machine. I couldn’t find a s2ram parameter combination still working (before just -f worked). Not even with current 2.6.30 it works. I contacted MSI about this and hope they provide a solution for this problem. … Meanwhile as a first response they sent me BIOS Version 1.0C, but that also has the suspend problem. Hopefully they either fixup their BIOS or provide me the working 0.9 Version.

APN: web.vodafone.de
Username: vodafone
Password: vodafone

Just for completeness the 1und1 special configs are:

APN: mail.partner.de
Username: D2
Password: Web

You can also find all those data on the providers homepages, but you have to dig quite deep until you find them. So hopefully this is of use for some future users.

acpi=off

as bootparameter while installing. This is due to a totally broken video BIOS of the machine bringing the machine to hang on loading the video module. Later on you can just add blacklist video to /etc/modprobe.d/blacklist to avoid its loaded and then remove the acpi=off bootparameter.

Second ugly thing is they broke brightness support in their BIOS and so you have to _down_grade your bios version in order to get it working correctly (and i haven’t encountered anything bad due to the downgrade). I installed Bios Version 08ST from http://www.samsungpc.com/products/q45/q45_bios.htm. In order for the downgrade to work you have to add those additional parameters to the firmware programming call:

/s /x /c /nomodel_partcheck

so e.g. the call will look like this > 22sd.exe /s /x /c /nomodel_partcheck

After this is done the brightness keys will work, well at least they did for me and many others. But please be aware that doing this is on your own risk (as BIOS flashing is always a bit dangerous). But now in with those changes (older BIOS and video module blacklisted) the machine is very nicely usable with openSuSE. Suspend to ram and to disk work out of the box. There is a minor glitch with the harddisk doing too many load cycles, but this can also easily be workarounded by installing the package storage-fixup and adding this to /etc/storage-fixup.conf:

# Reported drive model: Toshiba MK2546GSX
rule samsung-q45
dmi system-manufacturer         SAMSUNG ELECTRONICS CO., LTD.
dmi system-product-name         SQ45S70S
ata model                       TOSHIBA MK*46GSX
act hdparm -B 254 $DEV

I reported this upstream and this change was accepted so it will soon be included (in future releases).

After it now bothered me enough to take a closer look it seems t-online did it first and now vodafone does the same (proxying the images to much lower quality level). But at least for the firefox browser there is a workaround i found (besides reloading every page) that has no noticeable speed downside on my system but brings me full joy of websurfing back:

1. Install “Modify Headers” add-on for firefox
2. In “Extras” or “Tools” menu go on “modify headers”
3. at the top select “add” (from dropdown menu), “Pragma”, “no-transform” (each without the quotes) and click the Add-Button next to it
4. at the top select “add”, “Cache-Control”, “no-transform” and again click the Add-Button
5. You should now have two rule entries each activated (a big green dot on their right side)
6. close that window
7. enjoy surfing

After a nice week here in Ottawa/Canada full of interesting talks, presentations, tutorials and people OLS today sadfully already closes. Definitely a conference you should consider visiting if you are interested into linux and want to take part in discussions where and how to move forward. Especially the talks from James Bottomley and Mark Shuttleworth were very remarkable. Although i somehow doubt Marks idea of synchronising all major distros is doable or even worth the effort, it was nevertheless inspiring of how to break rules that are only in our minds.

Seems we need much more testing in this area. Anybody volunteering to provide me some more hardware of this area so i can make sure it works in future? (e.g. bluetooth stereo headset etc.)

I have to admit until now I hadn’t much to do with bluetooth at all, but after a salesman here in Ottawa (here for OLS) advertised me this thing would work nicely under linux.. i somehow expected we’d do nice here, but to my frustration i found we have plenty of room for improvement here.

To make it short, after a lot of hours going through hundreds all different howtos and documentation i really was desperate i never would get this thing workin… when i finally realized our packages are broken in 11.0. Or at least bluez-audio which contains libaudio.so that seems to segfault immediatelly after connecting to the device . Fortunatelly the latest factory rpms have this fixed and also works on a 11.0 system (i opened a bug report for this meanwhile, perhaps you wanna place a vote for this.. bnc#412464).

Download them either from official factory repo or from seifes home buildservice project.

Please be aware that i am absolutely no bluetooth expert and all this might just be crap and i was just lucky this worked by chance for me, but perhaps this is also helpful for some more (but try on your own risks..).

After you upgraded your bluez* packages and installed bluez-audio you can make the headphone functionality make work with two simple files to create, but that solution won’t work for simultanious voice transfer for current voip softphones (like e.g. ekiga, kcall, xten etc.).

But if thats enough just:

1. Install new bluez* packages (including bluez-audio) from above
2. as root create the file /etc/bluetooth/audio.service with this content:
[Bluetooth Service]
Identifier=audio
Name=Audio service
Description=Bluetooth Audio service
Autostart=true
3. as normal user create the file ~/.asoundrc (yes, really in your home directory) with this content:
   pcm.bluetooth {
type bluetooth
device "00:00:00:00:00:00"
profile "auto"
}
   (with the address after device being that of your bluetooth headset; findable e.g. via kbluetooth or with hcitool scan on pairing)
4. restart bluetooth service (rcbluetooth restart as root)

From then on you have new alsa device called “bluetooth” many newer audio-/multimediaplayers can acces (e.g. for mplayer: mplayer -ao alsa:device=bluetooth ….).

Sadfully for e.g. ekiga and other softphones this does not work. There we have to use/ i had to use the old snd-bt-sco module so i get a new additionall complete sound device. When back in germany or when time permits i’ll package this so this will just be simple kmp package to install, but for now heres a tar ball with the fixed sources (so they compile with a 11.0 kernel).

1. Deactivate above method by setting Autostart to false in /etc/bluetooth/audio.service
2. Download and extract the tar ball fseidels-btsco-05a.tar.bz2
3. cd into the new dir and issue (as normal user) “./configure“
4. then “make“
5. and as root in that directory a “make install“
6. if not already done install kernel-source package
7. cd /usr/src/linux
8. as root “make cloneconfig“
9. as root “make prepare“
10. cd back to the btsco directoy and there cd into its kernel subdirectoy
11. issue a “make“
12. copy the snd_bt_sco.ko module to /lib/modules/$(uname -r)/weak-updates/ (as root)
13. issue a “depmod -a” (as root)

To start/use it

1. restart bluetooth service
2. pair your headset (e.g. with help from kbluetooth)
3. issue “hciconfig hci0 voice 0x0060” as root
4. issue a modprobe snd_bt_sco as root
5. start “btsco 00:00:00:00:00” (as root; with the address parameter being that of your headset)
6. chmod 666 /dev/dsp1 (as root)
7. chmod 666 /dev/mixer1 (as root)
8. as root issue “alsamixer -c 1” to adjust volume settings

now you have a workin new /dev/dsp1 sound device you can set in the audio settings of ekiga and other voip softphones and use it like expected.

Hope this helps anybody and hopefully i will soon find the time to package this up as decent kmp with automated udev-rules so it gets easier to use.