So way back in the 2.4 Linux kernel days I operated a shell server. It was kind of like a “proto honeypot”. We set it up to portray a bunch of fake information about itself. We then released it into the wilds of IRC claiming it was a “machine we just owned here have a free account.” Then we monitored the systems use via ttysnoop applications and pulled copies of anything uploaded.
The result was we pulled about 4 gigs of exploit archives. And a bunch of logs of people trying to figure out why their x86 exploits weren’t running on the DEC alpha that was pretending to be the wopr.
Anyways, one of the neat side effects of my playing with the proc file system was finding a limit on the upper bounds of top. By setting up procfs to displays stats for 1048 processors I managed to get top to collapse after one iteration of showing cpu usage stats. Added a fun screenshot to the gallery.
Among the fun easy things you can do to tweak your kernel, upping your jiffy count to present 16 years of uptime, and customizing your HZ values are the most fun. I remember running HZ at 1024 back when the rest of the world was rocking it at 100 and slowing themselves down massively. Kernel tweaks are fun, relatively easy, and a great way to learn. Be fearless, be stupid, and be prepared to watch stuff explode catastrophically. But, don’t be afraid to slash and burn and start anew.
