On 06.05.2025 18:51, Oleksii Kurochko wrote: > @@ -72,6 +72,8 @@ void __init noreturn start_xen(unsigned long bootcpu_id, > > remove_identity_mapping(); > > + smp_prepare_boot_cpu(); > + > set_processor_id(0); > > set_cpuid_to_hartid(0, bootcpu_id);
Is this a good placement? I'd think that smp_prepare_boot_cpu() ought to be permitted to rely on set_processor_id() already having run, for example (even if right now there's no such dependency). Alternatively the set_processor_id() call may want to move into the new function? Jan
