sched: wakeup preempt when small overlap

Lin Ming reported a 10% OLTP regression against 2.6.27-rc4.

The difference seems to come from different preemption agressiveness,
which affects the cache footprint of the workload and its effective
cache trashing.

Aggresively preempt a task if its avg overlap is very small, this should
avoid the task going to sleep and find it still running when we schedule
back to it - saving a wakeup.

Reported-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

diff --git a/include/linux/sched.h b/include/linux/sched.h
index b3b7a8f32477aa096c170e1343df846d76df2846..d8e699b558586ac0ee0f3f46e6304e6ac2d5057f 100644 (file)
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -897,7 +897,7 @@ struct sched_class {
        void (*yield_task) (struct rq *rq);
        int  (*select_task_rq)(struct task_struct *p, int sync);
 
-       void (*check_preempt_curr) (struct rq *rq, struct task_struct *p);
+       void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
 
        struct task_struct * (*pick_next_task) (struct rq *rq);
        void (*put_prev_task) (struct rq *rq, struct task_struct *p);

diff --git a/kernel/sched.c b/kernel/sched.c
index 0d8905a1b8ca9fab76ae92e2b1b952b990d83b02..ad9d39b021f8454bf6b66158796f705d846c0645 100644 (file)
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -604,9 +604,9 @@ struct rq {
 
 static DEFINE_PER_CPU_SHARED_ALIGNED(struct rq, runqueues);
 
-static inline void check_preempt_curr(struct rq *rq, struct task_struct *p)
+static inline void check_preempt_curr(struct rq *rq, struct task_struct *p, int sync)
 {
-       rq->curr->sched_class->check_preempt_curr(rq, p);
+       rq->curr->sched_class->check_preempt_curr(rq, p, sync);
 }
 
 static inline int cpu_of(struct rq *rq)
@@ -2282,7 +2282,7 @@ out_running:
        trace_mark(kernel_sched_wakeup,
                "pid %d state %ld ## rq %p task %p rq->curr %p",
                p->pid, p->state, rq, p, rq->curr);
-       check_preempt_curr(rq, p);
+       check_preempt_curr(rq, p, sync);
 
        p->state = TASK_RUNNING;
 #ifdef CONFIG_SMP
@@ -2417,7 +2417,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags)
        trace_mark(kernel_sched_wakeup_new,
                "pid %d state %ld ## rq %p task %p rq->curr %p",
                p->pid, p->state, rq, p, rq->curr);
-       check_preempt_curr(rq, p);
+       check_preempt_curr(rq, p, 0);
 #ifdef CONFIG_SMP
        if (p->sched_class->task_wake_up)
                p->sched_class->task_wake_up(rq, p);
@@ -2877,7 +2877,7 @@ static void pull_task(struct rq *src_rq, struct task_struct *p,
         * Note that idle threads have a prio of MAX_PRIO, for this test
         * to be always true for them.
         */
-       check_preempt_curr(this_rq, p);
+       check_preempt_curr(this_rq, p, 0);
 }
 
 /*
@@ -6007,7 +6007,7 @@ static int __migrate_task(struct task_struct *p, int src_cpu, int dest_cpu)
        set_task_cpu(p, dest_cpu);
        if (on_rq) {
                activate_task(rq_dest, p, 0);
-               check_preempt_curr(rq_dest, p);
+               check_preempt_curr(rq_dest, p, 0);
        }
 done:
        ret = 1;

diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
index a10ac0bcee6448ac99076360c67dab2c345966da..7328383690f1e66a603c707c231f1388f2f1008d 100644 (file)
--- a/kernel/sched_fair.c
+++ b/kernel/sched_fair.c
@@ -1331,7 +1331,7 @@ static inline int depth_se(struct sched_entity *se)
 /*
  * Preempt the current task with a newly woken task if needed:
  */
-static void check_preempt_wakeup(struct rq *rq, struct task_struct *p)
+static void check_preempt_wakeup(struct rq *rq, struct task_struct *p, int sync)
 {
        struct task_struct *curr = rq->curr;
        struct cfs_rq *cfs_rq = task_cfs_rq(curr);
@@ -1367,6 +1367,13 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p)
        if (!sched_feat(WAKEUP_PREEMPT))
                return;
 
+       if (sched_feat(WAKEUP_OVERLAP) && sync &&
+                       se->avg_overlap < sysctl_sched_migration_cost &&
+                       pse->avg_overlap < sysctl_sched_migration_cost) {
+               resched_task(curr);
+               return;
+       }
+
        /*
         * preemption test can be made between sibling entities who are in the
         * same cfs_rq i.e who have a common parent. Walk up the hierarchy of
@@ -1649,7 +1656,7 @@ static void prio_changed_fair(struct rq *rq, struct task_struct *p,
                if (p->prio > oldprio)
                        resched_task(rq->curr);
        } else
-               check_preempt_curr(rq, p);
+               check_preempt_curr(rq, p, 0);
 }
 
 /*
@@ -1666,7 +1673,7 @@ static void switched_to_fair(struct rq *rq, struct task_struct *p,
        if (running)
                resched_task(rq->curr);
        else
-               check_preempt_curr(rq, p);
+               check_preempt_curr(rq, p, 0);
 }
 
 /* Account for a task changing its policy or group.

diff --git a/kernel/sched_features.h b/kernel/sched_features.h
index 9353ca78154e880c786376108d2ef1be4335cde4..bf027a7accf801ed1850e36f08edaf9eb1d2dfff 100644 (file)
--- a/kernel/sched_features.h
+++ b/kernel/sched_features.h
@@ -11,3 +11,4 @@ SCHED_FEAT(ASYM_GRAN, 1)
 SCHED_FEAT(LB_BIAS, 1)
 SCHED_FEAT(LB_WAKEUP_UPDATE, 1)
 SCHED_FEAT(ASYM_EFF_LOAD, 1)
+SCHED_FEAT(WAKEUP_OVERLAP, 1)

diff --git a/kernel/sched_idletask.c b/kernel/sched_idletask.c
index 3a4f92dbbe6609b786da0add62f2306e991b791f..dec4ccabe2f5c8af51566dcc323ce52858ff8764 100644 (file)
--- a/kernel/sched_idletask.c
+++ b/kernel/sched_idletask.c
@@ -14,7 +14,7 @@ static int select_task_rq_idle(struct task_struct *p, int sync)
 /*
  * Idle tasks are unconditionally rescheduled:
  */
-static void check_preempt_curr_idle(struct rq *rq, struct task_struct *p)
+static void check_preempt_curr_idle(struct rq *rq, struct task_struct *p, int sync)
 {
        resched_task(rq->idle);
 }
@@ -76,7 +76,7 @@ static void switched_to_idle(struct rq *rq, struct task_struct *p,
        if (running)
                resched_task(rq->curr);
        else
-               check_preempt_curr(rq, p);
+               check_preempt_curr(rq, p, 0);
 }
 
 static void prio_changed_idle(struct rq *rq, struct task_struct *p,
@@ -93,7 +93,7 @@ static void prio_changed_idle(struct rq *rq, struct task_struct *p,
                if (p->prio > oldprio)
                        resched_task(rq->curr);
        } else
-               check_preempt_curr(rq, p);
+               check_preempt_curr(rq, p, 0);
 }
 
 /*

diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c
index 552310798dadf13e3b2059d5f84e050f7e0926e2..6d2d0a5d030b5c99ac28cbbb46fd3784e14d812d 100644 (file)
--- a/kernel/sched_rt.c
+++ b/kernel/sched_rt.c
@@ -783,7 +783,7 @@ static void check_preempt_equal_prio(struct rq *rq, struct task_struct *p)
 /*
  * Preempt the current task with a newly woken task if needed:
  */
-static void check_preempt_curr_rt(struct rq *rq, struct task_struct *p)
+static void check_preempt_curr_rt(struct rq *rq, struct task_struct *p, int sync)
 {
        if (p->prio < rq->curr->prio) {
                resched_task(rq->curr);