Re: [PATCH v3 24/24] cpus: Remove CPUClass::has_work() handler

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 27/1/25 08:50, Philippe Mathieu-Daudé wrote:
> On 26/1/25 13:31, Richard Henderson wrote:
>> On 1/25/25 09:01, Philippe Mathieu-Daudé wrote:
>>> diff --git a/cpu-target.c b/cpu-target.c
>>> index 98e9e7cc4a1..778f622b07a 100644
>>> --- a/cpu-target.c
>>> +++ b/cpu-target.c
>>> @@ -230,6 +230,14 @@ void cpu_class_init_props(DeviceClass *dc)
>>>       device_class_set_props(dc, cpu_common_props);
>>>   }
>>> +void cpu_exec_class_post_init(CPUClass *cc, void *data)
>>> +{
>>> +#ifndef CONFIG_USER_ONLY
>>> +    /* Check mandatory SysemuCPUOps handlers */
>>> +    g_assert(cc->sysemu_ops->has_work);
>>> +#endif
>>> +}
>>> +
>>
>> Does this really need to be split from...
>>
> 
>  > diff --git a/hw/core/cpu-common.c b/hw/core/cpu-common.c
>  >
>>> @@ -319,6 +313,11 @@ static void cpu_common_class_init(ObjectClass 
>>> *klass, void *data)
>>>       dc->user_creatable = false;
>>>   }
>>> +static void cpu_common_post_class_init(ObjectClass *klass, void *data)
>>> +{
>>> +    cpu_exec_class_post_init(CPU_CLASS(klass), data);
>>> +}
>>
>> ... here?
> 
> cpu-target.c is in specific_ss[], cpu-common.c in common_ss[]
> where CONFIG_USER_ONLY is not available thus defined.
> 
> We could define cpu_common_post_class_init() in cpu-target.c,
> which is odd because not common; or name it differently 
> (cpu_exec_post_class_init?) and have it registered as common
> post_init handler but again it seems a dirty API mismatch, so
> I went this way, hoping that eventually cpu-common.c ends
> better abstracted and cpu-target.c disappearing...

I'll amend to the description:

---
Note, since cpu-common.c is in meson's common_ss[] source set, we
must define cpu_exec_class_post_init() in cpu-target.c (which is
in the specific_ss[] source set) to have CONFIG_USER_ONLY defined.
---