Hard (for me) to find pages in this Wiki:

?WikiMeta - Discussion of this wiki

The discussion seems to mostly happen in this discussion list : http://lists.debian.org/debian-www/ --AnneGoldenberg

Topics I am currently interested in:

Notes about upgrading to Etch

My Apache2 configurations from Sarge have conditional blocks on SSL being defined. After upgrading to Etch, Apache2 stopped listening for SSL requests. It seems that this is due to the IfDefine SSL blocks, and it has been reported as a bug. The bug doesn't seem to be resolved and I didn't find an official statement on the best way to handle this, so I'll try and start one at EtchApache2DefineSSL.


If I had realized I would be an early adopter using the Intel DG965RY I probably would have stayed away. I had picked another MB but switched to the 965 at the last minute for the onboard video (this will be a small server.) Fortunatly Andrew Barr had blazed the path and made some notes.

I want to replace an aging system with a pretty new large EIDE disk, so my goal was to get the IDE interface (PATA) and networking going. I wanted to identify any potential trouble and have work-arounds in place before moving the hard disk from the existing system into the new one.


His documentation should be sufficient, but I'd like to leave a few notes from my experiances.


Andrew's write-up starts up with this statement: /"The machine was configured initially in Native AHCI mode in the BIOS. This probably accounts for some problems, but not all: read on."/

I would like to emphasize this differently: *Your setup will be easiest using Native & AHCI mode in the bios.* Do this first. Go to Advanced, Drives, ATA/IDE Mode *Native*, Configure SATA as *AHCI*. Without this you will probably not have access to your sata drives. Without this your IDE devices will be hdi & hdj instead of hda & hdb. Without this Knoppix from an ATAPI/IDE/PATA drive won't find it's image & neither will GRML because they don't have device nodes for hdi & hdj (although this can be worked around with GRML's debug boot.)

So, step 1 is to change the settings in BIOS so you are using SATA as AHCI. Next you need to boot.

Debianized 2.6.8 kernel

The first hitch with this kernel is hanging on MMCONF. So use the pci=nommconf option to skip that. Later kernels blacklist mmconf on this board.

Next you will find that it doesn't see your IDE drives. No, switching to ATA/IDE Mode Legacy doesn't help, neither does Configure SATA as IDE. To get these drives working you will need the all-generic-ide kernel option and to have Generic PCI IDE Chipset Support (CONFIG_BLK_DEV_GENERIC) built-in, not as a module. This means CONFIG_IDE and CONFIG_BLK_DEV_IDE must also be compililed into the kernel (Y) and not modules (M). Don't bother though with this kernel. It doesn't have all-generic-ide.

2.6.17 via Knoppix and GRML

Knoppix 5.0 (KNOPPIX_V5.0.1CD-2006-06-01-EN.iso) and GRML 0.8 (grml_0.8.iso) both have 2.6.17 Linux kernel versions (probably with different patches.) Both do not require the nommconf workaround and work with the all-generic-ide option. I noticed that grml supports AHCI so the SATA devices will be available as well. Knoppix probably does. I didn't pay attention.


This is the 2.6.18 kernel patched with 2.6.19-rc1, then with 2.6.19-rc1-mm1, then with [PATCH] pata_marvell: Marvell 6101/6145 PATA driver.

It works great with or without IDE stuff compiled in or as modules (it doesn't use them.) Doesn't require all-generic-ide because libata picks up the ide (PATA) devices. It does require the scsi stuff as modules or built-in. My ide disk and cd-rom now show up as sda and sdb.

Hopefully this is rolled into 2.6.20 or so. It is not necessary to go this far as the 2.6.17 kernels seem to have done the job, but I appreciated seeing the whole thing running under libata.

* http://marc.theaimsgroup.com/?l=linux-kernel&m=116101195514054&w=2


I am interested in learning more about UserModeLinux. Currently all my systems are Sarge though.


Getting my head around the GNU Build system (AutoTools) is turning out to be more work than I expected. I'm writing up some notes to try and keep things straight and document what seems to work best on DebianSarge.

Upgrade a live RedHat 7.3 system to DebianSarge.

I have a spare 4G partition on /dev/hdc2 so debootstrap should hopefully work out. I am basing my upgrade on the notes by Joe Drew at http://capricorn.woot.net/~jdrew/debian-chroot.html.

I tried going right to Sarge, but it's debootstrap depends on a newer version of glibc: /usr/lib/debootstrap/pkgdetails: /lib/i686/libc.so.6: version `GLIBC_2.3' not found (required by /usr/lib/debootstrap/pkgdetails)=. Searching around I found more than one person suggesting that you use debootstrap for the version it was created and then dist-upgrade. So, as Joe's directions suggest, I bootstrapped to woody.

Differences from Joe Drew's notes:

Other points:

Trying to get my ADBKingstonOrbit trackball mouse working right under xserver-xfree86 on Debian 3.1.

My system is a PowerPCBeigeG3 and things are working pretty well, except that the right mouse button acts like the middle mouse button. I found a few pages discussing this on the debian-powerpc list (without that list, I'd think no-one used this thing on X...)

Despite what the last message says, I found that the xmodmap command, xmodmap -e "pointer = 1 3 2" works for me, at least when run by hand. I haven't tried setting it up to go automatcially. My next steps are to do that and tell XFree86 to emulate 3 mouse buttons. Beyond that, it would be nice to not need to use the xmodmap command but to have the right "driver" or whatever. This is what the Linux 2.6 kernel CUDA v0.5 driver for ADB sees (I only have one mouse plugged in):

 Macintosh CUDA driver v0.5 for Unified ADB.
 adb: starting probe task...
 adb devices: ["2"]: 2 2 ["3"]: 3 32 ["15"]: 3 1
 ADB keyboard at 2, handler set to 3
 Detected ADB keyboard, type ANSI.
 input: ADB keyboard on adb2:2.02/input
 ADB mouse at 3, handler 1
 input: ADB mouse on adb3:3.32/input
 ADB mouse at 15, handler set to 4 (TurboMouse 5)
 input: ADB mouse on adb15:3.01/input
 adb: finished probe task...

Power state management

Including powering off, halting, booting, and such (OffAndOnAgain) to work better with ?UninterruptiblePowerSupply setups and remote boot via WakeOnLan.

MS Front Page Server Extensions for Linux

?SargeApache2FPSE - Trying to get MS FrontPage 2002 server extensions working with the apache2 in Sarge. I've since decided that I'd rather stick with Apache1.3 for FPSE and mod_perl stuff, so see my notes at ?SargeApacheFPSE.

File systems

DevFAQ - udev, devfs, dev. What is the Debian way things are done? I need to refactor this. The notes about ttys disappearing are pretty dated - though I suppose the problem could happen again some time.

?KernfsFAQ - usbfs, proc(fs), ptyfs, devfs (dual-homed), and other kernel based file systems. Still very much in the research stage. I am expeciting this to be a well formatted FAQ on the in-memory filesystems the kernel auto-manages like proc. http://lists.debian.org/debian-devel/2003/debian-devel-200312/msg01149.html

I haven't read enough yet to know if other special/important filesystems like tempfs would require their own FAQ or not


PerlFAQ - I've had lots of questions cross my mind about how to do Perl the right way in Debian, especially when tempted with CPAN. Looking around it appears that others have wondered, or tried re-inventing the wheel.


?BuildYourOwnKernel - A quick mini-HOWTO and links to other pages talking about compiling a kernel on debian using the kernel-package tools. I'm interested in trying out a SELinux kernel.


?SecuringDebian - An overview of areas to consider when trying to make your Debian system more secure with links to related and sub topics like SELinux and MultiUserManagement. Using a 'change root jail' like described in Build a chroot can be good for security and for building packages or testing stuff. Maybe there should be a ?CategorySecurity available for these pages.

iptables - A page on iptables has been started. Expanding the sections on fw tools will be good, as would brief discusions on DROP vs REJECT vs REJECT with-icmp-port-unreachable.

SecuringNFS - Mostly taken from the Security section of the NFS-HOWTO adapted to DebianSarge.

autoconf & automake

I'm trying to get my head around autoconf and automake. While experimenting I made an error in my configure.ac, I forgot to close the version string and it gives this error:

 NONE:0: /usr/bin/m4: ERROR: EOF in string

So I Googled around and found pages full of people asking about this error. Most of them had no response, a few had lame responses like "update m4" or something. That may have been valid in some cases, but my autoconf & friends were all used to build the rest of the packages on my system so I was sure it wasn't that. It turns out the error really isn't too cryptic. It means I forgot to close a string. It would be nice if it said the line number the string started on. Oh well. It turned out to be the version string in my case. So, if you get that error, make sure your open and close brackets match up.

Shuttle XPC SK41G (FX41)


The Shuttle XPC systems generally use their heat pipe technology for cpu cooling. They also have a three stage cpu fan speed controller. Slow is ~2200 RPM, medium is ~3000 RPM and high is ~3300 RPM. You can set the bios to switch between low and medium at tempratures ranging from 40C to 52C. The high speed is hard coded to kick in at 56C.

I wanted to see what the temprature was and the fan speed was from time to time using xsensors and keep a log of when it the values go out of min/max levels using sensord. (For example when the CPU temprature hits 56C.) I'm running Linux Kernel 2.6.8.x and it was a simple matter of installing the aforementioned packages. This pulled in ?libsensors3 and [lm-sensors]. I could also monitor with mbmon and xmbmon without libsensors.

I then stumbled around at what modules to load before I read about sensors and sensors-detect. I let sensors-detect append to /etc/modules. This worked pretty good, except that +3.3V was 2x too high, -12V and -5V were way off (-20V and -9V, despite the fact that BIOS said they were practically right on). It also claimed my ?VCore 2 and tempratures were out of bounds and the max tempratures were lower than my normal operating range. (/mbmon also has issues with the 3/5/12 volt readings.)

Fix the 2x too high 3.3V issue:

 # If 3.3V reads 2X too high (Soyo Dragon and Asus A7V8X-X, for example),
 # comment out following line.
 # It was 2x too high on the ["SK41G"] (["FX41"])
 #    compute in2   2*@ , @/2

Fix the negative voltage readings.

 # The next two are negative voltages (-12 and -5).
 # The following formulas must be used. Unfortunately the datasheet
 # does not give recommendations for Rin, Rf, but we can back into
 # them based on a nominal +2V input to the chip, together with a 4.096V Vref.
 # Formula:
 #    actual V = (Vmeasured ''' (1 + Rin/Rf)) - (Vref ''' (Rin/Rf))
 #    For -12V input use Rin/Rf = 6.68
 #    For -5V input use Rin/Rf = 3.33
 # Then you can convert the forumula to a standard form like:
 #    compute in5 (7.67 * @) - 27.36  ,  (@ + 27.36) / 7.67
 #    compute in6 (4.33 * @) - 13.64  ,  (@ + 13.64) / 4.33


 # Neither of those calculations work for the ["SK41G"] (["FX41"]) :(
 # in5 (-12V) is 0.96 when BIOS shows was showing 11.95-12.00.
 # Adjusting -12V Rin/Rf to 4.16 gives a close value
 # in6 (-5V) is 2.42 when BIOS was showing 5v
 # Adjusting -5V Rin/Rf to 4.43 gives a close value
     compute in5 (5.16 * @) - 17.04  ,  (@ + 17.04) / 5.16
     compute in6 (5.43 * @) - 18.15  ,  (@ + 18.15) / 5.43

Fix the !?VCore 2 (BIOS calls it VDDQ) limits

    set in1_min 1.4
    set in1_max 1.6

Increase the tempratures.

    label temp1       "M/B Temp"
    set   temp1_over  48
    set   temp1_low   15
    label temp2       "CPU Temp"
    set   temp2_over  56
    set   temp2_low   15
 #  ignore temp3
    label temp3       "Temp3"
    set   temp3_over  48
    set   temp3_low   15

Fix up the fan settings.

 # Fans
     set fan1_min 2200
 #   set fan2_min 3000
 #   ignore fan3
 #   set fan3_min 3000
 # Ignore fans you don't actually have
     ignore fan2
     ignore fan3

make-kpkg, distcc and ccache

I was having trouble making make-kpkg use distcc and ccache via the immediatly obvious ways as per the make-kpkg man page so I searched the web and found a lot of other people had asked the same question. I then found an answer:


I tweaked that solution a little to a $HOME/bin "masquerade directory" of ccache, CCACHE_PREFIX="distcc", and the rest of the options mentioned in that article.

 DISTCC_HOSTS="fastest faster localhost"

Be sure to read the distcc and ccache man pages for help with the "masquerade directory" and other options. Once all this is set, use make-kpkg as per the standard suggestions.



Maybe perhaps someday if I get around to it, I'll become proficient enough at 'refactoring' other people's contributions to earn the right to be on the ?DebianWikiMasters list, but don't hold your breath. Until then I'll practice random acts of ?WikiGnome ing and hope that I spell things correctly.

Figure out a kwiki hack for extra key words and acronyms so that NMU, BTS, and RC would auto-link w/o the braces on a Debian focused wiki, and to link non ?WikiWords when a page exists.

Wiki Mail Box

Have a comment? Need me to 'take back' something I said? Post it below this line.

Hi Jacob. Welcome to the Debian wiki. Thank you for writing useful content here (such as WakeOnLan ). -- DavidCary

You're most welcome. I apologize for the delay in responding. I did read your note quite some time ago.