Index: linux-3.16-ck2/arch/x86/Kconfig
===================================================================
--- linux-3.16-ck2.orig/arch/x86/Kconfig	2014-08-25 09:45:57.910543881 +1000
+++ linux-3.16-ck2/arch/x86/Kconfig	2014-08-25 09:48:39.000000000 +1000
@@ -794,7 +794,7 @@ config SCHED_SMT
 
 config SMT_NICE
 	bool "SMT (Hyperthreading) aware nice priority and policy support"
-	depends on X86_HT && SCHED_BFS
+	depends on X86_HT && SCHED_BFS && SCHED_SMT
 	default y
 	---help---
 	  Enabling Hyperthreading on Intel CPUs decreases the effectiveness
Index: linux-3.16-ck2/kernel/sched/bfs.c
===================================================================
--- linux-3.16-ck2.orig/kernel/sched/bfs.c	2014-08-25 09:45:58.002543872 +1000
+++ linux-3.16-ck2/kernel/sched/bfs.c	2014-08-25 09:45:58.285543846 +1000
@@ -150,7 +150,7 @@ EXPORT_SYMBOL(__smp_mb__after_atomic);
 
 void print_scheduler_version(void)
 {
-	printk(KERN_INFO "BFS CPU scheduler v0.450 by Con Kolivas.\n");
+	printk(KERN_INFO "BFS CPU scheduler v0.456 by Con Kolivas.\n");
 }
 
 /*
@@ -1179,6 +1179,13 @@ static inline void return_task(struct ta
 	}
 }
 
+/* Enter with grq lock held. We know p is on the local cpu */
+static inline void __set_tsk_resched(struct task_struct *p)
+{
+	set_tsk_need_resched(p);
+	set_preempt_need_resched();
+}
+
 /*
  * resched_task - mark a task 'to be rescheduled now'.
  *
@@ -1505,6 +1512,18 @@ ttwu_stat(struct task_struct *p, int cpu
 #endif /* CONFIG_SCHEDSTATS */
 }
 
+#ifdef CONFIG_SMP
+void scheduler_ipi(void)
+{
+	/*
+	 * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting
+	 * TIF_NEED_RESCHED remotely (for the first time) will also send
+	 * this IPI.
+	 */
+	preempt_fold_need_resched();
+}
+#endif
+
 static inline void ttwu_activate(struct task_struct *p, struct rq *rq,
 				 bool is_sync)
 {
@@ -1776,7 +1795,7 @@ after_ts_init:
 			 * do child-runs-first in anticipation of an exec. This
 			 * usually avoids a lot of COW overhead.
 			 */
-			set_tsk_need_resched(parent);
+			__set_tsk_resched(parent);
 		} else
 			try_preempt(p, rq);
 	} else {
@@ -1788,7 +1807,7 @@ after_ts_init:
 		 	* be slightly earlier.
 		 	*/
 			rq->rq_time_slice = 0;
-			set_tsk_need_resched(parent);
+			__set_tsk_resched(parent);
 		}
 		time_slice_expired(p);
 	}
@@ -2913,9 +2932,10 @@ static void task_running_tick(struct rq
 
 	/* p->time_slice < RESCHED_US. We only modify task_struct under grq lock */
 	p = rq->curr;
+
 	grq_lock();
 	requeue_task(p);
-	set_tsk_need_resched(p);
+	__set_tsk_resched(p);
 	grq_unlock();
 }
 
@@ -4745,9 +4765,9 @@ EXPORT_SYMBOL(yield);
  */
 int __sched yield_to(struct task_struct *p, bool preempt)
 {
+	struct rq *rq, *p_rq;
 	unsigned long flags;
 	int yielded = 0;
-	struct rq *rq;
 
 	rq = this_rq();
 	grq_lock_irqsave(&flags);
@@ -4755,6 +4775,8 @@ int __sched yield_to(struct task_struct
 		yielded = -ESRCH;
 		goto out_unlock;
 	}
+
+	p_rq = task_rq(p);
 	yielded = 1;
 	if (p->deadline > rq->rq_deadline)
 		p->deadline = rq->rq_deadline;
@@ -4762,7 +4784,8 @@ int __sched yield_to(struct task_struct
 	rq->rq_time_slice = 0;
 	if (p->time_slice > timeslice())
 		p->time_slice = timeslice();
-	set_tsk_need_resched(rq->curr);
+	if (preempt && rq != rq)
+		resched_task(p_rq->curr);
 out_unlock:
 	grq_unlock_irqrestore(&flags);
 
@@ -5028,7 +5051,6 @@ void init_idle(struct task_struct *idle,
 #if defined(CONFIG_SMP)
 	sprintf(idle->comm, "%s/%d", INIT_TASK_COMM, cpu);
 #endif
-	set_tsk_need_resched(idle);
 }
 
 void resched_cpu(int cpu)
@@ -5195,7 +5217,7 @@ out:
 	task_grq_unlock(&flags);
 
 	if (running_wrong)
-		_cond_resched();
+		__cond_resched();
 
 	return ret;
 }
Index: linux-3.16-ck2/include/linux/sched.h
===================================================================
--- linux-3.16-ck2.orig/include/linux/sched.h	2014-08-25 09:45:58.094543864 +1000
+++ linux-3.16-ck2/include/linux/sched.h	2014-08-25 09:45:58.092543864 +1000
@@ -2510,12 +2510,7 @@ static inline void set_task_comm(struct
 extern char *get_task_comm(char *to, struct task_struct *tsk);
 
 #ifdef CONFIG_SMP
-#ifdef CONFIG_SCHED_BFS
-/* scheduler_ipi does nothing on BFS */
-static inline void scheduler_ipi(void) { }
-#else
 void scheduler_ipi(void);
-#endif
 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
 #else
 static inline void scheduler_ipi(void) { }