]> err.no Git - linux-2.6/commitdiff
[PATCH] Dynamic sched domains: ia64 changes
authorDinakar Guniguntala <dino@in.ibm.com>
Sat, 25 Jun 2005 21:57:36 +0000 (14:57 -0700)
committerLinus Torvalds <torvalds@ppc970.osdl.org>
Sat, 25 Jun 2005 23:24:45 +0000 (16:24 -0700)
ia64 changes similar to kernel/sched.c.

Signed-off-by: Dinakar Guniguntala <dino@in.ibm.com>
Acked-by: Paul Jackson <pj@sgi.com>
Acked-by: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
arch/ia64/kernel/domain.c

index afbde79c3b3df36977af352c63b89215f452ee3d..d65e87b6394fa2c636f9593718f6e84f29c9804c 100644 (file)
@@ -27,7 +27,7 @@
  *
  * Should use nodemask_t.
  */
-static int __devinit find_next_best_node(int node, unsigned long *used_nodes)
+static int find_next_best_node(int node, unsigned long *used_nodes)
 {
        int i, n, val, min_val, best_node = 0;
 
@@ -66,7 +66,7 @@ static int __devinit find_next_best_node(int node, unsigned long *used_nodes)
  * should be one that prevents unnecessary balancing, but also spreads tasks
  * out optimally.
  */
-static cpumask_t __devinit sched_domain_node_span(int node)
+static cpumask_t sched_domain_node_span(int node)
 {
        int i;
        cpumask_t span, nodemask;
@@ -96,7 +96,7 @@ static cpumask_t __devinit sched_domain_node_span(int node)
 #ifdef CONFIG_SCHED_SMT
 static DEFINE_PER_CPU(struct sched_domain, cpu_domains);
 static struct sched_group sched_group_cpus[NR_CPUS];
-static int __devinit cpu_to_cpu_group(int cpu)
+static int cpu_to_cpu_group(int cpu)
 {
        return cpu;
 }
@@ -104,7 +104,7 @@ static int __devinit cpu_to_cpu_group(int cpu)
 
 static DEFINE_PER_CPU(struct sched_domain, phys_domains);
 static struct sched_group sched_group_phys[NR_CPUS];
-static int __devinit cpu_to_phys_group(int cpu)
+static int cpu_to_phys_group(int cpu)
 {
 #ifdef CONFIG_SCHED_SMT
        return first_cpu(cpu_sibling_map[cpu]);
@@ -125,44 +125,36 @@ static struct sched_group *sched_group_nodes[MAX_NUMNODES];
 static DEFINE_PER_CPU(struct sched_domain, allnodes_domains);
 static struct sched_group sched_group_allnodes[MAX_NUMNODES];
 
-static int __devinit cpu_to_allnodes_group(int cpu)
+static int cpu_to_allnodes_group(int cpu)
 {
        return cpu_to_node(cpu);
 }
 #endif
 
 /*
- * Set up scheduler domains and groups.  Callers must hold the hotplug lock.
+ * Build sched domains for a given set of cpus and attach the sched domains
+ * to the individual cpus
  */
-void __devinit arch_init_sched_domains(void)
+void build_sched_domains(const cpumask_t *cpu_map)
 {
        int i;
-       cpumask_t cpu_default_map;
 
        /*
-        * Setup mask for cpus without special case scheduling requirements.
-        * For now this just excludes isolated cpus, but could be used to
-        * exclude other special cases in the future.
+        * Set up domains for cpus specified by the cpu_map.
         */
-       cpus_complement(cpu_default_map, cpu_isolated_map);
-       cpus_and(cpu_default_map, cpu_default_map, cpu_online_map);
-
-       /*
-        * Set up domains. Isolated domains just stay on the dummy domain.
-        */
-       for_each_cpu_mask(i, cpu_default_map) {
+       for_each_cpu_mask(i, *cpu_map) {
                int group;
                struct sched_domain *sd = NULL, *p;
                cpumask_t nodemask = node_to_cpumask(cpu_to_node(i));
 
-               cpus_and(nodemask, nodemask, cpu_default_map);
+               cpus_and(nodemask, nodemask, *cpu_map);
 
 #ifdef CONFIG_NUMA
                if (num_online_cpus()
                                > SD_NODES_PER_DOMAIN*cpus_weight(nodemask)) {
                        sd = &per_cpu(allnodes_domains, i);
                        *sd = SD_ALLNODES_INIT;
-                       sd->span = cpu_default_map;
+                       sd->span = *cpu_map;
                        group = cpu_to_allnodes_group(i);
                        sd->groups = &sched_group_allnodes[group];
                        p = sd;
@@ -173,7 +165,7 @@ void __devinit arch_init_sched_domains(void)
                *sd = SD_NODE_INIT;
                sd->span = sched_domain_node_span(cpu_to_node(i));
                sd->parent = p;
-               cpus_and(sd->span, sd->span, cpu_default_map);
+               cpus_and(sd->span, sd->span, *cpu_map);
 #endif
 
                p = sd;
@@ -190,7 +182,7 @@ void __devinit arch_init_sched_domains(void)
                group = cpu_to_cpu_group(i);
                *sd = SD_SIBLING_INIT;
                sd->span = cpu_sibling_map[i];
-               cpus_and(sd->span, sd->span, cpu_default_map);
+               cpus_and(sd->span, sd->span, *cpu_map);
                sd->parent = p;
                sd->groups = &sched_group_cpus[group];
 #endif
@@ -198,9 +190,9 @@ void __devinit arch_init_sched_domains(void)
 
 #ifdef CONFIG_SCHED_SMT
        /* Set up CPU (sibling) groups */
-       for_each_cpu_mask(i, cpu_default_map) {
+       for_each_cpu_mask(i, *cpu_map) {
                cpumask_t this_sibling_map = cpu_sibling_map[i];
-               cpus_and(this_sibling_map, this_sibling_map, cpu_default_map);
+               cpus_and(this_sibling_map, this_sibling_map, *cpu_map);
                if (i != first_cpu(this_sibling_map))
                        continue;
 
@@ -213,7 +205,7 @@ void __devinit arch_init_sched_domains(void)
        for (i = 0; i < MAX_NUMNODES; i++) {
                cpumask_t nodemask = node_to_cpumask(i);
 
-               cpus_and(nodemask, nodemask, cpu_default_map);
+               cpus_and(nodemask, nodemask, *cpu_map);
                if (cpus_empty(nodemask))
                        continue;
 
@@ -222,7 +214,7 @@ void __devinit arch_init_sched_domains(void)
        }
 
 #ifdef CONFIG_NUMA
-       init_sched_build_groups(sched_group_allnodes, cpu_default_map,
+       init_sched_build_groups(sched_group_allnodes, *cpu_map,
                                &cpu_to_allnodes_group);
 
        for (i = 0; i < MAX_NUMNODES; i++) {
@@ -233,12 +225,12 @@ void __devinit arch_init_sched_domains(void)
                cpumask_t covered = CPU_MASK_NONE;
                int j;
 
-               cpus_and(nodemask, nodemask, cpu_default_map);
+               cpus_and(nodemask, nodemask, *cpu_map);
                if (cpus_empty(nodemask))
                        continue;
 
                domainspan = sched_domain_node_span(i);
-               cpus_and(domainspan, domainspan, cpu_default_map);
+               cpus_and(domainspan, domainspan, *cpu_map);
 
                sg = kmalloc(sizeof(struct sched_group), GFP_KERNEL);
                sched_group_nodes[i] = sg;
@@ -266,7 +258,7 @@ void __devinit arch_init_sched_domains(void)
                        int n = (i + j) % MAX_NUMNODES;
 
                        cpus_complement(notcovered, covered);
-                       cpus_and(tmp, notcovered, cpu_default_map);
+                       cpus_and(tmp, notcovered, *cpu_map);
                        cpus_and(tmp, tmp, domainspan);
                        if (cpus_empty(tmp))
                                break;
@@ -293,7 +285,7 @@ void __devinit arch_init_sched_domains(void)
 #endif
 
        /* Calculate CPU power for physical packages and nodes */
-       for_each_cpu_mask(i, cpu_default_map) {
+       for_each_cpu_mask(i, *cpu_map) {
                int power;
                struct sched_domain *sd;
 #ifdef CONFIG_SCHED_SMT
@@ -359,13 +351,35 @@ next_sg:
                cpu_attach_domain(sd, i);
        }
 }
+/*
+ * Set up scheduler domains and groups.  Callers must hold the hotplug lock.
+ */
+void arch_init_sched_domains(const cpumask_t *cpu_map)
+{
+       cpumask_t cpu_default_map;
+
+       /*
+        * Setup mask for cpus without special case scheduling requirements.
+        * For now this just excludes isolated cpus, but could be used to
+        * exclude other special cases in the future.
+        */
+       cpus_andnot(cpu_default_map, *cpu_map, cpu_isolated_map);
+
+       build_sched_domains(&cpu_default_map);
+}
 
-void __devinit arch_destroy_sched_domains(void)
+void arch_destroy_sched_domains(const cpumask_t *cpu_map)
 {
 #ifdef CONFIG_NUMA
        int i;
        for (i = 0; i < MAX_NUMNODES; i++) {
+               cpumask_t nodemask = node_to_cpumask(i);
                struct sched_group *oldsg, *sg = sched_group_nodes[i];
+
+               cpus_and(nodemask, nodemask, *cpu_map);
+               if (cpus_empty(nodemask))
+                       continue;
+
                if (sg == NULL)
                        continue;
                sg = sg->next;