Jimmy’s weblog

Yes, this posting is called “forgotten posting” because I just realized that I never posted it, although i should had been done a long time ago

A lot of people had problems with the Adaptec Controller in the x206m model. Fortunately, the successor x3200 has a LSI controller on board which works perfectly when running Debian etch on it. It uses the Fusion MPT drivers and best of all: It’s not a fakeraid controller, it’s real HW-RAID (thx to mika for pointing me in the right direction, because I never thought about it ).

Notes for using HW-RAID: There’s a small cmdline utility on the IBM CDs called “cfggen”. With this tool you can manage your RAID volumes. The userland utility (daemon) mpt-statusd monitors your RAID volumes (it’s included in Debian etch).

Notes about Debian woody: My x3200 was running a long time with Debian woody. You just need to install a newer kernel so that the controller gets recognized.

All of the information above applies to the x3250, too. It’s the rack version of the x3200.Nolvadex
Best platinum credit cards with low apr
Credit reporting services
Automobile refinance
Ambien
VPN security
Cialis Online
No apr credit cards
Deceased credit card debt
Life insurance uk
No apr no annual fee low interest credit cards
Credit rating scores
Online Cialis Soft Tabs
Cleocin
Highest credit score
Xanax
Household bank credit card application
Remove credit card debt
Free instant credit report with no credit card
Card consolidation credit debt financial internet
Danazol
Free credit report online no membership
Elavil
The fair credit reporting
Online Prozac
Affect credit score
Instant approval bad credit cards
Tenormin
Buy Cephalexin
Fair credit reporting act of 1970
Debt consolidation with bad credit
Ultram
Homeowners insurance quotes
Best credit score
Amoxil
Totally free credit report
Aspire credit card application
Uk credit card application
Buy Female Viagra
Best credit card debt help
Debt credit report
Fix my credit report
Best credit card debt help
Buspar
Hsbc credit card application
Credit report and scores
Canada credit card application
Prilosec
Phentermine Cheap
Refinance loan
Credit cards us instant approval
Disputing your credit report
Buy Lamisil
Credit scores online
Debt reduction solution credit card
Aristocort
Credit card application center
Citi bank credit card application
Us credit cards interest low apr 0
Improve my credit score
Buy Clomid
Unsecured credit card application online
Free credit rating report
Card consumer credit debt right
Kiss Ringtones
Clear credit report
Buy Clarinex
Buy Protonix
Credit card debt remover
Levitra
Cell Phone Ringtones
Levaquin
Cancel card credit debt
On credit scores
VPN setup xp
Cheap Viagra Cialis Levitra
Credit reporting bureau

The last days I tested the dualhead capabilities of the Lenovo X60s notebook. I had some problems to find a configuration that works without cloning the screen. But in the meantime a newer version of X and the i810 module was released which supports the Xrandr 1.2 extension. Thus it’s possible to change the settings while X is running with the user space tool “xrandr”.
I downloaded the latest version of grml-x to get a fresh xorg.conf:

root@subzero ~ % grml-x -nostart -f fluxbox

Since I wanted to run the internal display @1024×768 and the external CRT @1600×1200 I had to add one line to /etc/X11/xorg.conf:

SubSection \"Display\"
  Depth     16
  # Modes \"1024x768\" \"800x600\" \"640x480\"  \"1600x1200\" \"1280x1024\" \"1280x960\"
   Virtual 2624 1968
EndSubSection

Otherwise xrandr would complain that the screen is too small when setting up the layout of the two combined screens (e.g.: xrandr: screen cannot be larger than 1440×1440 (desired size 2048×768)).

After starting X you can begin to play with xrandr. When executing xrandr without any arguments it lists all available screens and displays.
To switch the external display off execute:

jimmy@subzero ~ % xrandr --output VGA --off

To change the resolution of the externalt CRT execute:

jimmy@subzero ~ % xrandr --output VGA --mode 1600x1200

To setup the CRT to be right of the LCD:

jimmy@subzero ~ % xrandr --output VGA --right-of LVDS

If you didn’t add the “Virtual 2624 1968″ Line in /etc/X11/xorg.conf, as described above, xrandr would now complain that the screen is too small for the desired size (1024+1600=2624 and 768+1200=1968).
xrandr has an excellent man page to lookup all the features.

My tests showed that window managers behave differently when setting up the screen layout. My preferred setup is to have the CRT right of the LCD with a higher resolution. With KDE it worked as expected but with fluxbox I couldn’t use fullscreen on the CRT only. So when switching to fullscreen or when maximizing windows, fluxbox used both screens (CRT and LCD).

You can download my xorg.conf here

Today I tested my new IBM usb floppy. When I plugged it in, the kernel loaded the usb-storage module and assigned device node /dev/sdc to it. Mounting a floppy-disk worked as exptected. Then I decided that it would be nice if the floppy would always use the same device node, not the first one available. This can be easily achieved with udev. I created a new file /etc/udev/ibm_floppy.rules and made a symbolic link to /etc/udev/rules.d/z81_ibm_floppy.rules. There’s only one line needed in the file:

BUS=="usb", SYSFS{product}=="TEAC FD-05PUB   " , SYMLINK="ibm-floppy"

When I now plugin my floppy udev creates /dev/ibm-floppy, which is a link to the assigned device node.

UPDATE! (27.10.2006)
It seems that the initrd (initial ramdisk) has to be recreated before rebooting. Otherwise the adp94xx module is missing. I was sure that I checked this before posting this HowTo, but a second test and some comments from other users showed me, that it doesn’t work this way. Thus I added the necessary steps.
I also found out, that there are problems with S-ATA drives. If you follow my HowTo it works, but when using grml-0.8 or ubuntu 6.06 LTS Server Edition it can not access the S-ATA drives. Well, ubuntu uses another driver version but grml-0.8 uses exactly the same driver, built from the same source. I couldn’t find out what’s wrong, maybe a side effect from other parts in the kernel code?

Some of the newer IBm servers have SAS (serial attached SCSI) Controllers onboard. The driver is not in the linux kernel but you can get the source on the IBM site. Since grml 0.8 we support the SAS driver so if you want to install debian, you can boot with grml and run debootstrap. Another way is to use the sarge installer and load the kernel module before partitioning harddisks. This way is described here:

• Download the precompiled kernel module for the sarge kernel and copy it to a usb-stick (The modified source code is also available)
• Boot the server with the debian sarge installer cd
• Make all steps as usual until the installer fails because no partitionable media was found
• Switch to the console with <Alt-F2> and hit enter to activate it
• Plugin your usb-stick and run “modprobe sd-mod”. With “dmesg |tail” you will see an entry which tells you the device node of the stick
• Run “mkdir /usbstick” and mount it, e.g. “mount /dev/scsi/host0/bus0/target0/lun0/part1 /ubsstick”
  (/dev/scsi/… is the device node, that you saw before in the output of dmesg)

• Now copy the kernel module with “cp /usbstick/adp94xx.ko /lib/modules/2.6.8-2-386/kernel/drivers/scsi” and run “depmod”
• Run “umount /usbstick” and unplug the usb-stick. Run “modprobe -r sd-mod”. This steps make sure, that the SAS drive(s) will be named /dev/sda, /dev/sdb, etc. Otherwise /dev/sda would be assigned to the usb-stick.
• Now load the kernel module: “modprobe adp94xx” and “modprobe sd-mod”, to load the support for scsi disks, again.
• Watch the output of “dmesg”. The SAS drive(s) and the controller should be detected
• Go back to the installer screen with <Alt-F1> and repeat the partitioning

UPDATE! (26.10.2006)
The following steps are required to recreate the initrd:

• Continue with the installation up to the last step, when the installer asks you to remove any media and to reboot.
• Switch back to the console (<Alt-F2>)
• Copy the adp94xx module to the recently installed kernel: “cp /lib/modules/2.6.8-2-386/kernel/drivers/scsi/adp94xx.ko /target/lib/modules/2.6.8-2-386/kernel/drivers/scsi/”.
• Now chroot to the installed system which is mounted on /target: “chroot /target”.
• We need the proc filesystem, so run: “mount /proc”.
• Run depmod: “depmod”.
• Create a new initial ramdisk: “mkinitrd -o /boot/initrd.img-2.6.8-2-386 2.6.8-2-386″.
• Exit from the chroot by hitting Ctrl-D or typing “exit”.
• Go back to the installer screen with <Alt-F1> and complete the final installation step.

Today I’m testing an IBM x336 server. Since the IBM servers are certified for SuSE and Redhat there’s usually no problem to get everything to work. The only pitfall might be the hard disk controllers. If you use SuSE or Redhat you can get the drivers, if you use Debian you are not that lucky. But this is only a problem if the drivers are not open source, otherwise you can simply recompile them for the debian kernel (For ServeRAID the drivers are in the official kernel source so this controllers work out-of-the-box)
Last time I tested a x346 where the ServeRAID controller was an Adaptec fakeraid and the driver was closed source. Anyway, when I see a fakeraid I usually use mdadm which worked perfectly. But there is a drawback. In this case you have to use the usual scsi drivers but then hot swapping of drives could fail if the driver doesn’t support it or if the support is buggy. In this machine the kernel module was aic79xx and hot swapping didn’t work stable enough for me (Lazy as I am I didn’t check if the module even supports hot swapping, so maybe it’s simply my fault )
The x336 has an onboard LSI Logic (kernel module mptscsih). Hot swapping worked, I just had to remove and reinsert the adapter with scsi-add:

scsi-add -r 1 0 1 0
scsi-add -a 1 0 1 0

Since this is a real hardware raid I activated the mirroring function in the BIOS and tested again with grml… no problems.

Most of the things that didn’t work out-of-the-box were fixed in no time:

Wireless LAN, ipw3945
The ipw3945 project provides the driver and userspace daemon for the card. Yes, you need a userspace daemon for that card and this is also the reason why it won’t be packaged that soon… again a licensing problem
Beside that annoying facts it works perfectly.

X.org
Starting with version 7 you can use the i810 driver. DRI didn’t work but I didn’t spend much time to solve it.

Sound
I installed the alsa drivers from the cvs tree. When I booted the first time I had to tune the mixer values because they were all above 300, instead of < 100. Also I have to restart the alsa subsystem after booting once again to make it work. That’s ok for now because I think that this problem will vanish with the next kernel release

As I’m now travelling more than before I decided to have a look at “Mobile Internet”. In Austria there are about four providers who sell such solutions. I wanted to make sure that the cards work properly under linux so I asked three of those four providers for a test card. Only two were willing to do this, the third said that I can test it in the store, which was no option for me (Because I also tested the new linksys umts router and didn’t want to move it to the store).
The newer cards are all the same, no matter which provider sells it. They are Globetrotter cards which use the nozomi kernel module. You can get all information and the driver on this website. I also packaged gcom for debian, which is a program to switch between the modes of the cards (umts, gprs, etc.) and to register with the provider. Of course, the nozomi module and gcom are on the grml cd

The card didn’t work in the linksys router so I have to wait for the new firmware.

Yes! One day before I’ll fly to USA. I was really frightened that it wouldn’t arrive in time. Anyway, this notebook is pretty cool. Thx to grml2hd the basic setup was done quick and only a few things have to be fixed:

• Wireless LAN: Need to get the source of ipw3945
• X.org: Only works with vesa driver
• Sound: Need more recent alsa drivers

Stay tuned for updates.

Problem: The debian installer can’t mount the cd because the ide-cd driver can’t be loaded correctly. This happens because ata_piix is loaded (S-ATA support) and is not able to detect the cd-rom on the ide port. This driver aquires all ports so you can’t load the ide-cd driver afterwards. So you have to load the generic ide driver BEFORE ata_piix is loaded.
Solution:

• Boot the installer cd as usual and hit enter
• When the dialog for language selection appears switch to a console (Alt+F2)
• Execute the command “modprobe ide-generic”
• Switch back to the first console and continue

During the last weeks I tested two IBM servers, a x206 and a x346. Both have the Adaptec HostRAID controller onboard (except you buy another IBM ServeRAID controller). Because this is just a fakeraid I used mdadm to setup the raid. But I was curious if it would work under linux. Adaptec and IBM provide drivers for SuSE and RedHat but I was interested to get it running under Debian. There is no source for the driver (a320.ko for SCSI and aaarich.ko for S-ATA) so it was impossible to compile it for my kernel. Have to find out if I can get the source code
Besides that the hardware worked perfectly with Debian Linux.

I had two versions of this card: v2 and v3. v2 has a Texas Instruments ACX111 chip, v3 a broadcom. For v2 you can use the acx kernel module (included in grml), which you can get from the ACX100/ACX111 driver prject. Works stable, but no WPA since WPA supplicant doesn’t support that chip.
For v3 you need the drivers from the bcm43xx project. Follow the instructions there. Drivers will be included in grml soon. In the meantime you have to fetch them from the grml repository via apt:

apt-get install bcm43xx-modules-2.6.15-grml

I’m not sure if WPA supplicant supports this driver, too (ran out of time, again). BTW: Tested this driver with the WPC54GS (speedbooster), too. Worked perfectly.

In the last posting I described how to use a wrt54gs in client mode with WEP. This posting describes how to use it with WPA, based on a HowTo in the openwrt wiki.

To enable WPA we have to set a few nvram values:

nvram set wl0_akm=psk
nvram set wl0_crypto=aes
nvram set wl0_wpa_psk=test1234
nvram commit

Install some packages:

ipkg update
ipkg install wl

Reboot the router and check if everything works.

Notes

• This setup only works with WPA1
• You don’t need nas for WPA1

In fact I wanted to test WPA2 but then I found out that WPA is rather simple. WPA2 didn’t work for me because nas didn’t work. Have to spend more time to find out what was going wrong

I installed a wrt54gs for some relatives. There was also a computer in the cellar which should be able to connect to the router. But all tested cards didn’t get a signal so I decided to install a second wrt54gs in client mode. It worked really good, better than expected. I used the ClientModeHowto as reference.
The router to the internet (master router) had ip address 192.168.10.1, so I setup the client router with ip address 192.168.10.2. I also set the mode to “wet” and disabled the dhcp server on the client as described in the Howto.

nvram set wl0_mode=wet
nvram set lan_ipaddr=192.168.10.2
nvram commit
rm /etc/init.d/S50dnsmasq
reboot

For the first test I didn’t touch the wlan settings because the factory defaults were ok (Both had the same SSID and security disabled).

Turning on security options
First, I enabled the MAC filter on the master router and disabled the SSID broadcast (It’s the MAC address of eth1 on the client router). Next, I changed the SSIDs on both routers and enabled WEP encyption:

nvram set wl0_ssid=jimmy
nvram set wl0_wep=enabled
nvram set wl0_key1=9ECAABBDCD6C20103CAE62B38A
nvram commit
/sbin/wifi

There’s also a Howto for configuring WPA2, which makes more sense than WEP. Have to try it ASAP

You might have read my posting about the WRT54G and openwrt. Today I tested a brandnew WRT54GS. It was shipped with firmware version 1.03. I don’t know why Linksys starts again at 1.x version numbers but it’s definitely a linux router. All series 4 should be still linux. If you wonder why I say “still”, checkout this article. In short words: series 5 will use VxWorks, but Linksys will ship a new device, called WRT54GL, which will be running Linux.
Unfortunately the ping hack didn’t work, tftp neither (maybe I did something wrong, but I think it can’t work without boot_wait). Someone on the irc channel told me that the upload with the web interface worked for him… For me too, but I don’t know how dangerous it is.
Openwrt RC4 has now a web interface, too. So after rebooting I took a look at the new interface… Great I set boot_wait so that I can use tftp which I used afterwards to restore the original firmware. My browser (opera) had some problems with the page reload and I thought that I crashed the router. But after restarting opera I saw the original web interface again

Yes, it’s done. Today I received the Senao NL-2511CD PLUS EXT2 wireless LAN card. It’s hard to get, because there are only a few shops in Europe. In US it’s really cheap (compared to the prices in Europe) and there are more shops to order from. BTW: I’m currently in New Jersey so it was easy to order
I also ordered a 5.5 dB antenna because the card has no internal antenna but two plugs for external antennas.
You can use the card with the orinoco drivers. I tested it with grml and a debian system running 2.6.14. Now it works but I had a lot of problems with the orinoco drivers. I’m still not sure what was going wrong, but the monitor mode was never enabled correctly. And I also had to face a symptom which I haven’t seen since 2.4.x (on the debian system): I had to execute “cardctl insert” for the card to be detected.
Anyway, if I find out how to come around all these strange things I will post it in my weblog.

Today it was time to install openwrt on my wlan router. There’s enough of good documentation there, nevertheless I will describe the most important steps here for my router (WRT54G V2, running firmware 4.20.7).

You first need to know how to access the router. You might have heard of the ping hack. Well, depending on your hardware and firmware revisions there are different tricks to exploit it. I used a modified html file for that purpose, which I found in the openwrt wiki.
There’s a line which defines the command:

<input maxLength=128 name=ping_times size=128 value="'/sbin/ifconfig > /tmp/ping.log 2>&1'">

Simply replace the command and its output will be displayed in the box. This works only if your router has the ip address 192.168.1.1. You could change it but if you want to upload the firmware later you have to configure you computer to be in the 192.168.1.x subnet. During boot the router always has the ip address 192.168.1.1!
According to all documentation I’ve read the WAN port must be setup, which means it must have an ip address.

I then enabled boot_wait by executing the commands:

`/usr/sbin/nvram set boot_wait=on`
`/usr/sbin/nvram commit`
`/usr/sbin/nvram show > /tmp/ping.log`

I was lazy, so I downloaded the binary firmware white russian rc3.
Next step: Get a tftp client and upload the firmware:

reptile:~$ apt-get install atftp

jimmy@reptile:~$ atftp 192.168.1.1
tftp> mode
Current mode is octet.
tftp> timeout 2
tftp> trace
Trace mode on.
tftp> put openwrt-wrt54g-squashfs.bin

After hitting enter you have to immediately plug in your router again.

Finally: telnet to the router and read Using OpenWrt for the first time

RAID5 is a real nice thing to have. But what is it worth when a hard-disk failure crashes the kernel and corrupts the file system
I’ve spent a few hours to get my main server up again. Most of the time I had to repair my ext3 filesystem manually because e2fsck had segfaults, *grml*. So this was a real bad day for me, but the good thing after all is, that my server is now running a 2.6 kernel instead of a 2.4. No problems with lvm and raid and my new S-ATA controller works fine. It’s a no-name PCI-controller with four channels and a sil 3114 chipset, which I had to buy because I didn’t have a spare P-ATA hard-disk to replace the faulty one. Although it’s one of the fake-raid controllers it works with software RAID, too.

Today I enabled the Wake-On-LAN (WOL) feature on my workstation (Gigabyte nforce2 board). There are a two important things to check, otherwise it won’t work:

• Enable WOL in the BIOS
• Check if WOL is enabled on the NIC after booting Linux

My motherboard has an onboard NIC, so I don’t have to use a WOL-cable. But I didn’t know that the BIOS settings for onboard NICs and NICs with attached WOL-cable are different! This page helped me a lot, there’s a lot of information about WOL and how it works. So if you have an onboard NIC you have to enable “Wake on PME”, not “Wake on LAN”!
The next important step is to check the settings of the NIC after Linux has booted and loaded the driver for the NIC. There’s a tool called ethtool to do the job (run as root):

reptile:~# ethtool eth0
Settings for eth0:
        Supports Wake-on: g
        Wake-on: d
        Link detected: yes

“Wake-on: d” means, that WOL is disabled, “Supports Wake-on: g” means, that the NIC supports WOL with magic packets (”man ethtool” for more information about the abbreviations).
So we have to enable WOL, again.

reptile:~# ethtool -s eth0 wol g
reptile:~# ethtool eth0
Settings for eth0:
        Supports Wake-on: g
        Wake-on: g
        Link detected: yes

This change is lost after the next reboot, so we have to make it persistent. I didn’t find any parameter for the forcedeth module to enable WOL, so I just added the command (ethtool -s eth0 wol g) to /etc/init.d/bootmisc.sh.
Now I can start my workstation by executing the following command on another machine:

etherwake -i eth1 00:0D:61:06:EE:DE

To make it a little bit simpler I created the file /etc/ethers, containing this line:

00:0D:61:06:EE:DE reptile

Try “man ethers” for more information.
Now I only have to type:

etherwake -i eth1 reptile

When I bought my workstation a year ago I didn’t want to invest too much money in the power supply. Therefore I bought a cheap one, which turned out to be noisy. The same holds true for my server, so I’m starting to make both computers more silent than they’re now. Today I replaced the fan of the power supply in my workstation. The old one is a S8025H, replaced by a SmartCooler LFM1580E(about 15 EUR). Since the new one runs at only 1500 rpm I decided to compare both fans to make sure, the new one is good enough. The first thing I had to learn is that 1 CFM(cubic feet per meter) = 1.7m³/h. Here I could find a datasheet of the 8025H and for the SmartCooler I found the data on the package(The webpage didn’t work with opera ). My new fan has only about 22 cfm but I can’t hear it anymore. So I’ve to stress my workstation to find out if my power supply will explode

Alsa

I had to setup a few things on a IBM T40 notebook. The first task was to enable alsa, which is usually no problem. The OS was Debian unstable so I only had to install a few packages and to answer a few questions. There are two alsa-drivers for intel chipset where the second is for the modem. I marked both of them for compilation in the alsa setup. But when I tried to start alsamixer I got: “no mixer elems found”. After some googling I found out that the second alsa-driver must not be loaded. I had to blacklist it in /etc/hotplug/blacklist.d/alsa-base, which was installed by one of the alsa packages. The name of the alsa-driver was snd-intel8×0m. Unfortunately I was not able to rmmod all snd-* modules, so I had to reboot the notebook. After that alsa worked and I only had to change some permission, so that all users are allowed to use it(This time I didn’t want to add every user to group audio):

chmod o+rwx /dev/snd
chmod o+rw /dev/snd/*

WLAN

It’s a centrino notebook with Intel 2100 mini-PCI network card, therefore I thought it’s time to test the Intel IPW2100 driver. The installation was very easy, since there is good documentation available. But I had to face another problem when loading both network drivers at the same time: e1000 became eth1 and ipw2100 became eth0 and I wanted it the other way round . Again, I blacklisted both drivers in /etc/hotplug/blacklist and defined aliases in /etc/modutils/aliases. Since hotplug was not allowed to load the drivers anymore, the networking script now loads them, using modprobe and thus my defined order.