Le 28/04/2020 à 15:19, KONRAD Frederic a écrit :
>
>
> Le 4/27/20 à 9:53 AM, Laurent Vivier a écrit :
>> Le 20/04/2020 à 16:01, frederic.kon...@adacore.com a écrit :
>>> From: KONRAD Frederic <frederic.kon...@adacore.com>
...
>> I would prefer to have two macros with no class parameter, something
>> like DEFINE_M68K_CPU_TYPE_M68K() and DEFINE_M68K_CPU_TYPE_CF().
>>
>>> #define DEFINE_M68K_CPU_TYPE(cpu_model, initfn) \
>>> { \
>>> .name = M68K_CPU_TYPE_NAME(cpu_model), \
>>> @@ -314,11 +329,16 @@ static const TypeInfo m68k_cpus_type_infos[] = {
>>> .class_size = sizeof(M68kCPUClass),
>>> .class_init = m68k_cpu_class_init,
>>> },
>>> - DEFINE_M68K_CPU_TYPE("m68000", m68000_cpu_initfn),
>>> - DEFINE_M68K_CPU_TYPE("m68020", m68020_cpu_initfn),
>>> - DEFINE_M68K_CPU_TYPE("m68030", m68030_cpu_initfn),
>>> - DEFINE_M68K_CPU_TYPE("m68040", m68040_cpu_initfn),
>>> - DEFINE_M68K_CPU_TYPE("m68060", m68060_cpu_initfn),
>>> + DEFINE_M68K_CPU_TYPE_WITH_CLASS("m68000", m68000_cpu_initfn,
>>> + m68k_cpu_class_init_m68k_core),
>>> + DEFINE_M68K_CPU_TYPE_WITH_CLASS("m68020", m68020_cpu_initfn,
>>> + m68k_cpu_class_init_m68k_core),
>>> + DEFINE_M68K_CPU_TYPE_WITH_CLASS("m68030", m68030_cpu_initfn,
>>> + m68k_cpu_class_init_m68k_core),
>>> + DEFINE_M68K_CPU_TYPE_WITH_CLASS("m68040", m68040_cpu_initfn,
>>> + m68k_cpu_class_init_m68k_core),
>>> + DEFINE_M68K_CPU_TYPE_WITH_CLASS("m68060", m68060_cpu_initfn,
>>> + m68k_cpu_class_init_m68k_core),
>>> DEFINE_M68K_CPU_TYPE("m5206", m5206_cpu_initfn),
>>> DEFINE_M68K_CPU_TYPE("m5208", m5208_cpu_initfn),
>>> DEFINE_M68K_CPU_TYPE("cfv4e", cfv4e_cpu_initfn),
>
> But what about the "any" which is out of context here?
>
> DEFINE_M68K_CPU_TYPE("any", any_cpu_initfn),
>
> Should it be TYPE_M68K or TYPE_CF in which case which xml will it take?
> I guess
> we can take TYPE_CF so it doesn't change from what is done today.
>
Yes, "any" has been introduced with coldfire CPUs and defines CF_FPU, so
it uses 64bit float registers.
Thanks,
Laurent
