Re: [PATCH v4 2/2] rcu: Simplify the code logic of rcu_init_nohz()

From: Paul E. McKenney
Date: Thu Aug 25 2022 - 12:20:04 EST


On Tue, Aug 23, 2022 at 08:13:00PM +0800, Leizhen (ThunderTown) wrote:
>
>
> On 2022/8/23 0:35, Paul E. McKenney wrote:
> > On Wed, Aug 17, 2022 at 09:42:53AM +0800, Zhen Lei wrote:
> >> When CONFIG_RCU_NOCB_CPU_DEFAULT_ALL=y or CONFIG_NO_HZ_FULL=y, additional
> >> CPUs need to be added to 'rcu_nocb_mask'. But 'rcu_nocb_mask' may be not
> >> available now, due to 'rcu_nocbs' is not specified. Check and initialize
> >> 'rcu_nocb_mask' before using it. This code simplification strictly follows
> >> this logic, compared with old implementations, unnecessary crossovers are
> >> avoided and easy to understand.
> >>
> >> Signed-off-by: Zhen Lei <thunder.leizhen@xxxxxxxxxx>
> >
> > This looks like a nice simplification, but I will wait for your response
> > on Patch 1/1 before trying it out.
>
> How about I post v5 and just do this simplification? Patch 1/2 seems to require
> further discussion and in-depth analysis, which may take a long time.

Excellent choice, thank you!


Thanx, Paul

> >> ---
> >> kernel/rcu/tree_nocb.h | 32 +++++++++-----------------------
> >> 1 file changed, 9 insertions(+), 23 deletions(-)
> >>
> >> diff --git a/kernel/rcu/tree_nocb.h b/kernel/rcu/tree_nocb.h
> >> index ff763e7dc53551f..3c59b12f4465af1 100644
> >> --- a/kernel/rcu/tree_nocb.h
> >> +++ b/kernel/rcu/tree_nocb.h
> >> @@ -1209,44 +1209,30 @@ EXPORT_SYMBOL_GPL(rcu_nocb_cpu_offload);
> >> void __init rcu_init_nohz(void)
> >> {
> >> int cpu;
> >> - bool need_rcu_nocb_mask = false;
> >> - bool offload_all = false;
> >> struct rcu_data *rdp;
> >> + const struct cpumask *cpumask = NULL;
> >>
> >> #if defined(CONFIG_RCU_NOCB_CPU_DEFAULT_ALL)
> >> - if (!cpumask_available(rcu_nocb_mask)) {
> >> - need_rcu_nocb_mask = true;
> >> - offload_all = true;
> >> - }
> >> -#endif /* #if defined(CONFIG_RCU_NOCB_CPU_DEFAULT_ALL) */
> >> -
> >> -#if defined(CONFIG_NO_HZ_FULL)
> >> - if (tick_nohz_full_running && !cpumask_empty(tick_nohz_full_mask)) {
> >> - need_rcu_nocb_mask = true;
> >> - offload_all = false; /* NO_HZ_FULL has its own mask. */
> >> - }
> >> -#endif /* #if defined(CONFIG_NO_HZ_FULL) */
> >> + cpumask = cpu_possible_mask;
> >> +#elif defined(CONFIG_NO_HZ_FULL)
> >> + if (tick_nohz_full_running && !cpumask_empty(tick_nohz_full_mask))
> >> + cpumask = tick_nohz_full_mask;
> >> +#endif
> >>
> >> - if (need_rcu_nocb_mask) {
> >> + if (cpumask) {
> >> if (!cpumask_available(rcu_nocb_mask)) {
> >> if (!zalloc_cpumask_var(&rcu_nocb_mask, GFP_KERNEL)) {
> >> pr_info("rcu_nocb_mask allocation failed, callback offloading disabled.\n");
> >> return;
> >> }
> >> }
> >> +
> >> + cpumask_or(rcu_nocb_mask, rcu_nocb_mask, cpumask);
> >> }
> >>
> >> if (!cpumask_available(rcu_nocb_mask))
> >> return;
> >>
> >> -#if defined(CONFIG_NO_HZ_FULL)
> >> - if (tick_nohz_full_running)
> >> - cpumask_or(rcu_nocb_mask, rcu_nocb_mask, tick_nohz_full_mask);
> >> -#endif /* #if defined(CONFIG_NO_HZ_FULL) */
> >> -
> >> - if (offload_all)
> >> - cpumask_setall(rcu_nocb_mask);
> >> -
> >> if (!cpumask_subset(rcu_nocb_mask, cpu_possible_mask)) {
> >> pr_info("\tNote: kernel parameter 'rcu_nocbs=', 'nohz_full', or 'isolcpus=' contains nonexistent CPUs.\n");
> >> cpumask_and(rcu_nocb_mask, cpu_possible_mask,
> >> --
> >> 2.25.1
> >>
> > .
> >
>
> --
> Regards,
> Zhen Lei