Re: [RFC PATCH 06/11] target/arm: Introduce host_cpu_feature_supported()

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

On 12/8/25 02:44, Richard Henderson wrote:
> On 8/12/25 03:06, Philippe Mathieu-Daudé wrote:
>> +bool host_cpu_feature_supported(enum arm_features feat, bool 
>> can_emulate)
>> +{
>> +#if defined(CONFIG_KVM) || defined(CONFIG_HVF)
>> +    static enum { F_UNKN, F_SUPP, F_UNSUPP } supported[64] = { };
>> +
>> +    assert(feat < ARRAY_SIZE(supported));
>> +    if (supported[feat] == F_UNKN) {
>> +        supported[feat] = arm_hw_accel_cpu_feature_supported(feat, 
>> can_emulate);
>> +    }
>> +    return supported[feat] == F_SUPP;
>> +#elif defined(CONFIG_TCG)
>> +    return can_emulate;
>> +#else
>> +#error
>> +#endif
>> +}
> 
> How much startup time is saved by this cache?
> This feels like premature optimization.

It is called for each PSCI call here:

https://lore.kernel.org/qemu-devel/20250808070137.48716-6-mohamed@unpredictable.fr/

+static int hvf_psci_get_target_el(void)
+{
+    if (hvf_arm_el2_enabled()) {
+        return 2;
+    }
+    return 1;
+}

But as you mentioned in patch #9 "by the time we get [here] we have
already detected the host cpu features and have updated env to match",
so checking with:

   arm_feature(env, ARM_FEATURE_EL2)

should be sufficient.