diff -urNp 2.5.58/fs/exec.c 2.5.58-ms-ilb-nb/fs/exec.c
--- 2.5.58/fs/exec.c	2003-01-14 06:58:33.000000000 +0100
+++ 2.5.58-ms-ilb-nb/fs/exec.c	2003-01-14 16:31:08.000000000 +0100
@@ -1031,6 +1031,8 @@ int do_execve(char * filename, char ** a
 	int retval;
 	int i;
 
+	sched_balance_exec();
+
 	file = open_exec(filename);
 
 	retval = PTR_ERR(file);
diff -urNp 2.5.58/include/linux/sched.h 2.5.58-ms-ilb-nb/include/linux/sched.h
--- 2.5.58/include/linux/sched.h	2003-01-14 06:58:06.000000000 +0100
+++ 2.5.58-ms-ilb-nb/include/linux/sched.h	2003-01-14 16:31:08.000000000 +0100
@@ -160,7 +160,6 @@ extern void update_one_process(struct ta
 extern void scheduler_tick(int user_tick, int system);
 extern unsigned long cache_decay_ticks;
 
-
 #define	MAX_SCHEDULE_TIMEOUT	LONG_MAX
 extern signed long FASTCALL(schedule_timeout(signed long timeout));
 asmlinkage void schedule(void);
@@ -444,6 +443,28 @@ extern void set_cpus_allowed(task_t *p, 
 # define set_cpus_allowed(p, new_mask) do { } while (0)
 #endif
 
+#ifdef CONFIG_NUMA
+extern void sched_balance_exec(void);
+extern void node_nr_running_init(void);
+
+static inline void nr_running_inc(runqueue_t *rq)
+{
+	atomic_inc(rq->node_ptr);
+	rq->nr_running++;
+}
+
+static inline void nr_running_dec(runqueue_t *rq)
+{
+	atomic_dec(rq->node_ptr);
+	rq->nr_running--;
+}
+#else
+#define sched_balance_exec() {}
+#define node_nr_running_init() {}
+#define nr_running_inc(rq) rq->nr_running++
+#define nr_running_dec(rq) rq->nr_running--
+#endif
+
 extern void set_user_nice(task_t *p, long nice);
 extern int task_prio(task_t *p);
 extern int task_nice(task_t *p);
diff -urNp 2.5.58/init/main.c 2.5.58-ms-ilb-nb/init/main.c
--- 2.5.58/init/main.c	2003-01-14 06:58:20.000000000 +0100
+++ 2.5.58-ms-ilb-nb/init/main.c	2003-01-14 16:31:08.000000000 +0100
@@ -495,6 +495,7 @@ static void do_pre_smp_initcalls(void)
 
 	migration_init();
 #endif
+	node_nr_running_init();
 	spawn_ksoftirqd();
 }
 
diff -urNp 2.5.58/kernel/sched.c 2.5.58-ms-ilb-nb/kernel/sched.c
--- 2.5.58/kernel/sched.c	2003-01-14 06:59:11.000000000 +0100
+++ 2.5.58-ms-ilb-nb/kernel/sched.c	2003-01-14 16:31:57.000000000 +0100
@@ -67,6 +67,8 @@
 #define INTERACTIVE_DELTA	2
 #define MAX_SLEEP_AVG		(2*HZ)
 #define STARVATION_LIMIT	(2*HZ)
+#define NODE_BALANCE_RATIO	10
+#define NODE_THRESHOLD          125
 
 /*
  * If a task is 'interactive' then we reinsert it in the active
@@ -153,7 +155,10 @@ struct runqueue {
 	task_t *curr, *idle;
 	prio_array_t *active, *expired, arrays[2];
 	int prev_nr_running[NR_CPUS];
-
+#ifdef CONFIG_NUMA
+	atomic_t * node_ptr;
+	unsigned int nr_balanced;
+#endif
 	task_t *migration_thread;
 	struct list_head migration_queue;
 
@@ -294,7 +299,7 @@ static inline void activate_task(task_t 
 		p->prio = effective_prio(p);
 	}
 	enqueue_task(p, array);
-	rq->nr_running++;
+	nr_running_inc(rq);
 }
 
 /*
@@ -302,7 +307,7 @@ static inline void activate_task(task_t 
  */
 static inline void deactivate_task(struct task_struct *p, runqueue_t *rq)
 {
-	rq->nr_running--;
+	nr_running_dec(rq);
 	if (p->state == TASK_UNINTERRUPTIBLE)
 		rq->nr_uninterruptible++;
 	dequeue_task(p, p->array);
@@ -624,6 +629,105 @@ static inline void double_rq_unlock(runq
 		spin_unlock(&rq2->lock);
 }
 
+#if CONFIG_NUMA
+static atomic_t node_nr_running[MAX_NUMNODES] ____cacheline_maxaligned_in_smp =
+	{[0 ...MAX_NUMNODES-1] = ATOMIC_INIT(0)};
+
+__init void node_nr_running_init(void)
+{
+	int i;
+
+	for (i = 0; i < NR_CPUS; i++)
+		cpu_rq(i)->node_ptr = &node_nr_running[__cpu_to_node(i)];
+}
+
+/*
+ * If dest_cpu is allowed for this process, migrate the task to it.
+ * This is accomplished by forcing the cpu_allowed mask to only
+ * allow dest_cpu, which will force the cpu onto dest_cpu.  Then
+ * the cpu_allowed mask is restored.
+ */
+static void sched_migrate_task(task_t *p, int dest_cpu)
+{
+	unsigned long old_mask;
+
+	old_mask = p->cpus_allowed;
+	if (!(old_mask & (1UL << dest_cpu)))
+		return;
+	/* force the process onto the specified CPU */
+	set_cpus_allowed(p, 1UL << dest_cpu);
+
+	/* restore the cpus allowed mask */
+	set_cpus_allowed(p, old_mask);
+}
+
+/*
+ * Find the least loaded CPU.  Slightly favor the current CPU by
+ * setting its runqueue length as the minimum to start.
+ */
+static int sched_best_cpu(struct task_struct *p)
+{
+	int i, minload, load, best_cpu, node = 0;
+	unsigned long cpumask;
+
+	best_cpu = task_cpu(p);
+	if (cpu_rq(best_cpu)->nr_running <= 2)
+		return best_cpu;
+
+	minload = 10000000;
+	for (i = 0; i < numnodes; i++) {
+		load = atomic_read(&node_nr_running[i]);
+		if (load < minload) {
+			minload = load;
+			node = i;
+		}
+	}
+
+	minload = 10000000;
+	cpumask = __node_to_cpu_mask(node);
+	for (i = 0; i < NR_CPUS; ++i) {
+		if (!(cpumask & (1UL << i)))
+			continue;
+		if (cpu_rq(i)->nr_running < minload) {
+			best_cpu = i;
+			minload = cpu_rq(i)->nr_running;
+		}
+	}
+	return best_cpu;
+}
+
+void sched_balance_exec(void)
+{
+	int new_cpu;
+
+	if (numnodes > 1) {
+		new_cpu = sched_best_cpu(current);
+		if (new_cpu != smp_processor_id())
+			sched_migrate_task(current, new_cpu);
+	}
+}
+
+static int find_busiest_node(int this_node)
+{
+	int i, node = -1, load, this_load, maxload;
+	
+	this_load = maxload = atomic_read(&node_nr_running[this_node]);
+	for (i = 0; i < numnodes; i++) {
+		if (i == this_node)
+			continue;
+		load = atomic_read(&node_nr_running[i]);
+		if (load > maxload &&
+		    (100*load > ((NODE_THRESHOLD*100*this_load)/100))) {
+			maxload = load;
+			node = i;
+		}
+	}
+	if (maxload <= 4)
+		node = -1;
+	return node;
+}
+#endif /* CONFIG_NUMA */
+
 #if CONFIG_SMP
 
 /*
@@ -652,9 +756,9 @@ static inline unsigned int double_lock_b
 }
 
 /*
- * find_busiest_queue - find the busiest runqueue.
+ * find_busiest_queue - find the busiest runqueue among the cpus in cpumask
  */
-static inline runqueue_t *find_busiest_queue(runqueue_t *this_rq, int this_cpu, int idle, int *imbalance)
+static inline runqueue_t *find_busiest_queue(runqueue_t *this_rq, int this_cpu, int idle, int *imbalance, unsigned long cpumask)
 {
 	int nr_running, load, max_load, i;
 	runqueue_t *busiest, *rq_src;
@@ -689,7 +793,7 @@ static inline runqueue_t *find_busiest_q
 	busiest = NULL;
 	max_load = 1;
 	for (i = 0; i < NR_CPUS; i++) {
-		if (!cpu_online(i))
+		if (!cpu_online(i) || !((1UL << i) & cpumask) )
 			continue;
 
 		rq_src = cpu_rq(i);
@@ -736,9 +840,9 @@ out:
 static inline void pull_task(runqueue_t *src_rq, prio_array_t *src_array, task_t *p, runqueue_t *this_rq, int this_cpu)
 {
 	dequeue_task(p, src_array);
-	src_rq->nr_running--;
+	nr_running_dec(src_rq);
 	set_task_cpu(p, this_cpu);
-	this_rq->nr_running++;
+	nr_running_inc(this_rq);
 	enqueue_task(p, this_rq->active);
 	/*
 	 * Note that idle threads have a prio of MAX_PRIO, for this test
@@ -763,8 +867,21 @@ static void load_balance(runqueue_t *thi
 	prio_array_t *array;
 	struct list_head *head, *curr;
 	task_t *tmp;
+	int this_node = __cpu_to_node(this_cpu);
+	unsigned long cpumask = __node_to_cpu_mask(this_node);
 
-	busiest = find_busiest_queue(this_rq, this_cpu, idle, &imbalance);
+#if CONFIG_NUMA
+	/*
+	 * Avoid rebalancing between nodes too often.
+	 */
+	if (!(++(this_rq->nr_balanced) % NODE_BALANCE_RATIO)) {
+		int node = find_busiest_node(this_node);
+		if (node >= 0)
+			cpumask = __node_to_cpu_mask(node) | (1UL << this_cpu);
+	}
+#endif
+	busiest = find_busiest_queue(this_rq, this_cpu, idle, &imbalance,
+				    cpumask);
 	if (!busiest)
 		goto out;
 
@@ -2231,6 +2348,10 @@ void __init sched_init(void)
 		spin_lock_init(&rq->lock);
 		INIT_LIST_HEAD(&rq->migration_queue);
 		atomic_set(&rq->nr_iowait, 0);
+#if CONFIG_NUMA
+		rq->node_ptr = &node_nr_running[0];
+#endif /* CONFIG_NUMA */
+
 
 		for (j = 0; j < 2; j++) {
 			array = rq->arrays + j;