diff -urN oldtree/include/asm-i386/topology.h newtree/include/asm-i386/topology.h
--- oldtree/include/asm-i386/topology.h	2006-09-29 13:50:42.000000000 -0400
+++ newtree/include/asm-i386/topology.h	2006-09-29 14:09:33.000000000 -0400
@@ -74,6 +74,7 @@
 #define SD_NODE_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 8,			\
 	.max_interval		= 32,			\
diff -urN oldtree/include/asm-ia64/topology.h newtree/include/asm-ia64/topology.h
--- oldtree/include/asm-ia64/topology.h	2006-09-29 13:50:42.000000000 -0400
+++ newtree/include/asm-ia64/topology.h	2006-09-29 14:09:33.000000000 -0400
@@ -59,6 +59,7 @@
 #define SD_CPU_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 1,			\
 	.max_interval		= 4,			\
@@ -84,6 +85,7 @@
 #define SD_NODE_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 8,			\
 	.max_interval		= 8*(min(num_online_cpus(), 32)), \
diff -urN oldtree/include/asm-mips/mach-ip27/topology.h newtree/include/asm-mips/mach-ip27/topology.h
--- oldtree/include/asm-mips/mach-ip27/topology.h	2006-09-29 13:50:42.000000000 -0400
+++ newtree/include/asm-mips/mach-ip27/topology.h	2006-09-29 14:09:33.000000000 -0400
@@ -22,6 +22,7 @@
 #define SD_NODE_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 8,			\
 	.max_interval		= 32,			\
diff -urN oldtree/include/asm-powerpc/topology.h newtree/include/asm-powerpc/topology.h
--- oldtree/include/asm-powerpc/topology.h	2006-09-29 13:50:42.000000000 -0400
+++ newtree/include/asm-powerpc/topology.h	2006-09-29 14:09:33.000000000 -0400
@@ -43,6 +43,7 @@
 #define SD_NODE_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 8,			\
 	.max_interval		= 32,			\
diff -urN oldtree/include/asm-x86_64/topology.h newtree/include/asm-x86_64/topology.h
--- oldtree/include/asm-x86_64/topology.h	2006-09-29 13:50:42.000000000 -0400
+++ newtree/include/asm-x86_64/topology.h	2006-09-29 14:09:33.000000000 -0400
@@ -31,6 +31,7 @@
 #define SD_NODE_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 8,			\
 	.max_interval		= 32,			\
diff -urN oldtree/include/linux/sched.h newtree/include/linux/sched.h
--- oldtree/include/linux/sched.h	2006-09-29 14:09:24.000000000 -0400
+++ newtree/include/linux/sched.h	2006-09-29 14:09:33.000000000 -0400
@@ -646,6 +646,7 @@
 struct sched_domain {
 	/* These fields must be setup */
 	struct sched_domain *parent;	/* top domain must be null terminated */
+	struct sched_domain *child;	/* bottom domain must be null terminated */
 	struct sched_group *groups;	/* the balancing groups of the domain */
 	cpumask_t span;			/* span of all CPUs in this domain */
 	unsigned long min_interval;	/* Minimum balance interval ms */
diff -urN oldtree/include/linux/topology.h newtree/include/linux/topology.h
--- oldtree/include/linux/topology.h	2006-09-29 13:50:42.000000000 -0400
+++ newtree/include/linux/topology.h	2006-09-29 14:09:33.000000000 -0400
@@ -89,6 +89,7 @@
 #define SD_SIBLING_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 1,			\
 	.max_interval		= 2,			\
@@ -119,6 +120,7 @@
 #define SD_CPU_INIT (struct sched_domain) {		\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 1,			\
 	.max_interval		= 4,			\
@@ -146,6 +148,7 @@
 #define SD_ALLNODES_INIT (struct sched_domain) {	\
 	.span			= CPU_MASK_NONE,	\
 	.parent			= NULL,			\
+	.child			= NULL,			\
 	.groups			= NULL,			\
 	.min_interval		= 64,			\
 	.max_interval		= 64*num_online_cpus(),	\
diff -urN oldtree/kernel/sched.c newtree/kernel/sched.c
--- oldtree/kernel/sched.c	2006-09-29 14:09:24.000000000 -0400
+++ newtree/kernel/sched.c	2006-09-29 14:09:33.000000000 -0400
@@ -1286,21 +1286,29 @@
 	while (sd) {
 		cpumask_t span;
 		struct sched_group *group;
-		int new_cpu;
-		int weight;
+		int new_cpu, weight;
+
+		if (!(sd->flags & flag)) {
+			sd = sd->child;
+			continue;
+		}
 
 		span = sd->span;
 		group = find_idlest_group(sd, t, cpu);
-		if (!group)
-			goto nextlevel;
+		if (!group) {
+			sd = sd->child;
+			continue;
+		}
 
 		new_cpu = find_idlest_cpu(group, t, cpu);
-		if (new_cpu == -1 || new_cpu == cpu)
-			goto nextlevel;
+		if (new_cpu == -1 || new_cpu == cpu) {
+			/* Now try balancing at a lower domain level of cpu */
+			sd = sd->child;
+			continue;
+		}
 
-		/* Now try balancing at a lower domain level */
+		/* Now try balancing at a lower domain level of new_cpu */
 		cpu = new_cpu;
-nextlevel:
 		sd = NULL;
 		weight = cpus_weight(span);
 		for_each_domain(cpu, tmp) {
@@ -5444,12 +5452,18 @@
 		struct sched_domain *parent = tmp->parent;
 		if (!parent)
 			break;
-		if (sd_parent_degenerate(tmp, parent))
+		if (sd_parent_degenerate(tmp, parent)) {
 			tmp->parent = parent->parent;
+			if (parent->parent)
+				parent->parent->child = tmp;
+		}
 	}
 
-	if (sd && sd_degenerate(sd))
+	if (sd && sd_degenerate(sd)) {
 		sd = sd->parent;
+		if (sd)
+			sd->child = NULL;
+	}
 
 	sched_domain_debug(sd, cpu);
 
@@ -6284,6 +6298,8 @@
 		*sd = SD_NODE_INIT;
 		sd->span = sched_domain_node_span(cpu_to_node(i));
 		sd->parent = p;
+		if (p)
+			p->child = sd;
 		cpus_and(sd->span, sd->span, *cpu_map);
 #endif
 
@@ -6293,6 +6309,8 @@
 		*sd = SD_CPU_INIT;
 		sd->span = nodemask;
 		sd->parent = p;
+		if (p)
+			p->child = sd;
 		sd->groups = &sched_group_phys[group];
 
 #ifdef CONFIG_SCHED_MC
@@ -6303,6 +6321,7 @@
 		sd->span = cpu_coregroup_map(i);
 		cpus_and(sd->span, sd->span, *cpu_map);
 		sd->parent = p;
+		p->child = sd;
 		sd->groups = &sched_group_core[group];
 #endif
 
@@ -6314,6 +6333,7 @@
 		sd->span = cpu_sibling_map[i];
 		cpus_and(sd->span, sd->span, *cpu_map);
 		sd->parent = p;
+		p->child = sd;
 		sd->groups = &sched_group_cpus[group];
 #endif
 	}