Re: [PATCH v2 5/5] [RISCV_PM] Implement address masking functions required for RISC-V Pointer Masking extension

[Alexey Baturo]

> g_assert_not_reached();

Would fix, thanks.

> bool

Would fix.

>!s->pm_enabled

Same.

> Don't need the if. 

Would remove, thanks.

> And should it in fact be placed outside the ifdef? 

Sure, you're right.

Richard, thank you for your time and effort reviewing these changes!

чт, 15 окт. 2020 г. в 20:07, Richard Henderson <richard.henderson@linaro.org>:

On 10/15/20 8:21 AM, Alexey Baturo wrote:
> +        switch (priv) {
> +        case PRV_U:
> +            pm_enabled = env->mmte & U_PM_ENABLE;
> +            break;
> +        case PRV_S:
> +            pm_enabled = env->mmte & S_PM_ENABLE;
> +            break;
> +        case PRV_M:
> +            pm_enabled = env->mmte & M_PM_ENABLE;
> +            break;
> +        default:
> +            assert(0 && "Unreachable");

g_assert_not_reached();

> +    /* PointerMasking extension */
> +    uint8_t pm_enabled;

bool

> +    if (s->pm_enabled == 0) {

!s->pm_enabled

> +    if (riscv_has_ext(env, RVJ)) {
> +        ctx->pm_enabled = FIELD_EX32(tb_flags, TB_FLAGS, PM_ENABLED);
> +        int priv = cpu_mmu_index(env, false);
> +        ctx->pm_mask = pm_mask[priv];
> +        ctx->pm_base = pm_base[priv];
> +    } else {
> +        ctx->pm_enabled = 0;
> +    }

Don't need the if.  And should it in fact be placed outside the ifdef?  This
shouldn't be related to !CONFIG_USER_ONLY here and nowhere else.


r~