sched: Clean up remnants of sd_idle
authorPeter Zijlstra <a.p.zijlstra@chello.nl>
Mon, 21 Feb 2011 17:52:53 +0000 (18:52 +0100)
committerIngo Molnar <mingo@elte.hu>
Wed, 23 Feb 2011 10:33:55 +0000 (11:33 +0100)
With the wholesale removal of the sd_idle SMT logic we can clean up
some more.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nikhil Rao <ncrao@google.com>
Cc: Venkatesh Pallipadi <venki@google.com>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Mike Galbraith <efault@gmx.de>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
kernel/sched_fair.c

index d384e73..cd18600 100644 (file)
@@ -3150,25 +3150,23 @@ find_busiest_group(struct sched_domain *sd, int this_cpu,
        if (sds.this_load >= sds.avg_load)
                goto out_balanced;
 
-       /*
-        * In the CPU_NEWLY_IDLE, use imbalance_pct to be conservative.
-        * And to check for busy balance use !idle_cpu instead of
-        * CPU_NOT_IDLE. This is because HT siblings will use CPU_NOT_IDLE
-        * even when they are idle.
-        */
-       if (idle == CPU_NEWLY_IDLE || !idle_cpu(this_cpu)) {
-               if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load)
-                       goto out_balanced;
-       } else {
+       if (idle == CPU_IDLE) {
                /*
                 * This cpu is idle. If the busiest group load doesn't
                 * have more tasks than the number of available cpu's and
                 * there is no imbalance between this and busiest group
                 * wrt to idle cpu's, it is balanced.
                 */
-               if ((sds.this_idle_cpus  <= sds.busiest_idle_cpus + 1) &&
+               if ((sds.this_idle_cpus <= sds.busiest_idle_cpus + 1) &&
                    sds.busiest_nr_running <= sds.busiest_group_weight)
                        goto out_balanced;
+       } else {
+               /*
+                * In the CPU_NEWLY_IDLE, CPU_NOT_IDLE cases, use
+                * imbalance_pct to be conservative.
+                */
+               if (100 * sds.max_load <= sd->imbalance_pct * sds.this_load)
+                       goto out_balanced;
        }
 
 force_balance:
@@ -3862,8 +3860,7 @@ static void rebalance_domains(int cpu, enum cpu_idle_type idle)
                        if (load_balance(cpu, rq, sd, idle, &balance)) {
                                /*
                                 * We've pulled tasks over so either we're no
-                                * longer idle, or one of our SMT siblings is
-                                * not idle.
+                                * longer idle.
                                 */
                                idle = CPU_NOT_IDLE;
                        }