sicherheitsschwankung.de - some things I do for free

After a long break i'll start blogging again. :) All the software has been updated to the current versions and it seems to be working.

Yesterday i've recieved the confirmation that i'm in the Neo1973 Phase 0 early access program. Yay!

When the phone has arrived i'll write a short review and start working on some interesting applications (i.e. an OpenStreetMap client).

Today the long awaited lecture videos have been releases here. We have a complete mirror available.

If you want to create a public mirror, please contact me. I'll give you access to a non-public server with all the files for mirroring.

Thanks to FeM e.V. for creating the recordings!

Just some nice things from the Desktop side:

Leaftag: Tag files on your system (Homepage, Video)

XOrg Bling: Using OpenGL to make nice effects

• by Novell/SuSE:Compiz
• by RedHat: aiglx (Videos: 1 2 3)

The 22C3 in Berlin is coming up and after helping a friend convert his notebook to dm-crypt/lvm, i'm ready to try it myself :) It was not quite straightforward, so i'll describe the necessary steps here.

When i'm done, i'll have two physical partitions (hda1 and hda2). hda1 is /boot and must be unencrypted because it contains the kernel and initrd. hda2 is an encrypted LVM physical volume which contains / and all other volumes.

Before we begin some packages need to be installed: yaird (to build the initrd) cryptsetup-luks (supports a partition header which containes the encrypted key) lvm2 (to manage volumes) yaird and lvm2 are in debian and cryptsetup-luks is avaible from http://luks.endorphin.org/ (a debian repository is at http://einsteinmg.dyndns.org/debian/).

</http://einsteinmg.dyndns.org/debian/></http://luks.endorphin.org/>

The code needed to create two bitstreams with the same hash is public. It takes about 45 minutes to find a MD5 collision. This does not allow to create a bitstream which maps to a given hash, so md5 is still safe for signing cleartext. However you can make two different documents in PostScript/PDF which have the same hash (more details). Use SHA-256 or SHA-512 (not SHA-1). MD5 collisions could also be used as "watermarks" for mp3 files (players skip invalid data) to track their distribution on p2p networks.

Joey has blogged about thread patterns in mailing lists and how to select interesting mails based on these patterns.

I use evolution, which could learn some things from mutt about threaded display. It would be nice if one could sort threads not by the date of the original mail but by the last reply.

Maybe it would be a good idea to start a community-based service to tag mails. That would allow slashdot-like filtering based on tags by people whose votes correlate with ones own. I should probably look into writing a plugin for evolution :)

Vernor Vinge - Accelerating Change 2005 on IT Conversations - In this keynote address from Accelerating Change 2005, Vernor Vinge discusses the potential for a "technological singularity" - the event at which the creation of artificial superhuman intelligence changes the world so dramatically that it is impossible to imagine the world after that point. He explains that the singularity is not a given, nor is it necessarily a positive event. Many factors could arise that prevent the singularity from occurring and there is a potential for it to be a catastrophic event rather than a positive revolution.

I've been using bins for my photo galleries for some time now. It doesn't have all the features that Gallery has, but i like the clean themes. To use bins (which generates static html files) with Drupal, i use a module called wgHTML. wgHTML takes the static html files and wraps them with the Drupal theme and menu. After fixing some problems with the bins theme, the gallery looks quite nice :)

If you're running Debian unstable and have updated recently, your eth interfaces may be numbered differently now. This is because udev is now started earlier (at 05) than module-init-tools (at 20). If the order in /etc/modules is different from what udev detects, you get changed interface numbers.

So what can we do to prevent this from happening again?

We've just connected the new disk to our mirror, but they are not yet usable. At first we had a defective sata cable and after fixing that, it seems that the PCI ID of the SATA controller is not yet in the 2.6.13.4 sata_promise driver. I think i read somewhere that there is a patch in 2.6.14 :) Perhaps i'll figure it out tomorrow.

Update:

The controller works flawlessly after adding the PCI ID to sata_promise.c.