Re: [RFC PATCH v4 11/19] hw/core/machine: Allow dynamic registration of valid CPU types

[Pierrick Bouvier <pierrick.bouvier@linaro.org>]

On 4/22/25 10:54, Richard Henderson wrote:
> On 4/22/25 07:54, Philippe Mathieu-Daudé wrote:
>> index f52a4f2273b..8b40735ef98 100644
>> --- a/hw/core/machine.c
>> +++ b/hw/core/machine.c
>> @@ -1581,6 +1581,33 @@ static bool is_cpu_type_supported(const MachineState *machine, Error **errp)
>>                return false;
>>            }
>>        }
>> +    if (mc->get_valid_cpu_types) {
>> +        GSList *vct = mc->get_valid_cpu_types(machine);
>> +        bool valid = false;
>> +        unsigned count = 0;
>> +        GSList *l;
>> +
>> +        for (l = vct; !valid && l != NULL; l = l->next) {
>> +            valid |= !!object_class_dynamic_cast(oc, l->data);
>> +            count++;
>> +        }
>> +
>> +        if (!valid) {
>> +            g_autofree char *requested = cpu_model_from_type(machine->cpu_type);
>> +            vct = g_slist_reverse(vct);
>> +            error_setg(errp, "Invalid CPU model: %s", requested);
>> +            error_append_hint(errp, "The valid models are: ");
>> +            for (l = vct; l != NULL; l = l->next) {
>> +                g_autofree char *model = cpu_model_from_type(l->data);
>> +                error_append_hint(errp, "%s%s", model, --count ? ", " : "");
>> +            }
>> +            error_append_hint(errp, "\n");
>> +        }
>> +        g_slist_free_full(vct, g_free);
>> +        if (!valid) {
>> +            return false;
>> +        }
>> +    }
> 
> Why use GSList instead of GPtrArray?
> That would provide you the count without manually computing it,
> and it would avoid the need for any sort of reverse.
> I think it would also allow you to auto-free the set.
> 

Same remark than Richard, it would remove all the checks needed, as we 
can simply use this array "blindly".

> 
> r~