rcu: refactor RCU's context-switch handling

The addition of preemptible RCU to treercu resulted in a bit of
confusion and inefficiency surrounding the handling of context switches
for RCU-sched and for RCU-preempt.  For RCU-sched, a context switch
is a quiescent state, pure and simple, just like it always has been.
For RCU-preempt, a context switch is in no way a quiescent state, but
special handling is required when a task blocks in an RCU read-side
critical section.

However, the callout from the scheduler and the outer loop in ksoftirqd
still calls something named rcu_sched_qs(), whose name is no longer
accurate.  Furthermore, when rcu_check_callbacks() notes an RCU-sched
quiescent state, it ends up unnecessarily (though harmlessly, aside
from the performance hit) enqueuing the current task if it happens to
be running in an RCU-preempt read-side critical section.  This not only
increases the maximum latency of scheduler_tick(), it also needlessly
increases the overhead of the next outermost rcu_read_unlock() invocation.

This patch addresses this situation by separating the notion of RCU's
context-switch handling from that of RCU-sched's quiescent states.
The context-switch handling is covered by rcu_note_context_switch() in
general and by rcu_preempt_note_context_switch() for preemptible RCU.
This permits rcu_sched_qs() to handle quiescent states and only quiescent
states.  It also reduces the maximum latency of scheduler_tick(), though
probably by much less than a microsecond.  Finally, it means that tasks
within preemptible-RCU read-side critical sections avoid incurring the
overhead of queuing unless there really is a context switch.

Suggested-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Acked-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>

4 files changed, 21 insertions, 11 deletions

diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index 86bb9499aae6..e33631354b69 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -97,25 +97,32 @@ static int rcu_gp_in_progress(struct rcu_state *rsp)
  */
 void rcu_sched_qs(int cpu)
 {
-        struct rcu_data *rdp;
+        struct rcu_data *rdp = &per_cpu(rcu_sched_data, cpu);
 
-        rdp = &per_cpu(rcu_sched_data, cpu);
         rdp->passed_quiesc_completed = rdp->gpnum - 1;
         barrier();
         rdp->passed_quiesc = 1;
-        rcu_preempt_note_context_switch(cpu);
 }
 
 void rcu_bh_qs(int cpu)
 {
-        struct rcu_data *rdp;
+        struct rcu_data *rdp = &per_cpu(rcu_bh_data, cpu);
 
-        rdp = &per_cpu(rcu_bh_data, cpu);
         rdp->passed_quiesc_completed = rdp->gpnum - 1;
         barrier();
         rdp->passed_quiesc = 1;
 }
 
+/*
+ * Note a context switch.  This is a quiescent state for RCU-sched,
+ * and requires special handling for preemptible RCU.
+ */
+void rcu_note_context_switch(int cpu)
+{
+        rcu_sched_qs(cpu);
+        rcu_preempt_note_context_switch(cpu);
+}
+
 #ifdef CONFIG_NO_HZ
 DEFINE_PER_CPU(struct rcu_dynticks, rcu_dynticks) = {
         .dynticks_nesting = 1,
diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
index 687c4e90722e..f9bc83a047da 100644
--- a/kernel/rcutree_plugin.h
+++ b/kernel/rcutree_plugin.h
@@ -75,13 +75,19 @@ EXPORT_SYMBOL_GPL(rcu_force_quiescent_state);
  * that this just means that the task currently running on the CPU is
  * not in a quiescent state.  There might be any number of tasks blocked
  * while in an RCU read-side critical section.
+ *
+ * Unlike the other rcu_*_qs() functions, callers to this function
+ * must disable irqs in order to protect the assignment to
+ * ->rcu_read_unlock_special.
  */
 static void rcu_preempt_qs(int cpu)
 {
         struct rcu_data *rdp = &per_cpu(rcu_preempt_data, cpu);
+
         rdp->passed_quiesc_completed = rdp->gpnum - 1;
         barrier();
         rdp->passed_quiesc = 1;
+        current->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS;
 }
 
 /*
@@ -144,9 +150,8 @@ static void rcu_preempt_note_context_switch(int cpu)
          * grace period, then the fact that the task has been enqueued
          * means that we continue to block the current grace period.
          */
-        rcu_preempt_qs(cpu);
         local_irq_save(flags);
-        t->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS;
+        rcu_preempt_qs(cpu);
         local_irq_restore(flags);
 }
 
@@ -236,7 +241,6 @@ static void rcu_read_unlock_special(struct task_struct *t)
          */
         special = t->rcu_read_unlock_special;
         if (special & RCU_READ_UNLOCK_NEED_QS) {
-                t->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS;
                 rcu_preempt_qs(smp_processor_id());
         }
 
@@ -473,7 +477,6 @@ static void rcu_preempt_check_callbacks(int cpu)
         struct task_struct *t = current;
 
         if (t->rcu_read_lock_nesting == 0) {
-                t->rcu_read_unlock_special &= ~RCU_READ_UNLOCK_NEED_QS;
                 rcu_preempt_qs(cpu);
                 return;
         }
diff --git a/kernel/sched.c b/kernel/sched.c
index 3c2a54f70ffe..d8a213ccdc3b 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -3706,7 +3706,7 @@ need_resched:
         preempt_disable();
         cpu = smp_processor_id();
         rq = cpu_rq(cpu);
-        rcu_sched_qs(cpu);
+        rcu_note_context_switch(cpu);
         prev = rq->curr;
         switch_count = &prev->nivcsw;
 
diff --git a/kernel/softirq.c b/kernel/softirq.c
index 7c1a67ef0274..0db913a5c60f 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -716,7 +716,7 @@ static int run_ksoftirqd(void * __bind_cpu)
                         preempt_enable_no_resched();
                         cond_resched();
                         preempt_disable();
-                        rcu_sched_qs((long)__bind_cpu);
+                        rcu_note_context_switch((long)__bind_cpu);
                 }
                 preempt_enable();
                 set_current_state(TASK_INTERRUPTIBLE);