1. shared page tables support, based on Dave McCracken’s work. wli is planning to do it, question is, can I do it sooner?
2. remove the big kernel lock from trident.c.
3. continue ongoing work with “don’t let slab caches grow beyond a given size”, and sysfsication of slab caches.
More ideas welcome…
Muli Ben-Yehuda's journal 
Get it
1. Enigma based encryption. You open a socket, ioctl it to enigma mode, ioctl the wheels and voila – no one can really understand what you are talking about:)
Of course,there is little security value, but I feel amused:)
2. Raw network character devices. You open /dev/net/raw/ethernet/eth0, you get stream of segments. You open /dev/net/raw/ip/ipv4/192.168.0.1, you get a stream of datagrams.
TCP support should be omitted at first stage, since it is bigger pain in the ass than it may seem.
3. Something. I would like to do something in kernel again, since I feel getting stains of rust instead of wisdom.
Comment by omerm — June 23, 2003 @ 9:02 PM |
The Problem With Linux
I honestly don’t know of something useful Linux doesn’t do. So here is something non-useful that it doesn’t do. Why not rewrite GNU-mach, or L4, to be a kernel module. Then, the HURD can run on top of Linux. It will instantly have access to all of Linux’s hardware support, and installing it would be much easier. Plus, it would be dead easy to play with it.
Comment by moshez — June 24, 2003 @ 10:49 AM |
Re: The Problem With Linux
Sounds like a fun project … for someone else. Everything I’ve heard about the Mach does not make me particularly interested in getting to know it better.
I would love to work with k42 on linux, though. Their build system was a nightmare the last I looked at it and that drove me off, so maybe it’s time to look again.
k42: http://www.research.ibm.com/K42/
Comment by mulix — June 25, 2003 @ 4:50 AM |