Paul Brook wrote:
> On Saturday 02 May 2009, Jan Kiszka wrote:
>> Paul Brook wrote:
>>> On Friday 01 May 2009, Jan Kiszka wrote:
>>>> Add the parameter 'order' to qemu_register_reset and sort callbacks on
>>>> registration. On system reset, callbacks with lower order will be
>>>> invoked before those with higher order. Update all existing users to the
>>>> standard order 0.
>>>>
>>>> Note: At least for x86, the existing users seem to assume that handlers
>>>> are called in their registration order. Therefore, the patch preserves
>>>> this property. If someone feels bored, (s)he could try to identify this
>>>> dependency and express it properly on callback registration.
>>> Why do we need this? Why isn't creation order good enough?
>> At latest when properly deregistering reset handlers again on device
>> unplug, the registration order is no longer a static thing, manifested
>> in the code organization - which can also break due to refactoring, BTW.
> 
> I'm afraid I can't make any sense of this. What exactly are you trying to 
> solve?

Thinking about it again, device hot-plugging was a bad example as its
natural order usually also ensures the right reset order.

The problem I'm trying to solve are tricky dependencies on x86 between
CPU, APIC, and KVM's VCPU. But I got an idea how to solve it differently.

Jan