On Tue, 24 Jan 2017 10:09:25 +0000
Marc-André Lureau <marcandre.lur...@gmail.com> wrote:

>
> diff --git a/exec.c b/exec.c
> index 47835c1dc1..66f8187281 100644
> --- a/exec.c
> +++ b/exec.c
> @@ -672,9 +672,11 @@ void cpu_exec_unrealizefn(CPUState *cpu)
>      if (cc->vmsd != NULL) {
>          vmstate_unregister(NULL, cc->vmsd, cpu);
>      }
> +#ifndef CONFIG_USER_ONLY
>      if (qdev_get_vmsd(DEVICE(cpu)) == NULL) {
>          vmstate_unregister(NULL, &vmstate_cpu_common, cpu);
>      }
> +#endif
> 
> 
> Shouldn't this #ifndef block also include the vmsd
> vmstate_unregister() above? That would be matching
> cpu_exec_realizefn() #ifndef.
> 

IIRC the idea here was to break a depency introduced by
"vmstate_cpu_common".
I'll include the previous if statement in the `ifndef`.

Let me know if I missed a comment on this patch.

--
Alessandro Di Federico
PhD student at Politecnico di Milano

Reply via email to