Well, it intermittently hangs the PC. And I mean "have to reach down and hit the power switch" hang.
Since I have a bit of experience with Linux kernel debugging, I figured I would take a look. It's surprising though how much I have taken for granted with my development environment at Netilla. What I thought would be a simple operation was, well, not so much.
First off, there's X11. Having a system with X makes it much more difficult to debug since when it locks up you can't get access to the console. So you need a serial console, which means you need another computer. Having the logs go to a second computer is a good idea anyway, since you can save them to a file (something which you typically can't do on a system that has panic'd)
For me, that meant I needed to:
- Get a USB to serial adapter for my iBook
- Download and install OpenDarwin
- Install minicom
- Find my serial cable
- Find my null modem adapter
- Reconfigure inittab to spawn a console on /dev/ttyS0
- Modify grub to support console on both tty0 and ttyS0 (setting the baud rate appropriately)
- Increasing the kernel log level
- Enable the magic SysRq key
- Download the Suse kernel source
- Figure out where the hell Suse keeps third-party kernel module source code
I am familar with each of these steps since I was responsible for kernel debugging at Netilla, but I guess I didn't appreciate what a pain in the ass it would be to reproduce such an environment in my apartment.
And so I do all this and I am now getting kernel debug output to dump to minicom and save to a file on my iBook. And of course, I have been successfully connected through my wireless card for the last hour and now I can't reproduce the problem. That's just my luck....
