On Wed, Jun 17, 2026 at 08:20:04PM +0100, Kiryl Shutsemau wrote:
> +/*
> + * Raise callback for nmi_trigger_cpumask_backtrace(): signal event 0
> + * at every CPU still pending in @mask. The framework excludes the local
> + * CPU from @mask before calling us.
> + */
> +static void sdei_nmi_raise_backtrace(cpumask_t *mask)
> +{
> + unsigned int cpu;
> +
> + for_each_cpu(cpu, mask)
> + sdei_nmi_fire(cpu);
> +}
Does this need a barrier as in patch 4? Sashiko spotted that
backtrace_mask won't be visible without a dsb(ishst). Again, as per my
comment in patch 4, we get away with this because of the TF-A
implementation but that's not guaranteed (we could ask for the SDEI spec
to be updated assuming all firmware implementations do this).
--
Catalin
