Re: [PATCH v9 4/6] arm/sysctl: Implement cpu hotplug ops

[Jan Beulich <jbeulich@xxxxxxxx>]

On 18.08.2026 10:48, Mykyta Poturai wrote:
> SMT-disable enforcement check is moved into a separate
> architecture-specific function.
> 
> For now this operations only support Arm64. For proper Arm32 support,
> there needs to be a mechanism to free per-cpu page tables, allocated in
> init_domheap_mappings. Also, hotplug is not supported if ITS enabled,
> and partially supported FFA, or TEE is enabled, as they use non-static
> IRQ actions.
> 
> Remove ifdef guards for x86 in flask, as cpu hotplug is now
> supported on more architectures.
> 
> Signed-off-by: Mykyta Poturai <mykyta_poturai@xxxxxxxx>
> ---
> v8->v9:
> * simplify dependencies of config CPU_ONLINE_OFFLINE again

You must also have re-based over XSA-499. As non-trivial re-basing can go
wrong, I think at least in those cases it wants noting in the revlog.
(Personally I try to remember to always add the remark, even if the re-base
was the trivial resolution of e.g. fuzz.)

> @@ -104,6 +105,40 @@ void smp_call_function_interrupt(void)
>      irq_exit();
>  }
>  
> +#ifdef CONFIG_CPU_ONLINE_OFFLINE
> +long cf_check cpu_up_helper(void *data)
> +{
> +    unsigned int cpu = (unsigned long)data;
> +    int ret = cpu_up(cpu);
> +
> +    /* Have one more go on EBUSY. */
> +    if ( ret == -EBUSY )
> +        ret = cpu_up(cpu);
> +
> +    if ( !ret && !arch_cpu_can_stay_online(cpu) )

I'm sorry, I should have noticed this already before: This isn't "can", at least
on x86. It is a requirement in certain situations that CPUs be kept kind-of-
online. Since arch_cpu_must_stay_online() feels clumsy as a name, and since that
might also conflict with another arch perhaps really meaning "can", not "must",
maybe arch_cpu_keep_online()?

> --- a/xen/common/sysctl.c
> +++ b/xen/common/sysctl.c
> @@ -475,6 +475,40 @@ long do_sysctl(XEN_GUEST_HANDLE_PARAM(xen_sysctl_t) 
> u_sysctl)
>              copyback = 1;
>          break;
>  
> +    case XEN_SYSCTL_cpu_hotplug:
> +    {
> +        unsigned int hp_op = op->u.cpu_hotplug.op;

This variable is used ...

> +        long (*fn)(void *data);
> +        void *hcpu = _p(op->u.cpu_hotplug.cpu);
> +
> +        ret = -EOPNOTSUPP;
> +        if ( !IS_ENABLED(CONFIG_CPU_ONLINE_OFFLINE) )
> +            break;
> +
> +        switch ( hp_op )

... solely here. Is such a variable really warranted?

> +        {
> +        case XEN_SYSCTL_CPU_HOTPLUG_ONLINE:
> +            fn = cpu_up_helper;
> +            break;
> +
> +        case XEN_SYSCTL_CPU_HOTPLUG_OFFLINE:
> +            fn = cpu_down_helper;
> +            break;
> +
> +        default:
> +            fn = NULL;
> +            break;
> +        }
> +
> +        if ( fn )
> +        {
> +            ret = continue_hypercall_on_cpu(0, fn, hcpu);

Same (maybe to slightly lesser degree) for "hcpu", used solely here.

For both variables, the situation is/was different in the original code.

Jan