Pare Tech.com Thoughts and Ideas From Somewhere

While working on the Barrelfish OS I needed a better way to interact with SimNow's serial interface. The problem is that SimNow allows you to set up pipes for serial communication to the development system. One file is created for the input and another for the output. Using cat for the output and echo for the output can be a real pain, especially when you want to easily configure a system to use a debugging system like GDB on the same physical machine. Around this time I found a document on the coreboot website that directly address this issue with a simple program called snserial.

At work I have a Samsung SyncMaster 213T that I connect my laptop to as a second display. It has been driving me crazy that I could only select 1024x768 or 1360x768 as screen resolutions while the native resolution is 1600x1200. Not only was this below the native resolution but as a result the displayed images would wiggle and just make my eyes sore by the day's end. The temporary solution to this problem turned out to be very simple.

As of a few weeks ago I was very new to boot loaders. Of course I used them but it was just as a conciquence of installers and that's how a OS set things up. But over the last few weeks I've had to deal with them a lot, booting multiboot compliant operating systems that must be booted over a network. Anyways those tales are for another time but I was starting to work on some kernel development and wanted a quick and easy way to make grub bootable images. I found QEMU to be a great way to do this without requiring another computer, a grub boot disk, etc...

I decided I would make some more of the documentation that is supplied with Barrelfish easier for others to view. The attached pdfs are generated from source from the December 2009 snapshot of Barrelfish. Barrelfish is distributed under a three clause BSD style license and as such I've included the license with the files.

I needed a TFTP/PXE server at home so that I can continue development of Barrelfish under the same conditions that I have at work. This document is a rough outline of the steps I took to complete this goal.

I needed to remove Network-Manager from my Ubuntu setup as I'm getting ready to implement some new network services for creating a test setup for a PXE boot cluster. It is my understanding from a few minutes of googling and one previous attempt that automatic network managers like Network-Manager can conflict with these sorts of setups. Particularly when trying to have both a wired Ethernet connection and a wireless connection active simultaneously.

Here are the exact steps I took on a Toshiba Satellite laptop running Ubunto 9.10. First make sure that you can connect to your wireless network and that you can connect to the internet.

Download a copy of gPXE, formally known as "Etherboot," by your favorite method.

# git clone git://git.etherboot.org/scm/gpxe.git

Of course you could always download an official snapshot release from http://kernel.org/pub/software/utils/boot/gpxe/.

gPXE does have precompiled rom images available at http://rom-o-matic.net/ but for what we're doing that won't work. We actually need to compile our image to be able to embed our payload.

For months, possibly years, I said I wanted to set up an SVN repository on our mediatemple (dv) 3.0 server. Well I finally installed SVN and trac. I wish I did a better job at documenting the actual install but I was a little inebriated at the time so the only documentation I have is a rather lengthing page or random commands and links to other sites. I will say that I used YUM package manager which made things a lot easier for a relatively inexperienced, and at the time inebriated, servadmin like myself.

A little while back ago I approached one of my professors, Dr. Mary Baker, about research lab opportunities for undergraduate students. We talked for a while about various opportunities that existed within the department and if I was far enough along in my studies to where it would be beneficial for all parties. One thing led to another and today I was given my first task in the Neuro-Imaging and Cognition Engineering Lab (NICE).

"The mission of the NICE lab is to develop models of perception, memory, neurological diseases and language as they relate to the underlying structure and neural circuitry of the human brain. It also addresses neurologically brain-damaged individuals and how their impairments reveal fundamental principles of brain development and neuro-cognitive functioning."

So the spring 2009 semester is quickly coming to end here in Lubbock and that means that projects should be nearing completion. This semester we were tasked with building a line following robot controlled by an FPGA, but there was a twist. Instead of traditional line followers that detect the line by optical means we had to build sensors that would detect a line of aluminum tape. I'll have a lot more details over the next week or two but for now I've posted a quick video clip after the break.