Skip navigation.

exploreopera

| Help

Sign up | Help

My Wi-Fi on Linux - The Empire Strikes Back

, , , ...

Guess what? The legacy x11-drivers/nvidia-drivers-96.43.05 fails to compile on 2.6.26.

Let me try the version that I have on my laptop, which is 169.12... Huh? It fails to compile too.

Maybe I have better luck with the latest stable version on gentoo, 173.14.09. You should remember that 173.14.09 was completely messed up on my laptop, but maybe I have a better luck on my desktop.

Hum... 173.14.09 compiles fine. Let me reboot with the new kernel and try it. No, wait! *Almost* forgot to update /boot/grub/grub.conf (menu.lst on some systems; on mine, one is a symlink to the other). How many of you have rebooted a system but forgot to update the boot manager? :D

Rebooting... It boots! Let me try X.org and this nvidia driver... Good, it appears to work, at least for now. I need to keep an eye on it. We never know when it will crash.

And... does the Wi-Fi card work as I want it?

Answer: I don't know yet! :D

I can confirm that I can put the card on ad-hoc mode, and that wpa_supplicant tools on my notebook can detect a new wi-fi network that uses IBSS. This is much better than my previous try. I can finally feel that things might start working!

However, I could not associate with that network yet, and I'm not sure why. Since I am in a hurry, I can't investigate it yet.

Will it work for real? Will it work in ad-hoc mode? Will it work in the very experimental master mode? Don't miss the answer to these questions, and also the answer to life, the universe, and everything in my next blog posts!

My Wi-Fi on Linux - A New Hope

, , ,

People who read my blog probably remember my previous attempts at setting up my own home wireless network, transforming my desktop computer into a wireless access point. My most recent try was using a Ralink RT2561 PCI card, with rt61pci driver from rt2x00 driver family. If you remember, that card was supposed to be Linux-compatible and so on, but the truth wasn't exactly what we expected.

The problem with that card was that the drivers did not support either ad-hoc or master modes, making it completely useless for my needs.

However... Things change... Drivers (hopefully) get improved...

A few minutes ago I found a forum post from the beginning of the year that says "Adhoc and master mode are scheduled for 2.6.26". Then I found that experimental drivers are available at Linux Wireless site. These under-development drivers are backported to kernel versions 2.6.21 and newer.

Well... If they are scheduled for 2.6.26, and I'm already using 2.6.25.x, it shouldn't take too long, right? What is the latest available stable kernel? Woohoo! 2.6.26 was released a few days ago! :hat:

Linux Wireless homepage/news page states that "Access point mode has received a lot of attention in the stack and drivers, but is not ready yet.". This statement is also supported by the todo-list: "AP support is still unfinished.". In addition, the 2.6.26 changelog says this:
commit adfdbb79c06154cd3cc7b5983106ace324aa3b02
Author: Ivo van Doorn <ivdoorn@gmail.com>
Date: Sun Feb 3 15:48:03 2008 +0100

rt2x00: Enable master and adhoc mode again

This will enable the creation of master mode and adhoc
interfaces again. This does not mean the issues surrounding
beaconing have been resolved, but this will make testing
easier and perhaps we can discover which cards are actually
working and which ones not.

So, it seems that AP and ad-hoc modes are still very experimental, but they might work. Even though they might not be production-ready, this means they are being actively developed, and sooner or later they will be ready. Hopefully sooner.

All of this are really good news. There is a new hope that my home wi-fi network will stop being just blog posts and start being a real working network. And, what's more, without buying another card!

It appears to be even more feasible when we start seeing forum posts of people talking about their success [1] [2].

nVidia (driver) strikes back yet another time

, , , ...

As usual, I'm here again to blog about... guess what? Problems with nvidia driver.

Right now I have an Asus M51Sn notebook, which has nvidia GeForce 9500M GS card. I'm using 64-bit Gentoo/Linux with vanilla kernel 2.6.25.5. At the time I installed it, the 2.6.25.x versions were not marked as stable in Gentoo, but I chose to install it anyway. However, the then stable x11-drivers/nvidia-drivers-169.09-r1 did not compile with those kernel versions (bug 218178).

Solution? Well, I just installed the next version (x11-drivers/nvidia-drivers-169.12) and it worked pretty well.

In fact, it worked very well until this week, when x11-drivers/nvidia-drivers-173.14.09 was marked stable (but 169.12 was still ~testing). After this update, my 3D desktop simply stopped working. Well, it starts, I can start one app or two, but things are frozen or almost frozen in a weird way that it is simply not usable.

Solution? I'm going back to 169.12 version.

This comment on bug 218178 was a bit scary for me:
Which is why you use the latest stable version, 173.14.09 which is designed for
2.6.25.

What does that mean? Does that mean that 169.x series were not designed for 2.6.25? Or just that 169.09 was not designed, but 169.12 is ok (because it was left "undefined" in his comment)?

Well, anyway I'm masking 173.14.09. I'm going to stick with 169.12, at least this version works for me... until the next nvidia strike.

My system, my notebook - xkcd

, , ,

xkcd comic rocks! At least for geeks like me us.

Read more...

July 2008
SMTWTFS
June 2008August 2008
12345
6789101112
13141516171819
20212223242526
2728293031