sched: Avoid expensive initial update_cfs_load()

Since cfs->{load_stamp,load_last} are zero-initalized the initial load update
will consider the delta to be 'since the beginning of time'.

This results in a lot of pointless divisions to bring this large period to be
within the sysctl_sched_shares_window.

Fix this by initializing load_stamp to be 1 at cfs_rq initialization, this
allows for an initial load_stamp > load_last which then lets standard idle
truncation proceed.

We avoid spinning (and slightly improve consistency) by fixing delta to be
[period - 1] in this path resulting in a slightly more predictable shares ramp.
(Previously the amount of idle time preserved by the overflow would range between
[period/2,period-1].)

Signed-off-by: Paul Turner <pjt@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20110122044852.102126037@google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

diff --git a/kernel/sched.c b/kernel/sched.c
index e0fa3ff7f194f8a2c7508eae78e1e78b0cfcd2a5..6820b5b3a9697fedd1ef96495d651dda92cd4183 100644 (file)
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -7796,6 +7796,8 @@ static void init_cfs_rq(struct cfs_rq *cfs_rq, struct rq *rq)
        INIT_LIST_HEAD(&cfs_rq->tasks);
 #ifdef CONFIG_FAIR_GROUP_SCHED
        cfs_rq->rq = rq;
+       /* allow initial update_cfs_load() to truncate */
+       cfs_rq->load_stamp = 1;
 #endif
        cfs_rq->min_vruntime = (u64)(-(1LL << 20));
 }

diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
index 0c550c841eeee51734f63d8eba02fc83f345e370..4cbc9121094cf72fe224b8db5d47d40008eb247b 100644 (file)
--- a/kernel/sched_fair.c
+++ b/kernel/sched_fair.c
@@ -733,6 +733,7 @@ static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
            now - cfs_rq->load_last > 4 * period) {
                cfs_rq->load_period = 0;
                cfs_rq->load_avg = 0;
+               delta = period - 1;
        }
 
        cfs_rq->load_stamp = now;