Thanks for the information guys


Bip

After a somewhat longer 'test-time': I found occasional and badly-tracable problems with the built-in wired network 'card', combined with the 'kernel.org' r8169 driver. There are two problems:

1) the rx-dropped register is read as 32 bits, but the built-in card has only a 24-bits register. Nothing serious, but the rx-dropped 'entry' in any ifconfig result will look very odd... The upper 8 bits are sometimes reported as ono-zero, and that gives an immense (reported..) number of rx-dropped packets.

2) sometimes, the card/driver pretends it is initialized into 1000Mbps mode. Seems that the actual card does not really support 1Gbps mode, and the result is that the wired network does not function. unloading/insmoding the driver does not seem to help. The symptoms are that ethtool eth0 reports a link speed of 1Gbps and that a lspci reports an unknown version 0xff for the device.

I first suspected dhcpcd, but that (sorry Gerard... :-) ) was a dead end: the problem showed up as 'no IP address from dhcp server', but the real problem was that the NIC itself was not working good (problem #2 of the list above..)...

At the moment, i'm test-driving with the 'kernel.org' r8169 driver blacklisted, and the driver from realtek (r8101) installed. So far, this seems to work reliably. But: it is an erratic and rare problem, so some long-term testing will be needed.

I've 'stolen' the driver source code from the url provided by archlinux:

ftp://202.65.194.211/cn/nic/r8101-1.009.00.tar.bz2

It compiles almost cleanly with my 2.6.25.4 (current) kernel.

Hmm, just saw that there is also a 1.011.00 version of the driver source; i may switch to that one still.



[edit] After over a week with the 1.009.00 version of this driver, i have not a single problem with it to report. This driver seems to work 100% 'ok' for my machine and environment.

Another status update: I've noticed very occasional problems with xorg/video. 'the thing' has happened 4 times all-in-all. What happens is the following: the screen goes 'artistic': multi-colored stripes, no mouse cursor, no response to caps-lock (led..), idem numlock (idem led...). No response to [CTRL]-[ALT]-[backspace]... All-in-all: a fully unusable desktop. The issue is known at 'www.x.org', and seems actively being tracked down there. One 'tip' is to add:

Option "ExaNoComposite" "true"

to xorg like this:



(http://bugs.freedesktop.org/show_bug.cgi?id=17638)

I'm now test-running with this switch..

Update (29-jan-2009): I just got 'another one'. This 'switch' is not enough to totally ban this problem. I've disabled it again... Now i'm urgently awaiting an update of Xorg.... We shall see

Oh; the one 'side effect' i did notice of this switch is that the 'post-edit' / 'new post' www-page of the zenwalk forum becomes *very* slow: the animated gif-fies (smileys) at the left of the screen 'chew' the full CPU. In konqueror, i can 'bring peace' by right-clicking 'stop animations'. I have seen the same effect on my kde-4 buildbox (totally different hardware..), probably i'll have to see if i can do something about the video-accel settings there too to bring joy to running kde4...

Note: the machine has no workable gui when this happens, but it still *is* ssh-able, so it's quite possible to reboot it cleanly by taking control of it 'from the net'... My earlier 'power-button' reboots are not really needed. Still: it's not nice that you need another machine to 'bring down' the netbook. That does dent the portability of this thing 'a bit'

Not really unexpected, i'd think: the WiFi driver seems 'on the way' into the kernel proper...

http://git.kernel.org/?p=linux/kernel/g ... 320a1bf2a5

[edit]

I don't find this worthy of a new 'post', but: [Fn]-[F3] disables the 'tickle-pad' I never did have any love for it (i use an USB-mouse....), and the thing was most annoing when it triggered a 'cursor-quantum-jump' because some part of my hand had 'triggered' the pad while i was typing... If now i could find a way to disable the thing when an USB mouse is 'inserted' i'd be perfectly happy with the 'pad'.

Another one: the 'good-folk' at Realtek (?) have put a password on the site where i'd downloaded the non-kernel.org driver for the wired network I have copies of the 1.0....9 and 1.0....11 drivers sources, so pm me if you need them Alternative: have a good look at the driver source filename & google for it: there are a few places where it can be downloaded from.

On Zenwalk 5.2, I made the wireless work by compiling a driver from Realtek, rtl8187se_linux_26.1023.1118.2008

I upgraded to Zenwalk 6.0, and now, the driver gives errors when compiling.
Does anyone have experience with Zenwalk 6.0 on the MSI Wind U90, or does someone have a solution? Thanks.

Do you have kernelsource installed? On 5.4beta i didn't need kernelsource to compile my ralink wireless driver,,but in 6 I need it.

In which package is the kernel-source?

In the kernelsource package

http://packages.zenwalk.org/?p=kernelso ... on=current

#1: my 'U90' is -for the moment- locked to 5.2: There are a few projects installed on it that will not bear upgrading: custom-built packages (opencv, ffmpeg) for which i do not want to risk making them unusable. That -i expect- will change in the next two weeks, after which i plan a 'clean-house' 6.0 re-install (blasting SuSE and SuSE restore partition).

#2: i've used the coffee version of the wifi driver: it had no errors/warnings build-cycle on the 5.2 kernel.
(http://msi-wind-linux.googlecode.com/fi ... 04.tar.bz2)

#3: the kernel source package is -oh what a surprise- called 'kernelsource' . Please take care to select the one that matches the kernel (version..) you are running... (uname -a)

If you still encounter problems compiling/installing the driver, please be so good as to post an extract of the error messages. No promises made, but it could happen that we can make sense of them

Thanks, it works now
I didn't use the coffee-driver because I found a newer version of the original manafacturer's driver.
http://boskastrona.ovh.org/download/rtl ... 008.tar.gz
With the kernelsource package installed, it compiled without problems.

Btw, Why would you blast SLED10 from your system? It's pretty useless, because you can't install new software on it, but it's cool to have a paid Linux distribution on your system If you leave it on your disk, there still is enough space left for Zenwalk, since the Wind has got an 80 GB hard drive.

#1 glad you got a driver to build that works Just a question: my experience with the RTL drivers is that they did produce a lot of warning messages when building them. Is that still so with this version?

#2 i've carefully noted down the url where you got the driver source: i had an IP address before, but that turned to requiring a usercode/password which i of course did not have.

#3 about blasting the SLED: i've not done that yet. The 'spare' partition that the machine came with 'holds' my zenwalk at the moment. It still has enough room. But: i like to have two partitions (ok, 3: one swap...): a '/' partition for zenwalk itself, plus one '/home' partition. That way, i can clean-re-install zenwalk when a release is made, but keep all my user data, the package cache that i redirect to /home/packages in /etc/netpkg.conf and my packages building work which i keep in /home/build. The zenwalk install i have now is more or less a 'test-install': i never bothered with the fact that everything was installed in one partition... But: i find it a bit wastefull to have a 20 Gig SLED partition (which would be perfect for the '/' zenwalk), plus some 10 Gig 'recovery' partition (perfect for a 'snapshot'/test zenwalk installl). So: i do expect that -in the end- i will remove SLED completely. No hurry yet though

I didn't remind whether there were a lot of warnings... At least no fatal errors, the compilation didn't stop.