On 05/08/2018 09:48 AM, Peter Maydell wrote:
> (A v8.x implementation can include any arbitrary
> subset of the v8.(x+1) features, so this is a legal combination.)
I didn't realize this was a possibility.
> In fact we already have a mandatory-in-v8.1 feature:
> ARM_FEATURE_V8_RDM, which we've just given its own feature bit.
Yes, I'd been planning to send follow-up patches to "tidy" this.
> --- a/target/arm/cpu.h
> +++ b/target/arm/cpu.h
> @@ -1449,8 +1449,7 @@ enum arm_features {
> ARM_FEATURE_V8_SHA3, /* implements SHA3 part of v8 Crypto Extensions */
> ARM_FEATURE_V8_SM3, /* implements SM3 part of v8 Crypto Extensions */
> ARM_FEATURE_V8_SM4, /* implements SM4 part of v8 Crypto Extensions */
> - ARM_FEATURE_V8_1,
> - ARM_FEATURE_V8_ATOMICS = ARM_FEATURE_V8_1, /* mandatory extension */
> + ARM_FEATURE_V8_ATOMICS, /* ARMv8.1-Atomics feature */
> ARM_FEATURE_V8_RDM, /* implements v8.1 simd round multiply */
> ARM_FEATURE_V8_FP16, /* implements v8.2 half-precision float */
> ARM_FEATURE_V8_FCMA, /* has complex number part of v8.3 extensions. */
>
> Is that OK?
Yes, that's fine. Thanks.
> We can have a separate patch which adds ARM_FEATURE_V8_1 and the
> code in realizefn to do
> if (arm_feature(env, ARM_FEATURE_V8_1)) {
> set_feature(env, ARM_FEATURE_V8);
> set_feature(env, ARM_FEATURE_V8_ATOMICS);
> set_feature(env, ARM_FEATURE_V8_RDM);
> }
Yep.
r~
