I recently upgraded from RedHat Linux 6.0 to 6.2, expecting that the shutdown -h / halt bug would be fixed. I was wrong. I have a K6-2/400 and a TMC ATX motherboard. Every time i try to halt the system, it spits out 2 pages of a core dump: The system is halted. stopping al md devices. Power Down. [...] The kernel is unable to handle virtual paging request at address 0x00008885 [...] Then it continues to dump the core. the address 0x00008885 is the same every time. I don't know why it doesn't work. Its also the same in 6.0 and 6.2.
Your BIOS is crashing after being told to power the machine off. Your options are a) try upgrading the BIOS b) apply an experimental patch that switches to real mode before powering off (it's in 2.2.15pre18) c) remove the '-p' from the last line of /etc/rc.d/init.d/halt, (this tells it not to try to power the machine off.)
I have the same problem, same cpu, Epox EP-MVP3G motherboard 7.16.99 Award BIOS v4.51 PG. Maybe it is a bug in the halt code for power off for an AMD K6-2? Also came up with the same solution (ie. remove -p from halt command). PS MeSsy Win98 shuts down and powers off ok (on those ever rarer occasions it gets the opportunity).