Re: [PATCH v2 05/12] smp: Free call_function_data via RCU in smpcfd_dead_cpu

[Muchun Song]

> On Mar 2, 2026, at 15:52, Chuyi Zhou <zhouchuyi@xxxxxxxxxxxxx> wrote:
> 
> Use rcu_read_lock to protect the hole scope of smp_call_function_many_cond
                                  ^
                                 whole

BTW, The protection we often refer to actually pertains to data, not code.
Therefore, the rcu_read_lock() here is actually preventing smp_call_function_many_cond()
from accessing percpu csd data that has already been freed (which is
what smpcfd_dead_cpu() does).

> and wait for all read critical sections to exit before releasing percpu csd
> data. This is a preparation for enabling preemption during csd_lock_wait().
> 
> Signed-off-by: Chuyi Zhou <zhouchuyi@xxxxxxxxxxxxx>
> ---
> kernel/smp.c | 3 +++
> 1 file changed, 3 insertions(+)
> 
> diff --git a/kernel/smp.c b/kernel/smp.c
> index 9728ba55944d..ad6073b71bbd 100644
> --- a/kernel/smp.c
> +++ b/kernel/smp.c
> @@ -77,6 +77,7 @@ int smpcfd_dead_cpu(unsigned int cpu)
> {
> 	struct call_function_data *cfd = &per_cpu(cfd_data, cpu);
> 
> + 	synchronize_rcu();
> 	free_cpumask_var(cfd->cpumask);
> 	free_cpumask_var(cfd->cpumask_ipi);
> 	free_percpu(cfd->csd);
> @@ -810,6 +811,7 @@ static void smp_call_function_many_cond(const struct cpumask *mask,
> 
> 	lockdep_assert_preemption_disabled();
> 
> + 	rcu_read_lock();
> 	cfd = this_cpu_ptr(&cfd_data);
> 	cpumask = cfd->cpumask;
> 
> @@ -907,6 +909,7 @@ static void smp_call_function_many_cond(const struct cpumask *mask,
> 
> 	if (preemptible_wait)
> 		free_cpumask_var(cpumask_stack);
> + 	rcu_read_unlock();

We could call rcu_read_unlock() before free_cpumask_var() above.

> }
> 
> /**
> -- 
> 2.20.1