Noites do Pinball no Garoa Hacker Clube

A partir de 3 de Outubro. Semanalmente, toda segunda-feira, às 19h30.

Rua Vitorino Carmilo, 459
Santa Cecília - São Paulo, SP

Coordinating efforts towards free firmware implementations

Some months ago I have started to mantain a wikipage at LibrePlanet.org with a list of device drivers that require non-free firmware. For each device I created a wiki page to aggregate technical information about the device in order to make it easier for us to develop free firmware.

Then, during November I have travelled around Europe and have had the opportunity to hack on a laser projector in the Prague hackerspace. The special thing about it is that we developed free software for it but we still rely on the binary-only firmware to startup the device. And I was annoyed by that so I continued investigating and ended up discovering that the device has an EZUSB (which is a 8051 microcontroller with native hardware support for USB communication). Then I have found the datasheet for this microcontroller and started to study it. I even used a disassembler in order to inspect the firmware and things started making sense, but I got concerned about the legality of reusing code from this disassembly process.

At that point I figured out that I should check the wiki page again to see which other devices are using the same microcontroller because that was valuable knowledge that I had just acquired that could be used to free some firmware. So I searched for "ezusb" in the kernel source code and found 3 devices that also use it. One of these devices (DABUSB/TerraTec-DrBox1: a digital radio receiver) has its binary firmware distributed under copyright terms that allow free, derivative works:

* Copyright (C) 1999 BayCom GmbH
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that redistributions of source
* code retain the above copyright notice and this comment without
* modification.

So this means that by disassembling the firmware, the resulting source code is free. I've done it and now we just have to understand the code and give nice names to the variables and routines in order to have a free firmware for the device. I've been doing it yesterday and today but then I stopped for a while and started to question myself whether it is actually useful work. I am concerned that this is a device released 11 years ago and I don't know anybody who uses it nowadays. So I don't know what is the real demand of support for this specific device.

What I mean is that we need better understanding of the meaning of the LinuxLibre blacklist of devices. We need to figure out a way to sort these devices in a relevance order so that we can focus our reverse engineering efforts on the most critical ones. I suggest that we start a working group to define the criteria for sorting that list and continue this work with some better strategies. One of the possible ways to sort these devices by relevance would be to build statistics based on collecting hardware information of libre-distro users.

I hope someday we'll be providing fixes to the issues raised by the LinuxLibre project instead of simply removing functionality. I think that this perception that lots of people have of LinuxLibre as a less functional, technically inferior alternative to Linux is a real danger. I know that we should not judge software only by its technical aspects. Nevertheless, this is still no good advertising for LinuxLibre.

I'd like to hear your opinions about these questions.

Happy Hacking,
Felipe Sanches

Brmlab & Lasers!

I just got back to my home in Brazil yesterday. Travelling in Europe was great and I'd like to share here the details of some cool things I've done in my last week there.

My last stop in Europe was Prague and there was a special reason for that. Three months ago I received an email from Cestmir Houska, one of the members of Brmlab - the Prague hackerspace. Cestmir was contacting me because he got inspired by my photos and blogposts about the Party Land Remake project. He told me that building a pinball machine was one of his dreams from childhood (which happens to be my case also) and that he wanted to keep in contact so that we can help each other in making it become a reality.

Since we are both members of recently inaugurated hackerspaces we figured out that our pinball projects could be understood as a healthy collaboration between them, mutually. So we are planning to design, document and simultaneously build two pinball machines: one in Garoa Hacker Clube, in São Paulo, Brazil; and the other one in Brmlab, in Prague, Czech Republic.

So, since I had the opportunity to visit Europe again due to the Mozilla Drumbeat Festival in Barcelona, I decided to schedule a visit to Prague (which resulted in my visit to Metalab in Vienna also).

In one of the days that I was there at Brmlab, one guy showed me his personal project: a homebrew laser projector. It is composed of 3 lasers (Red, Green and Blue) and a pair of mirrors that are rotated by motors so that the laser beam aims at the wall in different possible angles. By properly controlling those lasers and mirrors, one can draw colorful shapes on the wall. The effect is even more interesting if you add some fog to the room :-)

So I promptly asked him what software he was using to run the laser projector and the answer was: iShow, a proprietary software. I still inquired about free software implementations of drivers or controlling software but he was not aware of any existent efforts. So we decided to start doing that right away. We installed an USB sniffer on his Windows machine and started reverse engineering the protocol used by the device. In half an hour we had our first line drawn by sending URBs through a python script running on my GNU/Linux. Then we spent more 3 or 4 days improving the python module (all source code is in my personal svn repository at Google Code) and implementing some demos:

GML, SVG and ILDA

We wanted to display the Brmlab logo, so I decided to do that by implementing a simple SVG loader. It parses the SVG file and reads the d attribute of all tags. Then each path command from the SVG path description results in the respective drawing routines for our laser device.

The result can be seen in this photo:



Then I remembered GML! Grafitti Markup Language is an XML specification for storing grafitti data. That basically means x,y,z and time coordinates. I have used a python module called PyGML to render some GML files with our laser.



Another relevant data format is called ILDA. It is a file format for storing laser-show animations. It was developed by the International Laser Display Association (which explains the acronym ILDA). Using this python module we got some ILDA animations running on our projector :-)

A clock!

This is a simple analog clock displaying the current local time.

A pong game

Cestmir has implemented a pong game using the pygame library and rendering on the laser display.



When we decided to display scores in the game, we faced the challenge of creating a font for our laser display and font rendering routines. The first 3 charachteres (0, 1 and 2) were designed by imagining the shapes and manually inputing the values of the coordinates of the bezier control points for the glyphs. At that point we decided we should implement a tool... (more on that topic in a future blogpost)

A server for sharing the display

While we were working on this project it became clear that the laser was a scarce resource. We had 4 people hacking on it and it was starting to become common to hear the sentence "Can I use the laser now?" (same sentence I've heard a lot in Vienna a week earlier, but refering to a significantly different machine...).

It reminded me of stories about the early days in computing when people disputed CPU time in universities and research institutes computers. And then multi-user time-sharing systems were developed. So I decided to develop something analogous for our laser display: a server that collects rendering requests from clients and manages to merge them on the real device.

The server receives commands through telnet and manages individual double buffering for each connected user. When a user disconnects, only his/her portion of the drawing disappears from the screen. I also adapted our python module so that it would delegate rendering to a remote server when you pass an IP address to its contructor; otherwise it would render in a local device connected to the USB bus.

That way we could have several users rendering content to the display simultaneously. One possible use of that would be running a script that draws a clock all the time on a corner of the display screen. Here is an example of that (the clock script and another scipt drawing some bezier curves been run remotely by two different users simultaneously):

27C3

Several people from the Brmlab hackerspace will attend the 27th Chaos Communication Congress (27C3) and they are planning to present the laser projector there!

Pinball Hacking in Vienna

I am currently in Vienna, Austria, visiting the Metalab Hackerspace. They have a very nice EPILOG 36EXT laser cutter there and I couldn't avoid using it to cut more parts for my pinball project! So I focused on the upper playfield (ramps, loops, P-U-K-E platform...).

This is what those parts are supposed to look like in the end:



And you can see how they look like now in these photos:





I still need to apply artwork to these acrylic parts. The stickers that I'll use in the project are these:



I also presented a speech about the project trying to get more people interested in hacking on pinball machines DIY manufacturing fun.

Then I also got the opportunity to help on the Roböxotica festival by laser-cutting the wood trophies for the cocktail robots contest!

Another trip to Europe

I am in Europe again! I arrived here las November 2nd to attend the Mozilla Drumbeat Festival, and now I am touring around until December 12th.

In Barcelona I had the pleasure of visiting La Sagrada Familia. It is an amazing place, even for a non-religious person like me. I also met Dave and we worked a bit more on our Firefox addon.

I also visited the Barcelona FabLab and discovered that there's a chance we might have a FabLab in University of São Paulo in the near future. The FabLab was filled with wood/acrylic sculptures of various complex shapes that were built by employing digital fabrication techniques and tools such as laser cutters, 3d printers and CNC machines. Very exciting!

During my stay in Barcelona I was hosted by a very friendly couple in the Valldoreix neighbourhood. Train rides to Barcelona took something around 40 minutes from there.



Now I am in Nice, France. I was planning to visit a Torino hackerspace but got not much response from those guys. So I changed plans and decided to go directly to Milano, instead. But the trains would stop in Nice, France. So I contacted the Nice hackerspace (NICELAB) and discovered that they do not exist yet! Their first meeting with the purpose of organizing a hackerspace was exactly in the same night I would be visiting the city (yesterday night) :-D!!! So I joined them at a bar for some interesting conversation and then slept in the house of one of them. Today I am going to get a train to Milano.



While travelling from Barcelona to Nice, I passed by a train station called Beziérs. As I am deeply involved in vector graphics software development, I could not avoid getting a picture. For a while I wondered whether the name could have some real relation with the beziér curves, but I think that it is just a coincidence.

New York, September 2010

Tonight I'm flying to New York city and I'll stay there until October 3rd.

I'm going to attend the Open Video Conference, the Open Subtitles Design Summit, the Maker Faire and I'll try to visit the NYCResistor hackerspace. I will also participate in a party to "celebrate" the 15 years of the release of "Hackers", the movie.


NYC Skyline by =sp1te

My few last days in Europe were spent attending a demoparty in
Maaseik, Belgium. I always wanted to be in a demoparty and this was my
first opportunity to do so. I didn't want to simply attend it. I
wanted to actually release a demo! My choice of development platform
was the Phillips Videopac G7000 (european name), also known as
Magnavox Odyssey² (in USA). In Brazil it was released as "Odyssey" since the original one has never been released here, so the number 2 in the name would not make much sense in the brazilian market.



I've had one of these in Brazil since my childhood, but I did not bring
it with me to my Europe trip (as I didn't want to risk loosing it or
damaging it for whatever reason). Because of that I tried to buy a
Videopac on eBay and have it delivered to Brussels, where my
brazillian friend Leandro Lameiro lives nowadays, so that we could
develop and showcase the demo in a real machine during Rewired 2010.
But then, unfortunately, the french guy who sold me a Phillips c52
didn't send the item at all. Also, I have contacted Sören Gust,
trying to buy a development cartridge from him, but he didn't have a
spare one to sell. I though about building one myself in some european
hackerspace (I have visited many of them during june), but then the
fact that the real console was not delivered made me give up and
simply focus on emulator-based development.

I knew that it would potentially not be the same as developing in the
real machine, but I needed to release something at the demoparty and I
wanted to do something wild instead of simply making a PC demo. I have
never coded demos before, so I was not expecting to do something
awesome. So, my strategy was to do something interesting based on the
novelty of it instead of technical sophistication. Indeed, this was
(as far as I can tell) the first demo ever for the Videopac/Odyssey² !

I was happy when the emulator properly rendered the results that I
intended to see when I implemented the Video Display Controller (VDC)
trick to change charactere configuration in the middle of the screen
in order to overcome the hardware restriction of only being able to
use only 12 characteres oncreen at a given time. Initially I expected
that the emulator would simply not emulate the trick at all. When it
showed up, I though it was nice and I supposed that it was the correct
emulation. My friend Leandro Lameiro have commented about the issue of
overlapping characteres (that he had read about in some videopac
forum) but we had no way to test it on a real machine to be sure.



I was also interested in releasing the demo as free software, because
I am not happy with the overall secrecy aspect of the demoscene. I
would like to encourage more people to learn how do code demos and
then I think it is essential that we release our source codes and let
people learn! So, now that Mark Guttenbrunner (a.k.a. "manopac") has contacted me by email pointing out this
discussion on the Netherlands Videopac Forum, it makes me really proud
to know that the source code was actually checked out by at least one
person who found it useful.

I'd like to give a brief explanation of the name of our demoscene group.
When we were planning our participation in this demoparty we have
submited an entry in their website informing our name & country so
that the organizers and other people in the scene could have a general
idea of who would attend. This form obligated us to provide a group
name. But I had no group! I noticed that other people (probably in the
same context) had simply provided "-" as the group name (a dash
charactere). So I just followed the pattern. Then, in the party I have
met Niels van Dijkhuizen (NebulaH) who joined me and my brazilian
friend Leandro Lameiro and we decided to release the demo under the
group name "DashCrew" :-)

There were 7 demos competing at Rewired and our demo ranked 3rd :-D !!!



The entry for this demo at pouet.net is here.