Re: [PATCH V6 03/14] cpus: stop vm in suspended runstate

[Peter Xu <peterx@redhat.com>]

On Fri, Dec 01, 2023 at 12:11:32PM -0500, Steven Sistare wrote:
> >> diff --git a/include/sysemu/runstate.h b/include/sysemu/runstate.h
> >> index f6a337b..1d6828f 100644
> >> --- a/include/sysemu/runstate.h
> >> +++ b/include/sysemu/runstate.h
> >> @@ -40,6 +40,11 @@ static inline bool shutdown_caused_by_guest(ShutdownCause cause)
> >>      return cause >= SHUTDOWN_CAUSE_GUEST_SHUTDOWN;
> >>  }
> >>  
> >> +static inline bool runstate_is_started(RunState state)
> > 
> > Would runstate_has_vm_running() sound better?  It is a bit awkward when
> > saying something like "start a runstate".
> 
> I have been searching for the perfect name for this accessor.
> IMO using "running" in this accessor is confusing because it applies to both
> the running and suspended state.  So, I invented a new aggregate state called
> started.  vm_start transitions the machine to a started state.
> 
> How about runstate_was_started?  It works well at both start and stop call sites:
> 
>     void vm_resume(RunState state)
>         if (runstate_was_started(state)) {

This one looks fine, but...

>             vm_start();
> 
>     int vm_stop_force_state(RunState state)
>         if (runstate_was_started(runstate_get())) {

.. this one makes the past tense not looking good.

>             return vm_stop(state);

How about runstate_is_alive()?  So far the best I can come up with. :)

Even if you prefer "started", I'd vote for not using past tense, hence
runstate_is_started().

Thanks,

-- 
Peter Xu