We lost power this morning. It shouldn't have mattered to me, since I was working on the iBook at the time. Just to be safe, I went upstairs to perform a graceful shutdown on my desktop machine (attached to a UPS).
The UPS did work, except that it wasn't exactly uninterruptable. As I got to the console, the machine was in the process of rebooting off of battery power. So I did a graceful recovery-and-shutdown.
About 45 minutes later, power came back. I tried to reboot, only to find that FreeBSD was barfing on the same kernel panics I thought I had fixed one week ago (trying to set a value to the net.link.ether.bridge_cfg sysctl variable was supposed to configure ethernet bridging, but instead caused a kernel panic).
After an hour of booting, crashing and rebooting, (when nothing looked broken, yet the kernel panics occurred practically on demand), I saw that there were some extra messages in the boot log that weren't echoed to the console during boot. It turns out that the bootloader was loading the bridge module (and presumably failing), and the bridge module was failing to load properly during normal boot, which would explain why the system would crash when trying to set one of its sysctl variables.
I'm not sure how it got there, but the line in /boot/loader.conf that said bridge_load="YES" seems like the most likely cause of my recent unstability. Once I commented it out, there were no more module load failures, and no more kernel panics.
Two+ hours down. Now back to work...
