Re: [Xen-devel] [PATCH v6] xen/arm : emulation of arm's PSCI v0.2 standard

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Sat, 2014-07-26 at 01:22 +0530, Parth Dixit wrote:
> +    case PSCI_cpu_off:
> +        {
> +            uint32_t pstate = PSCI_ARG32(regs,1);
> +            PSCI_RESULT_REG(regs) = (int32_t) do_psci_cpu_off(pstate);

do_psci_cpu_off returns int32_t, so no need for this cast. The same is
true of all of the helpers, I think. Except perhaps MIGRATE_INFO_UP_CPU
which differs between 32- and 64-bit.

IOW I think this...

> +    case PSCI_0_2_FN_MIGRATE_INFO_UP_CPU:
> +        if ( psci_mode_check(current->domain, fid) )
> +            PSCI_RESULT_REG(regs) =
> +               (uint32_t) do_psci_0_2_migrate_info_up_cpu();

... is the only cast which might need to remain, and even there I'm not
sure.

do_psci_0_2_migrate_info_up_cpu() returns a register_t, which given the
currently trivial implementation I think is likely to be sufficient
without the cast.

> +register_t do_psci_0_2_cpu_suspend(uint32_t power_state, register_t 
> entry_point,
> +                            register_t context_id)
> +{
> +    struct vcpu *v = current;
> +
> +    /* affinity values are ignored in this implementation as
> +     * at present xen does not supports affinity level greater
> +     * than 0, for all  affinity values passed we power down/ standby
> +     * the current core */
> +
> +    vcpu_block_unless_event_pending(v);
> +
> +    /* if power down is requested context_id must be returned in x0/r0
> +     * as the function return will be written in x0/r0 we return context_id
> +     * instead os success value */

The spec isn't clear on this, but I don't think this can be correct.

I expect the requirement is that you either return PSCI_SUCCESS to the
instruction after the smc or you return to entry_point with
x0==context_id. I don't think it makes sense to return after the smc
with the x0==context_id. This is implied by section 5.4.2 bullet #4
which talks about handling PSCI return codes.

I'm going to ask ARM for clarification, please stand by.

Ian.



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel