I've started to look at the TCP handling code of the Linux kernel, it's a huge body of code and I need to get a little familiar with it for my upcoming MSc studies.
The professor with whom I'm going to work published his code patches to the kernel and some updated articles so I've got a lot to read about, I intend to organize all my data on a Wiki for which I got a domain HighPerf.net. I'm also looking into keeping the articles that I find, I just know I'll forget where I found them a few months from now, just when I'll need them.
I've gotten some prods from Debian fellows and gotten back on track to fix my packages, I've done most of them, now I've got syscalltrack left, which is up for adoption but no-one wants to handle it.
Kernel modules and patches are the most annoying types of packages, they require quite a bit of work to get them done properly.
During the initial stages of my packaging work I looked around at the code, found some things to improve, and some bugs waiting to happen in sctrace, fixed them, sent them upstream and they were accepted (or so I'm told, didn't check the CVS repository).
I still need to fix some bugs in ChkTeX, a program I maintain, some bugs on it were filed in the Debian BTS, but the code is so annoying there with lots of weird defines, that everytime I touch it I feel the urge to clean this mess, which then goes to waste the time I needed to fix the actual bugs. Hopefully, I'll get it all cleaned up soon enough and fix the bugs themselves.