lkml.org 
[lkml]   [2018]   [Jul]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH RFC tip/core/rcu 1/2] rcu: Defer reporting RCU-preempt quiescent states when disabled
Hi Paul,

On Wed, Jun 27, 2018 at 01:49:14PM -0700, Paul E. McKenney wrote:
[...]
> diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
> index c1b17f5b9361..ff5c70eae47d 100644
> --- a/kernel/rcu/tree_plugin.h
> +++ b/kernel/rcu/tree_plugin.h
> @@ -371,6 +371,9 @@ static void rcu_preempt_note_context_switch(bool preempt)
> * behalf of preempted instance of __rcu_read_unlock().
> */
> rcu_read_unlock_special(t);
> + rcu_preempt_deferred_qs(t);
> + } else {
> + rcu_preempt_deferred_qs(t);
> }
>
> /*
> @@ -464,54 +467,51 @@ static bool rcu_preempt_has_tasks(struct rcu_node *rnp)
> }
>
> /*
> - * Handle special cases during rcu_read_unlock(), such as needing to
> - * notify RCU core processing or task having blocked during the RCU
> - * read-side critical section.
> + * Report deferred quiescent states. The deferral time can
> + * be quite short, for example, in the case of the call from
> + * rcu_read_unlock_special().
> */
> -static void rcu_read_unlock_special(struct task_struct *t)
> +static void
> +rcu_preempt_deferred_qs_irqrestore(struct task_struct *t, unsigned long flags)
> {
> bool empty_exp;
> bool empty_norm;
> bool empty_exp_now;
> - unsigned long flags;
> struct list_head *np;
> bool drop_boost_mutex = false;
> struct rcu_data *rdp;
> struct rcu_node *rnp;
> union rcu_special special;
>
> - /* NMI handlers cannot block and cannot safely manipulate state. */
> - if (in_nmi())
> - return;
> -
> - local_irq_save(flags);
> -
> /*
> * If RCU core is waiting for this CPU to exit its critical section,
> * report the fact that it has exited. Because irqs are disabled,
> * t->rcu_read_unlock_special cannot change.
> */
> special = t->rcu_read_unlock_special;
> + rdp = this_cpu_ptr(rcu_state_p->rda);
> + if (!special.s && !rdp->deferred_qs) {
> + local_irq_restore(flags);
> + return;
> + }
> if (special.b.need_qs) {
> rcu_preempt_qs();
> t->rcu_read_unlock_special.b.need_qs = false;
> - if (!t->rcu_read_unlock_special.s) {
> + if (!t->rcu_read_unlock_special.s && !rdp->deferred_qs) {
> local_irq_restore(flags);
> return;
> }
> }
>
> /*
> - * Respond to a request for an expedited grace period, but only if
> - * we were not preempted, meaning that we were running on the same
> - * CPU throughout. If we were preempted, the exp_need_qs flag
> - * would have been cleared at the time of the first preemption,
> - * and the quiescent state would be reported when we were dequeued.
> + * Respond to a request by an expedited grace period for a
> + * quiescent state from this CPU. Note that requests from
> + * tasks are handled when removing the task from the
> + * blocked-tasks list below.
> */
> - if (special.b.exp_need_qs) {
> - WARN_ON_ONCE(special.b.blocked);
> + if (special.b.exp_need_qs || rdp->deferred_qs) {
> t->rcu_read_unlock_special.b.exp_need_qs = false;
> - rdp = this_cpu_ptr(rcu_state_p->rda);
> + rdp->deferred_qs = false;
> rcu_report_exp_rdp(rcu_state_p, rdp, true);
> if (!t->rcu_read_unlock_special.s) {
> local_irq_restore(flags);
> @@ -519,19 +519,6 @@ static void rcu_read_unlock_special(struct task_struct *t)
> }
> }
>
> - /* Hardware IRQ handlers cannot block, complain if they get here. */
> - if (in_irq() || in_serving_softirq()) {
> - lockdep_rcu_suspicious(__FILE__, __LINE__,
> - "rcu_read_unlock() from irq or softirq with blocking in critical section!!!\n");
> - pr_alert("->rcu_read_unlock_special: %#x (b: %d, enq: %d nq: %d)\n",
> - t->rcu_read_unlock_special.s,
> - t->rcu_read_unlock_special.b.blocked,
> - t->rcu_read_unlock_special.b.exp_need_qs,
> - t->rcu_read_unlock_special.b.need_qs);
> - local_irq_restore(flags);
> - return;
> - }
> -
> /* Clean up if blocked during RCU read-side critical section. */
> if (special.b.blocked) {
> t->rcu_read_unlock_special.b.blocked = false;
> @@ -602,6 +589,66 @@ static void rcu_read_unlock_special(struct task_struct *t)
> }
> }
>
> +/*
> + * Is a deferred quiescent-state pending, and are we also not in
> + * an RCU read-side critical section? It is the caller's responsibility
> + * to ensure it is otherwise safe to report any deferred quiescent
> + * states. The reason for this is that it is safe to report a
> + * quiescent state during context switch even though preemption
> + * is disabled. This function cannot be expected to understand these
> + * nuances, so the caller must handle them.
> + */
> +static bool rcu_preempt_need_deferred_qs(struct task_struct *t)
> +{
> + return (this_cpu_ptr(&rcu_preempt_data)->deferred_qs ||
> + READ_ONCE(t->rcu_read_unlock_special.s)) &&
> + !t->rcu_read_lock_nesting;
> +}
> +
> +/*
> + * Report a deferred quiescent state if needed and safe to do so.
> + * As with rcu_preempt_need_deferred_qs(), "safe" involves only
> + * not being in an RCU read-side critical section. The caller must
> + * evaluate safety in terms of interrupt, softirq, and preemption
> + * disabling.
> + */
> +static void rcu_preempt_deferred_qs(struct task_struct *t)
> +{
> + unsigned long flags;
> +
> + if (!rcu_preempt_need_deferred_qs(t))
> + return;
> + local_irq_save(flags);
> + rcu_preempt_deferred_qs_irqrestore(t, flags);
> +}
> +
> +/*
> + * Handle special cases during rcu_read_unlock(), such as needing to
> + * notify RCU core processing or task having blocked during the RCU
> + * read-side critical section.
> + */
> +static void rcu_read_unlock_special(struct task_struct *t)
> +{
> + unsigned long flags;
> + bool preempt_bh_were_disabled = !!(preempt_count() & ~HARDIRQ_MASK);
> + bool irqs_were_disabled;
> +
> + /* NMI handlers cannot block and cannot safely manipulate state. */
> + if (in_nmi())
> + return;
> +
> + local_irq_save(flags);
> + irqs_were_disabled = irqs_disabled_flags(flags);
> + if ((preempt_bh_were_disabled || irqs_were_disabled) &&
> + t->rcu_read_unlock_special.b.blocked) {
> + /* Need to defer quiescent state until everything is enabled. */
> + raise_softirq_irqoff(RCU_SOFTIRQ);
> + local_irq_restore(flags);
> + return;
> + }
> + rcu_preempt_deferred_qs_irqrestore(t, flags);
> +}
> +
> /*
> * Dump detailed information for all tasks blocking the current RCU
> * grace period on the specified rcu_node structure.
> @@ -737,10 +784,20 @@ static void rcu_preempt_check_callbacks(void)
> struct rcu_state *rsp = &rcu_preempt_state;
> struct task_struct *t = current;
>
> - if (t->rcu_read_lock_nesting == 0) {
> - rcu_preempt_qs();
> + if (t->rcu_read_lock_nesting > 0 ||
> + (preempt_count() & (PREEMPT_MASK | SOFTIRQ_MASK))) {
> + /* No QS, force context switch if deferred. */
> + if (rcu_preempt_need_deferred_qs(t))
> + resched_cpu(smp_processor_id());


Hi Paul,

I had a similar idea of checking the preempt_count() sometime back but didn't
believe this path can be called with preempt enabled (for some reason ;-)).
Now that I've convinced myself that's possible, what do you think about
taking advantage of the opportunity to report a RCU-sched qs like below from
rcu_check_callbacks ?

Did some basic testing, can roll into a patch later if you're Ok with it.

thanks.

---8<-----------------------

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index fb440baf8ac6..caa1e68f4168 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2683,6 +2683,12 @@ void rcu_check_callbacks(int user)
rcu_note_voluntary_context_switch(current);

} else if (!in_softirq()) {
+ /*
+ * Report RCU-sched qs if not in an RCU-sched read-side
+ * critical section.
+ */
+ if (!(preempt_count() & PREEMPT_MASK))
+ rcu_sched_qs();

/*
* Get here if this CPU did not take its interrupt from
\
 
 \ /
  Last update: 2018-07-01 20:38    [W:0.103 / U:0.856 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site