Re: [PATCH v2] arm64: topology: add source check in arch_cpu_idle_enter()

On 10/08/26 15:50, seanwang1 wrote:
> External email: Use caution opening links or attachments
>
>
> arch_cpu_idle_enter() directly calls amu_scale_freq_tick() to update
> arch_freq_scale when a CPU enters idle. This bypasses the sft_data
> pointer check that topology_clear_scale_freq_source() relies on.
>
> As a result, even after calling topology_clear_scale_freq_source()
> with SCALE_FREQ_SOURCE_ARCH to disable AMU-based frequency scaling,
> the arch_freq_scale value can still be modified by AMU counters when
> the CPU goes idle through the arch_cpu_idle_enter() path.
>
> Add topology_scale_freq_source() helper to check whether a specific
> frequency scaling source is currently registered for a CPU. Use it
> in arch_cpu_idle_enter() to verify that AMU is the active source
> before calling amu_scale_freq_tick().
>
> This ensures that topology_clear_scale_freq_source() properly
> disables AMU updates in both the tick path (already handled by
> topology_scale_freq_tick()) and the idle path.
>
> Co-developed-by: Xuewen Yan <[email protected]>
> Signed-off-by: Sean Wang <[email protected]>
> Signed-off-by: Xuewen Yan <[email protected]>
> ---
>   arch/arm64/kernel/topology.c  |  3 +++
>   drivers/base/arch_topology.c  | 14 ++++++++++++++
>   include/linux/arch_topology.h |  1 +
>   3 files changed, 18 insertions(+)
>
> diff --git a/arch/arm64/kernel/topology.c b/arch/arm64/kernel/topology.c
> index b32f13358fbb..cae5da68d5ec 100644
> --- a/arch/arm64/kernel/topology.c
> +++ b/arch/arm64/kernel/topology.c
> @@ -173,6 +173,9 @@ void arch_cpu_idle_enter(void)
>          if (!amu_fie_cpu_supported(cpu))
>                  return;
>
> +       if (!topology_scale_freq_source(SCALE_FREQ_SOURCE_ARCH, cpu))
> +               return;
> +
>          /* Kick in AMU update but only if one has not happened already */
>          if (housekeeping_cpu(cpu, HK_TYPE_TICK) &&
>              time_is_before_jiffies(per_cpu(cpu_amu_samples.last_scale_update, cpu)))

How about moving it into the condition below, so we only take the RCU
read side when an update is due rather than on every idle entry?

     if (housekeeping_cpu(cpu, HK_TYPE_TICK) &&
time_is_before_jiffies(per_cpu(cpu_amu_samples.last_scale_update, cpu)) &&
         topology_scale_freq_source(SCALE_FREQ_SOURCE_ARCH, cpu))
         amu_scale_freq_tick();

> diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
> index 8c5e47c28d9a..dfc2574a5588 100644
> --- a/drivers/base/arch_topology.c
> +++ b/drivers/base/arch_topology.c
> @@ -127,6 +127,20 @@ void topology_clear_scale_freq_source(enum scale_freq_source source,
>   }
>   EXPORT_SYMBOL_GPL(topology_clear_scale_freq_source);
>
> +bool topology_scale_freq_source(enum scale_freq_source source, unsigned int cpu)

%s/topology_scale_freq_source/topology_is_scale_freq_source/ ?


> +{
> +       struct scale_freq_data *sfd;
> +       bool ret;
> +
> +       rcu_read_lock_sched();
> +       sfd = rcu_dereference_sched(*per_cpu_ptr(&sft_data, cpu));
> +       ret = (sfd && sfd->source == source);
> +       rcu_read_unlock_sched();
> +
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(topology_scale_freq_source);
> +

Do we need to export it?

Thanks,
Sumit
....