Re: [PATCH 22/22] exec/cpu: Call cpu_exec_realizefn() once in cpu_common_realize()

[Richard Henderson]

On 9/18/23 09:02, Philippe Mathieu-Daudé wrote:
> +++ b/target/hppa/cpu.c
> @@ -121,22 +121,11 @@ void hppa_cpu_do_unaligned_access(CPUState *cs, vaddr 
> addr,
>   
>   static void hppa_cpu_realizefn(DeviceState *dev, Error **errp)
>   {
> -    CPUState *cs = CPU(dev);
>       HPPACPUClass *acc = HPPA_CPU_GET_CLASS(dev);
> -    Error *local_err = NULL;
> -
> -    cpu_exec_realizefn(cs, &local_err);
> -    if (local_err != NULL) {
> -        error_propagate(errp, local_err);
> -        return;
> -    }
>   
>   #ifndef CONFIG_USER_ONLY
> -    {
> -        HPPACPU *cpu = HPPA_CPU(cs);
> -        cpu->alarm_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
> -                                        hppa_cpu_alarm_timer, cpu);
> -    }
> +    cpu->alarm_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
> +                                    hppa_cpu_alarm_timer, HPPA_CPU(dev));
>   #endif
>   
>       acc->parent_realize(dev, errp);

Quoting one example.

In a previous patch you moved cpu_exec_realizefn() earlier, now you're moving it later, 
into parent_realize().  I think the previous patch needs adjustment.


r~