[PATCH] sched: new sched domain for representing multi-core

Add a new sched domain for representing multi-core with shared caches
between cores.  Consider a dual package system, each package containing two
cores and with last level cache shared between cores with in a package.  If
there are two runnable processes, with this appended patch those two
processes will be scheduled on different packages.

On such systems, with this patch we have observed 8% perf improvement with
specJBB(2 warehouse) benchmark and 35% improvement with CFP2000 rate(with 2
users).

This new domain will come into play only on multi-core systems with shared
caches.  On other systems, this sched domain will be removed by domain
degeneration code.  This new domain can be also used for implementing power
savings policy (see OLS 2005 CMP kernel scheduler paper for more details..
I will post another patch for power savings policy soon)

Most of the arch/* file changes are for cpu_coregroup_map() implementation.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 68 insertions, 5 deletions

diff --git a/kernel/sched.c b/kernel/sched.c
index a96a05d..8a8b71b 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -5574,11 +5574,31 @@ static int cpu_to_cpu_group(int cpu)
 }
 #endif
 
+#ifdef CONFIG_SCHED_MC
+static DEFINE_PER_CPU(struct sched_domain, core_domains);
+static struct sched_group sched_group_core[NR_CPUS];
+#endif
+
+#if defined(CONFIG_SCHED_MC) && defined(CONFIG_SCHED_SMT)
+static int cpu_to_core_group(int cpu)
+{
+	return first_cpu(cpu_sibling_map[cpu]);
+}
+#elif defined(CONFIG_SCHED_MC)
+static int cpu_to_core_group(int cpu)
+{
+	return cpu;
+}
+#endif
+
 static DEFINE_PER_CPU(struct sched_domain, phys_domains);
 static struct sched_group sched_group_phys[NR_CPUS];
 static int cpu_to_phys_group(int cpu)
 {
-#ifdef CONFIG_SCHED_SMT
+#if defined(CONFIG_SCHED_MC)
+	cpumask_t mask = cpu_coregroup_map(cpu);
+	return first_cpu(mask);
+#elif defined(CONFIG_SCHED_SMT)
 	return first_cpu(cpu_sibling_map[cpu]);
 #else
 	return cpu;
@@ -5676,6 +5696,17 @@ void build_sched_domains(const cpumask_t *cpu_map)
 		sd->parent = p;
 		sd->groups = &sched_group_phys[group];
 
+#ifdef CONFIG_SCHED_MC
+		p = sd;
+		sd = &per_cpu(core_domains, i);
+		group = cpu_to_core_group(i);
+		*sd = SD_MC_INIT;
+		sd->span = cpu_coregroup_map(i);
+		cpus_and(sd->span, sd->span, *cpu_map);
+		sd->parent = p;
+		sd->groups = &sched_group_core[group];
+#endif
+
 #ifdef CONFIG_SCHED_SMT
 		p = sd;
 		sd = &per_cpu(cpu_domains, i);
@@ -5701,6 +5732,19 @@ void build_sched_domains(const cpumask_t *cpu_map)
 	}
 #endif
 
+#ifdef CONFIG_SCHED_MC
+	/* Set up multi-core groups */
+	for_each_cpu_mask(i, *cpu_map) {
+		cpumask_t this_core_map = cpu_coregroup_map(i);
+		cpus_and(this_core_map, this_core_map, *cpu_map);
+		if (i != first_cpu(this_core_map))
+			continue;
+		init_sched_build_groups(sched_group_core, this_core_map,
+					&cpu_to_core_group);
+	}
+#endif
+
+
 	/* Set up physical groups */
 	for (i = 0; i < MAX_NUMNODES; i++) {
 		cpumask_t nodemask = node_to_cpumask(i);
@@ -5797,11 +5841,31 @@ void build_sched_domains(const cpumask_t *cpu_map)
 		power = SCHED_LOAD_SCALE;
 		sd->groups->cpu_power = power;
 #endif
+#ifdef CONFIG_SCHED_MC
+		sd = &per_cpu(core_domains, i);
+		power = SCHED_LOAD_SCALE + (cpus_weight(sd->groups->cpumask)-1)
+					    * SCHED_LOAD_SCALE / 10;
+		sd->groups->cpu_power = power;
+
+		sd = &per_cpu(phys_domains, i);
 
+ 		/*
+ 		 * This has to be < 2 * SCHED_LOAD_SCALE
+ 		 * Lets keep it SCHED_LOAD_SCALE, so that
+ 		 * while calculating NUMA group's cpu_power
+ 		 * we can simply do
+ 		 *  numa_group->cpu_power += phys_group->cpu_power;
+ 		 *
+ 		 * See "only add power once for each physical pkg"
+ 		 * comment below
+ 		 */
+ 		sd->groups->cpu_power = SCHED_LOAD_SCALE;
+#else
 		sd = &per_cpu(phys_domains, i);
 		power = SCHED_LOAD_SCALE + SCHED_LOAD_SCALE *
 				(cpus_weight(sd->groups->cpumask)-1) / 10;
 		sd->groups->cpu_power = power;
+#endif
 
 #ifdef CONFIG_NUMA
 		sd = &per_cpu(allnodes_domains, i);
@@ -5823,7 +5887,6 @@ void build_sched_domains(const cpumask_t *cpu_map)
 next_sg:
 		for_each_cpu_mask(j, sg->cpumask) {
 			struct sched_domain *sd;
-			int power;
 
 			sd = &per_cpu(phys_domains, j);
 			if (j != first_cpu(sd->groups->cpumask)) {
@@ -5833,10 +5896,8 @@ next_sg:
 				 */
 				continue;
 			}
-			power = SCHED_LOAD_SCALE + SCHED_LOAD_SCALE *
-				(cpus_weight(sd->groups->cpumask)-1) / 10;
 
-			sg->cpu_power += power;
+			sg->cpu_power += sd->groups->cpu_power;
 		}
 		sg = sg->next;
 		if (sg != sched_group_nodes[i])
@@ -5849,6 +5910,8 @@ next_sg:
 		struct sched_domain *sd;
 #ifdef CONFIG_SCHED_SMT
 		sd = &per_cpu(cpu_domains, i);
+#elif defined(CONFIG_SCHED_MC)
+		sd = &per_cpu(core_domains, i);
 #else
 		sd = &per_cpu(phys_domains, i);
 #endif