AK's weblog

"If you want to recruit killers for foreign conquest, you need to be able to offer them the three basics: treasure, murder and pussy. This is why Iraq is a dead end. There is no pussy in Iraq, absolutely none. No 'me so horny' scenes will be shot in the inevitable Iraq movies."

Design Patterns are so late 1990's, the new hype for 2004 are...Resign Patterns!!! Knowing these resign patterns by heart should be considered extremely important by every software engineer.

I'm so totally fucking wasted. Going out at 11 pm and coming home at 7:30 am is just not the right thing for me, at least when I have not more than 5 hours of sleep afterwards. That kind of partying is just too much for me, I think.



The photo that you can see here was taken at a new coiffeur in Linz right at Bürgerstraße. The last time I saw this error message it was there for already more than 2 days. So, nobody really seems to care about it...

During the last 3 evenings, I worked on ContraPolice, a libc extension for protecting applications from heap-smashing attacks. The idea is simple: when memory is dynamically allocated via malloc(3), so-called decoys are placed around it. Should a heap overflow occur, it is very that one of the decoys is overwritten. The decoys are then checked by the libc whether they still match, and if they don't, the program is aborted immediately. This means that all the potentially critical library routines like strcpy(3), strcat(3), sprintf(3), etc. are modified so that they do this check before they return. Due to the way ContraPolice manages the allocated memory blocks, it is extremely inefficient. The current implementation is only available for dietlibc, a small libc for Linux. I chose it because it is easy to understand and modify, and very compact. When I have enough time, implementations for other libcs may follow.

For the sensation-savvy people, here are photos of the DHL aircraft that was hit by a SAM-7 surface-to-air missile near Baghdad on Nov. 22. As you can see, the emergency exit went right into the barbwire, but the crew is OK.

(via Lunatic Fringe)

Thanks to Michael for bringing up the server that hosts synflood.at again. It's really a PITA when a server doesn't work for quite some time, then works for a day or two, and then does not anymore. Expect more blogging real soon now (finally).

After some problems with DNS and the webserver itself, synflood.at and this weblog is back online.



Yesterday I saw a movie that I can really recommend. It's called una casa con vista al mar. Moviemento played it, in the original language (spanish) with german and french subtitles. Paula, a friend from Argentina, said that the spanish spoken in the movie was not quite authentic, since the main actor was actually speaking Spanish dialect, while all the other people spoke some more or less local dialect (the movie plays in Venezuela). Anyway, it's quite interesting, and I can only recommend it (and no, I won't tell you anything about the plot).

/ the following nonsense is required

  because casts turn an Lvalue

  into an Rvalue so two cheats

  are necessary, one for each context.

 */

union { int _cheat;};

#define Lcheat(a)       ((a)._cheat)

#define Rcheat(a)       ((int)(a))

Found in the sh source code of 32V UNIX, which is written in the (in)famous BOURNEGOL, an Algol derivate based on C.

Because of regular crashes when reading certain emails (Hello Paula! :-), I compiled mutt 1.5.5 under Mac OS X 10.3.1, since fink doesn't yet provide anything more recent than mutt 1.4. Before I started using fink, I compiled mutt (1.3.28 IIRC) for OSX by myself, which was always quite tricky, and never went smoothly without any tinkering with the source. Anyway, mutt 1.5.5 compiled out of the box, without any changes, which is really fine.

I returned from Luxemburg on Sunday, but so far, I didn't have the time to update my weblog.



I went to Luxemburg by car, together with CM, Chris and Hans, which was quite boring, since it took almost 8 hours to get from Austria to Luxemburg. I actually expected to sleep in a youth hostel (that's what the organizers wrote in the emails), but it was actually an "Ibis" hotel. The hotel room was really small, but everything else was quite OK. So, after the check-in, we drove to some pizza restaurant. Well, pizza in Luxemburg doesn't seem to be very good, and also the beer (Mousel) was, well, sub-standard.



The next day, we had breakfast at the hotel, which I really enjoyed, since everything was there: fresh croissants, butter, jam, orange juice. Just nice. After that, we went to the IST by car, where the whole event was already starting. The whole event was quite OK, there was free coffee and free cookies available, and the lectures were quite OK, too. After that, we again went eating pizza, also with a b-class pizza, but at least the beer was OK.



On Saturday, the same thing as the before, but this time, I didn't really have the time to have that much breakfast, since I had my lecture starting at 9:30 in the morning. The lecture went pretty smooth, although not more than 10 people showed up. After the lecture, I even got the book "Secure Programming Cookbook" from a sponsor. At 1:30 pm I went to the Utopolis cinema, and watched "The Matrix Revolutions", in english, of course (hey, when do I have the chance to do that again?), with french and netherlandish (?) subtitles. Then I went back to the IST, watching the last lecture, and joining the social event. After the social event, we again went out eating (chinese this time), and then back to the hotel.



On Sunday, we again got up, had breakfast, and then drove home to Austria, which was extremely tiring. :-/



Photos about the event can be found here: http://synflood.at/gallery/view_album.php?set_albumName=linuxdays-lu.

In less than two hours I'll be on the way to Luxemburg. According to the Michelin route planner, the journey will take more than 7 hours. :-/ Well, at least I have something to read with me, "Der Standard" (a very good newspaper) and the latest "Communications of the ACM", which contains quite interesting articles this time. For the journey, I prepared my camera, so I will return with a lot of photos from Luxemburg.

During the last few days, I had to use the Samsung V200 several times, and so I noticed the following points why the Samsung V200 sucks:

• There seem to be two different kinds of input fields, e.g. the URL input field for the WAP browser behaves differently than an input field in a WML page. It looks as if the same widget was implemented twice.


• The WAP browser can't show bold and italic text in WML pages. Even first-generation WAP mobiles can do that.


• The WAP browser's decksize is extremely small. It reminds me of old mobile phones. I didn't measure it, but it's definitely way below 3.5 kB.

My advise: don't buy it! If you bought it, bring it back, and get another one. I really like the Nokia n-Gage and the SonyEricsson P800, both really neat mobiles. The Nokia n-Gage is the only Series60 mobile phone where the cursors do not suck, and the P800 is more a PDA than a mobile, even with a little stick included (the P800 has a touch screen), and it comes with handwriting recognition.

KDE appears in 24 and Alias:

• http://slashdotted.19inch.net/24-kde-slashdotted.html


• http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&selm=v24q9fgs1hq869%40corp.supernews.com

Pat Villani mentioned (1, 2) on the Unix Heritage Society mailing list that he's working on a port of 32V UNIX for i386. For those who don't know: 32V UNIX was the very first version of UNIX that was ported to the VAX architecture. So, this is definitely a nice project which I will observe for sure, and, if I have enough spare time, will try to contribute to.

Damn, somehow I couldn't manage to get ISDN working with Linux 2.6.0-test9. Well, I will have to stick to 2.4, then. :-/ What is also annoying is that the swap partition must be of type 1 (on my installation it was type 0), otherwise it doesn't get added and you have a system without swap space, which is not so nice, especially when you want to recompile a kernel.



Well, in the next few days, I should concentrate more on hacking cinderella, since in less than a week, there are Linuxdays in Luxemburg. I took part in the CfP, and my paper (about my NIDS cinderella) was taken, so I'll drive to Luxemburg (with Christian Mock and some other Quintessenz guys), do a lecture and have a good time. What I still have to do is to correct a few things in the slides I used at Chaos Communication Camp, since a number of things changed since then (e.g. the rewrite in C wasn't even considered during CCCamp). I also hope to take a few good pictures in Luxemburg, to have a few more nice pictures to show here in the blog.