ABIT (BP6?) dual-Celery motherboard (two 500 Mhz Celeries). SMP kernel. Apparently, the shutdown cycle disables the power switch! After the "Power Down" state, the power switch does not work :-) The only way to kill power to the system is to either: A) Physically yank the power cord out. B) Three-fingered salute, then turn the power off during the power-on self test.
Halt does not poweroff also on a Toshiba 320cds, kernel 2.2.16, tried both with and without realmode apm calls.
Problem still exists in beta4, but I confirmed that holding the power button for about five seconds will end up in the power being removed. Perhaps close this bug, and add a note to the release errata? This doesn't seem to be a big problem.
In a laptop with the power button configured to suspend and not to power off, the only option would be to yank the batteries out of their place. There must be a better option.
You can call 'halt' instead of 'poweroff', and it should not try to turn off the power.
I think the problem for atleast borgia is that he calls halt and not poweroff, hint for those above who are having trouble with halt not powering of there atx machine. Use poweroff or halt -p. halt now does just that: halt the machine (take a look at the last message printed by the kernel, if it is System halted then you should call poweroff or halt -p) If you are indead halting your machine then it is not unusal for the atx powerbutton to need 5 seconds, depending on the mb it needs that under plain dos too for example.