Contributed by weerd on from the arming-the-zoo dept.
Next up in our line of hackathon reports is Brandon Mercer (bmercer@). Brandon is a relatively new addition to the team, focusing mainly on the arm port:
In the weeks leading up to the hackathon there was a lot of work going on in the beagle port. We wanted to get things tested and into the tree so that by the time the hackathon rolled around we could be cleaning things up and focusing on getting some snapshots put together. On the first day I was able to do a make build and make release and got some sets to look at. It was fairly satisfying to have them sitting there, however we still needed a working ramdisk with ethernet and a good way to install. There had been lots of emails and discussion going on about the best way to do this and after a quick talk with Theo it all became clear how to proceed. I hacked up the miniroot and got it in so the other arm guys could start working with it.
The days started running together but I'm fairly certain that I tried to fix the ramdisk next. I was building the ramdisk and releases on the pandaboard ES at this point but the ramdisk would not boot on anything. It was slightly frustrating to run into this along with broken builds and other fun things that come on with an ever changing tree. On the flip side, there are a few other guys working on arm right now and even though they weren't at the hackathon, diffs and fixes came in that got things back on track quickly! In order to make an install useful we needed to get networking working which requires some USB work that patrick@ did that still isn't quite ready to go in. Florian@ created a ramdisk with the usb diff applied so we could test the installation process. It worked remarkably well! There were several install tests, and iterations of building sets while we ironed out a few more things... most notably fdisk did not want to play nice when we were creating our miniroot which led to easy mistakes during the install process.
Since Patrick@ was working on his usb I decided to tinker with the networking on the beaglebone black. The network does not hang off of usb like it does on the pandaboard. I began porting the netbsd cpsw driver over. It took me an entire day just to get it to build and even then the driver was in pretty ugly shape. I took the next morning to clean things up and do some more testing. Rapha@ said he tweaked a couple things and got the device to try and attach. I added those changes in and saw the same thing. It immediately panics on a bogus interrupt which was to be expected. It was very satisfying to port my first network driver (if not completely eye opening), and have it actually try and attach the device. There is a bit of work to be done on the driver still, but not a few hours after it went in, improvements were committed by others. It's great to see!
I'd like to thank everyone who helped put this hackathon together. It was a great success and everyone thoroughly enjoyed themselves!
Many thanks to Brandon for his report. We're looking forward to seeing a snapshot to install on our beagles and pandas!
(Comments are closed)