Re: [PATCH] Fix emergency_restart (sysrq-b) with kvm loaded on Intel hosts

[Avi Kivity <avi@qumranet.com>]

Ingo Molnar wrote:
> * Avi Kivity <avi@qumranet.com> wrote:
>
>   
>>> reboot was always a bit fragile - i think we should only do that if 
>>> we find a box where the FADT reset works better than the first-wave 
>>> approaches we try.
>>>       
>> It worked on my host.  Since it will fall back to keyboard reset and 
>> triple fault, it seems fairly safe.
>>     
>
> ... except if it hangs in ACPI/SMM code for whatever reason.
>   

acpi reboot doesn't call into the aml interpreter.  It just bangs on a 
port that it reads from a static table.  See acpi_reboot().

It's true that SMM could be set up to intercept that port, but in that 
case, it is even more likely that it intercepts the keyboard controller 
port (to translate usb keyboards etc).

-- 
error compiling committee.c: too many arguments to function