node affine NUMA scheduler 3/5

PATCH New Erich Focht 1 participant 1 message Oct 14, 2002 Scheduler

lore

Share & tools

lore

Erich Focht

author

2002-10-14 11:05 UTC #0 lore raw

PATCH 1/1 node affine NUMA scheduler 3/5

---------- Resent Message ----------

Subject: [PATCH] node affine NUMA scheduler 3/5
Date: Fri, 11 Oct 2002 19:58:14 +0200

This is the third part of the node affine NUMA scheduler.

Quoted 4 lines

03-node_affine-2.5.39-10.patch :
This is the heart of the node affine part of the patch. Tasks are assigned a homenode during initial load balancing and they are attracted to the homenode.

Erich

include/linux/sched.h +7 -0

diff -urNp a/include/linux/sched.h b/include/linux/sched.h
@@ -301,6 +301,7 @@ struct task_struct {
 	unsigned long policy;
 	unsigned long cpus_allowed;
 	unsigned int time_slice, first_time_slice;
+	int node;
 
 	struct list_head tasks;
 	struct list_head ptrace_children;

@@ -462,8 +463,14 @@ extern void build_pools(void);
 extern void pooldata_lock(void);
 extern void pooldata_unlock(void);
 extern void sched_balance_exec(void);
+extern void set_task_node(task_t *p, int node);
+# define homenode_inc(rq,node) (rq)->nr_homenode[node]++
+# define homenode_dec(rq,node) (rq)->nr_homenode[node]--
 #else
 #define sched_balance_exec() {}
+#define set_task_node(p,n) {}
+# define homenode_inc(rq,node) {}
+# define homenode_dec(rq,node) {}
 #endif
 extern void sched_migrate_task(task_t *p, int cpu);

kernel/sched.c +105 -2

210 lines not on this page · raw message

kernel/softirq.c +1 -0

diff -urNp a/kernel/softirq.c b/kernel/softirq.c
@@ -368,6 +368,7 @@ static int ksoftirqd(void * __bind_cpu)
 	set_cpus_allowed(current, 1UL << cpu);
 	if (smp_processor_id() != cpu)
 		BUG();
+	set_task_node(current, __cpu_to_node(cpu));
 
 	sprintf(current->comm, "ksoftirqd_CPU%d", cpu);