rcu: Yield simpler

The rcu_yield() code is amazing. It's there to avoid starvation of the
system when lots of (boosting) work is to be done.

Now looking at the code it's functionality is:

 Make the thread SCHED_OTHER and very nice, i.e. get it out of the way
 Arm a timer with 2 ticks
 schedule()

Now if the system goes idle the rcu task returns, regains SCHED_FIFO
and plugs on. If the systems stays busy the timer fires and wakes a
per node kthread which in turn makes the per cpu thread SCHED_FIFO and
brings it back on the cpu. For the boosting thread the "make it FIFO"
bit is missing and it just runs some magic boost checks. Now this is a
lot of code with extra threads and complexity.

It's way simpler to let the tasks when they detect overload schedule
away for 2 ticks and defer the normal wakeup as long as they are in
yielded state and the cpu is not idle.

That solves the same problem and the only difference is that when the
cpu goes idle it's not guaranteed that the thread returns right away,
but it won't be longer out than two ticks, so no harm is done. If
that's an issue than it is way simpler just to wake the task from
idle as RCU has callbacks there anyway.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/r/20120716103948.131256723@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

3 files changed, 41 insertions, 184 deletions

diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index f280e542e3e9..f08ee3bc5741 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -139,7 +139,7 @@ DEFINE_PER_CPU(char, rcu_cpu_has_work);
 
 #endif /* #ifdef CONFIG_RCU_BOOST */
 
-static void rcu_node_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu);
+static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu);
 static void invoke_rcu_core(void);
 static void invoke_rcu_callbacks(struct rcu_state *rsp, struct rcu_data *rdp);
 
@@ -1469,7 +1469,7 @@ static void rcu_cleanup_dead_cpu(int cpu, struct rcu_state *rsp)
 
         /* Adjust any no-longer-needed kthreads. */
         rcu_stop_cpu_kthread(cpu);
-        rcu_node_kthread_setaffinity(rnp, -1);
+        rcu_boost_kthread_setaffinity(rnp, -1);
 
         /* Remove the dead CPU from the bitmasks in the rcu_node hierarchy. */
 
@@ -2594,11 +2594,11 @@ static int __cpuinit rcu_cpu_notify(struct notifier_block *self,
                 break;
         case CPU_ONLINE:
         case CPU_DOWN_FAILED:
-                rcu_node_kthread_setaffinity(rnp, -1);
+                rcu_boost_kthread_setaffinity(rnp, -1);
                 rcu_cpu_kthread_setrt(cpu, 1);
                 break;
         case CPU_DOWN_PREPARE:
-                rcu_node_kthread_setaffinity(rnp, cpu);
+                rcu_boost_kthread_setaffinity(rnp, cpu);
                 rcu_cpu_kthread_setrt(cpu, 0);
                 break;
         case CPU_DYING:
diff --git a/kernel/rcutree.h b/kernel/rcutree.h
index 4d29169f2124..f08176172546 100644
--- a/kernel/rcutree.h
+++ b/kernel/rcutree.h
@@ -491,13 +491,8 @@ static void invoke_rcu_callbacks_kthread(void);
 static bool rcu_is_callbacks_kthread(void);
 #ifdef CONFIG_RCU_BOOST
 static void rcu_preempt_do_callbacks(void);
-static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp,
-                                          cpumask_var_t cm);
 static int __cpuinit rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
-                                                 struct rcu_node *rnp,
-                                                 int rnp_index);
-static void invoke_rcu_node_kthread(struct rcu_node *rnp);
-static void rcu_yield(void (*f)(unsigned long), unsigned long arg);
+                                                 struct rcu_node *rnp);
 #endif /* #ifdef CONFIG_RCU_BOOST */
 static void rcu_cpu_kthread_setrt(int cpu, int to_rt);
 static void __cpuinit rcu_prepare_kthreads(int cpu);
diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
index 7f3244c0df01..0f8b5ec64a7d 100644
--- a/kernel/rcutree_plugin.h
+++ b/kernel/rcutree_plugin.h
@@ -1069,6 +1069,16 @@ static void rcu_initiate_boost_trace(struct rcu_node *rnp)
 
 #endif /* #else #ifdef CONFIG_RCU_TRACE */
 
+static void rcu_wake_cond(struct task_struct *t, int status)
+{
+        /*
+         * If the thread is yielding, only wake it when this
+         * is invoked from idle
+         */
+        if (status != RCU_KTHREAD_YIELDING || is_idle_task(current))
+                wake_up_process(t);
+}
+
 /*
  * Carry out RCU priority boosting on the task indicated by ->exp_tasks
  * or ->boost_tasks, advancing the pointer to the next task in the
@@ -1141,17 +1151,6 @@ static int rcu_boost(struct rcu_node *rnp)
 }
 
 /*
- * Timer handler to initiate waking up of boost kthreads that
- * have yielded the CPU due to excessive numbers of tasks to
- * boost.  We wake up the per-rcu_node kthread, which in turn
- * will wake up the booster kthread.
- */
-static void rcu_boost_kthread_timer(unsigned long arg)
-{
-        invoke_rcu_node_kthread((struct rcu_node *)arg);
-}
-
-/*
  * Priority-boosting kthread.  One per leaf rcu_node and one for the
  * root rcu_node.
  */
@@ -1174,8 +1173,9 @@ static int rcu_boost_kthread(void *arg)
                 else
                         spincnt = 0;
                 if (spincnt > 10) {
+                        rnp->boost_kthread_status = RCU_KTHREAD_YIELDING;
                         trace_rcu_utilization("End boost kthread@rcu_yield");
-                        rcu_yield(rcu_boost_kthread_timer, (unsigned long)rnp);
+                        schedule_timeout_interruptible(2);
                         trace_rcu_utilization("Start boost kthread@rcu_yield");
                         spincnt = 0;
                 }
@@ -1213,8 +1213,8 @@ static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
                         rnp->boost_tasks = rnp->gp_tasks;
                 raw_spin_unlock_irqrestore(&rnp->lock, flags);
                 t = rnp->boost_kthread_task;
-                if (t != NULL)
-                        wake_up_process(t);
+                if (t)
+                        rcu_wake_cond(t, rnp->boost_kthread_status);
         } else {
                 rcu_initiate_boost_trace(rnp);
                 raw_spin_unlock_irqrestore(&rnp->lock, flags);
@@ -1231,8 +1231,10 @@ static void invoke_rcu_callbacks_kthread(void)
         local_irq_save(flags);
         __this_cpu_write(rcu_cpu_has_work, 1);
         if (__this_cpu_read(rcu_cpu_kthread_task) != NULL &&
-            current != __this_cpu_read(rcu_cpu_kthread_task))
-                wake_up_process(__this_cpu_read(rcu_cpu_kthread_task));
+            current != __this_cpu_read(rcu_cpu_kthread_task)) {
+                rcu_wake_cond(__this_cpu_read(rcu_cpu_kthread_task),
+                              __this_cpu_read(rcu_cpu_kthread_status));
+        }
         local_irq_restore(flags);
 }
 
@@ -1245,21 +1247,6 @@ static bool rcu_is_callbacks_kthread(void)
         return __get_cpu_var(rcu_cpu_kthread_task) == current;
 }
 
-/*
- * Set the affinity of the boost kthread.  The CPU-hotplug locks are
- * held, so no one should be messing with the existence of the boost
- * kthread.
- */
-static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp,
-                                          cpumask_var_t cm)
-{
-        struct task_struct *t;
-
-        t = rnp->boost_kthread_task;
-        if (t != NULL)
-                set_cpus_allowed_ptr(rnp->boost_kthread_task, cm);
-}
-
 #define RCU_BOOST_DELAY_JIFFIES DIV_ROUND_UP(CONFIG_RCU_BOOST_DELAY * HZ, 1000)
 
 /*
@@ -1276,15 +1263,19 @@ static void rcu_preempt_boost_start_gp(struct rcu_node *rnp)
  * Returns zero if all is well, a negated errno otherwise.
  */
 static int __cpuinit rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
-                                                 struct rcu_node *rnp,
-                                                 int rnp_index)
+                                                 struct rcu_node *rnp)
 {
+        int rnp_index = rnp - &rsp->node[0];
         unsigned long flags;
         struct sched_param sp;
         struct task_struct *t;
 
         if (&rcu_preempt_state != rsp)
                 return 0;
+
+        if (!rcu_scheduler_fully_active || rnp->qsmaskinit == 0)
+                return 0;
+
         rsp->boost = 1;
         if (rnp->boost_kthread_task != NULL)
                 return 0;
@@ -1328,20 +1319,6 @@ static void rcu_kthread_do_work(void)
 }
 
 /*
- * Wake up the specified per-rcu_node-structure kthread.
- * Because the per-rcu_node kthreads are immortal, we don't need
- * to do anything to keep them alive.
- */
-static void invoke_rcu_node_kthread(struct rcu_node *rnp)
-{
-        struct task_struct *t;
-
-        t = rnp->node_kthread_task;
-        if (t != NULL)
-                wake_up_process(t);
-}
-
-/*
  * Set the specified CPU's kthread to run RT or not, as specified by
  * the to_rt argument.  The CPU-hotplug locks are held, so the task
  * is not going away.
@@ -1366,45 +1343,6 @@ static void rcu_cpu_kthread_setrt(int cpu, int to_rt)
 }
 
 /*
- * Timer handler to initiate the waking up of per-CPU kthreads that
- * have yielded the CPU due to excess numbers of RCU callbacks.
- * We wake up the per-rcu_node kthread, which in turn will wake up
- * the booster kthread.
- */
-static void rcu_cpu_kthread_timer(unsigned long arg)
-{
-        struct rcu_data *rdp = per_cpu_ptr(rcu_state->rda, arg);
-        struct rcu_node *rnp = rdp->mynode;
-
-        atomic_or(rdp->grpmask, &rnp->wakemask);
-        invoke_rcu_node_kthread(rnp);
-}
-
-/*
- * Drop to non-real-time priority and yield, but only after posting a
- * timer that will cause us to regain our real-time priority if we
- * remain preempted.  Either way, we restore our real-time priority
- * before returning.
- */
-static void rcu_yield(void (*f)(unsigned long), unsigned long arg)
-{
-        struct sched_param sp;
-        struct timer_list yield_timer;
-        int prio = current->rt_priority;
-
-        setup_timer_on_stack(&yield_timer, f, arg);
-        mod_timer(&yield_timer, jiffies + 2);
-        sp.sched_priority = 0;
-        sched_setscheduler_nocheck(current, SCHED_NORMAL, &sp);
-        set_user_nice(current, 19);
-        schedule();
-        set_user_nice(current, 0);
-        sp.sched_priority = prio;
-        sched_setscheduler_nocheck(current, SCHED_FIFO, &sp);
-        del_timer(&yield_timer);
-}
-
-/*
  * Handle cases where the rcu_cpu_kthread() ends up on the wrong CPU.
  * This can happen while the corresponding CPU is either coming online
  * or going offline.  We cannot wait until the CPU is fully online
@@ -1476,7 +1414,7 @@ static int rcu_cpu_kthread(void *arg)
                 if (spincnt > 10) {
                         *statusp = RCU_KTHREAD_YIELDING;
                         trace_rcu_utilization("End CPU kthread@rcu_yield");
-                        rcu_yield(rcu_cpu_kthread_timer, (unsigned long)cpu);
+                        schedule_timeout_interruptible(2);
                         trace_rcu_utilization("Start CPU kthread@rcu_yield");
                         spincnt = 0;
                 }
@@ -1533,48 +1471,6 @@ static int __cpuinit rcu_spawn_one_cpu_kthread(int cpu)
 }
 
 /*
- * Per-rcu_node kthread, which is in charge of waking up the per-CPU
- * kthreads when needed.  We ignore requests to wake up kthreads
- * for offline CPUs, which is OK because force_quiescent_state()
- * takes care of this case.
- */
-static int rcu_node_kthread(void *arg)
-{
-        int cpu;
-        unsigned long flags;
-        unsigned long mask;
-        struct rcu_node *rnp = (struct rcu_node *)arg;
-        struct sched_param sp;
-        struct task_struct *t;
-
-        for (;;) {
-                rnp->node_kthread_status = RCU_KTHREAD_WAITING;
-                rcu_wait(atomic_read(&rnp->wakemask) != 0);
-                rnp->node_kthread_status = RCU_KTHREAD_RUNNING;
-                raw_spin_lock_irqsave(&rnp->lock, flags);
-                mask = atomic_xchg(&rnp->wakemask, 0);
-                rcu_initiate_boost(rnp, flags); /* releases rnp->lock. */
-                for (cpu = rnp->grplo; cpu <= rnp->grphi; cpu++, mask >>= 1) {
-                        if ((mask & 0x1) == 0)
-                                continue;
-                        preempt_disable();
-                        t = per_cpu(rcu_cpu_kthread_task, cpu);
-                        if (!cpu_online(cpu) || t == NULL) {
-                                preempt_enable();
-                                continue;
-                        }
-                        per_cpu(rcu_cpu_has_work, cpu) = 1;
-                        sp.sched_priority = RCU_KTHREAD_PRIO;
-                        sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
-                        preempt_enable();
-                }
-        }
-        /* NOTREACHED */
-        rnp->node_kthread_status = RCU_KTHREAD_STOPPED;
-        return 0;
-}
-
-/*
  * Set the per-rcu_node kthread's affinity to cover all CPUs that are
  * served by the rcu_node in question.  The CPU hotplug lock is still
  * held, so the value of rnp->qsmaskinit will be stable.
@@ -1583,17 +1479,17 @@ static int rcu_node_kthread(void *arg)
  * no outgoing CPU.  If there are no CPUs left in the affinity set,
  * this function allows the kthread to execute on any CPU.
  */
-static void rcu_node_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
+static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
 {
+        struct task_struct *t = rnp->boost_kthread_task;
+        unsigned long mask = rnp->qsmaskinit;
         cpumask_var_t cm;
         int cpu;
-        unsigned long mask = rnp->qsmaskinit;
 
-        if (rnp->node_kthread_task == NULL)
+        if (!t)
                 return;
-        if (!alloc_cpumask_var(&cm, GFP_KERNEL))
+        if (!zalloc_cpumask_var(&cm, GFP_KERNEL))
                 return;
-        cpumask_clear(cm);
         for (cpu = rnp->grplo; cpu <= rnp->grphi; cpu++, mask >>= 1)
                 if ((mask & 0x1) && cpu != outgoingcpu)
                         cpumask_set_cpu(cpu, cm);
@@ -1603,50 +1499,17 @@ static void rcu_node_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
                         cpumask_clear_cpu(cpu, cm);
                 WARN_ON_ONCE(cpumask_weight(cm) == 0);
         }
-        set_cpus_allowed_ptr(rnp->node_kthread_task, cm);
-        rcu_boost_kthread_setaffinity(rnp, cm);
+        set_cpus_allowed_ptr(t, cm);
         free_cpumask_var(cm);
 }
 
 /*
- * Spawn a per-rcu_node kthread, setting priority and affinity.
- * Called during boot before online/offline can happen, or, if
- * during runtime, with the main CPU-hotplug locks held.  So only
- * one of these can be executing at a time.
- */
-static int __cpuinit rcu_spawn_one_node_kthread(struct rcu_state *rsp,
-                                                struct rcu_node *rnp)
-{
-        unsigned long flags;
-        int rnp_index = rnp - &rsp->node[0];
-        struct sched_param sp;
-        struct task_struct *t;
-
-        if (!rcu_scheduler_fully_active ||
-            rnp->qsmaskinit == 0)
-                return 0;
-        if (rnp->node_kthread_task == NULL) {
-                t = kthread_create(rcu_node_kthread, (void *)rnp,
-                                   "rcun/%d", rnp_index);
-                if (IS_ERR(t))
-                        return PTR_ERR(t);
-                raw_spin_lock_irqsave(&rnp->lock, flags);
-                rnp->node_kthread_task = t;
-                raw_spin_unlock_irqrestore(&rnp->lock, flags);
-                sp.sched_priority = 99;
-                sched_setscheduler_nocheck(t, SCHED_FIFO, &sp);
-                wake_up_process(t); /* get to TASK_INTERRUPTIBLE quickly. */
-        }
-        return rcu_spawn_one_boost_kthread(rsp, rnp, rnp_index);
-}
-
-/*
  * Spawn all kthreads -- called as soon as the scheduler is running.
  */
 static int __init rcu_spawn_kthreads(void)
 {
-        int cpu;
         struct rcu_node *rnp;
+        int cpu;
 
         rcu_scheduler_fully_active = 1;
         for_each_possible_cpu(cpu) {
@@ -1655,10 +1518,10 @@ static int __init rcu_spawn_kthreads(void)
                         (void)rcu_spawn_one_cpu_kthread(cpu);
         }
         rnp = rcu_get_root(rcu_state);
-        (void)rcu_spawn_one_node_kthread(rcu_state, rnp);
+        (void)rcu_spawn_one_boost_kthread(rcu_state, rnp);
         if (NUM_RCU_NODES > 1) {
                 rcu_for_each_leaf_node(rcu_state, rnp)
-                        (void)rcu_spawn_one_node_kthread(rcu_state, rnp);
+                        (void)rcu_spawn_one_boost_kthread(rcu_state, rnp);
         }
         return 0;
 }
@@ -1672,8 +1535,7 @@ static void __cpuinit rcu_prepare_kthreads(int cpu)
         /* Fire up the incoming CPU's kthread and leaf rcu_node kthread. */
         if (rcu_scheduler_fully_active) {
                 (void)rcu_spawn_one_cpu_kthread(cpu);
-                if (rnp->node_kthread_task == NULL)
-                        (void)rcu_spawn_one_node_kthread(rcu_state, rnp);
+                (void)rcu_spawn_one_boost_kthread(rcu_state, rnp);
         }
 }
 
@@ -1706,7 +1568,7 @@ static void rcu_stop_cpu_kthread(int cpu)
 
 #endif /* #ifdef CONFIG_HOTPLUG_CPU */
 
-static void rcu_node_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
+static void rcu_boost_kthread_setaffinity(struct rcu_node *rnp, int outgoingcpu)
 {
 }