>>>>> "Jared" == Jared D McNeill <jmcneill@invisible.ca> writes:

    Jared> On Mon, 3 Sep 2007, Markus W Kilbinger wrote:
    >> ppb1 ehci0 ohci1 ohci0 pcib0 ppb0 pchb0 pckbc0 npx0 pcppi0
    >> attimer0 acpibut0 pci0 acpi0 ioapic0 cpu1panic: kernel
    >> diagnostic assertion "mutex_owned(&cpu_lock)" failed: file
    >> "../../../../kern/kern_cpu.c", line 216

    Jared> Hrm, I'll look into this. In the meantime, can you comment
    Jared> out the call to cpu_setonline in arch/i386/i386/cpu.c?

Commenting this function call (and its following err check) out makes
a jmcneill-pm branch kernel cleanly power down.

But: The same does a normal -current kernel built of sources 7 days
     ago!

So, something inbetween broke the power down behavior.

-> Does anybody see the culprit for breaking power down within this
   time window?

Markus.