[PATCH] sched/fair: Fix the util_avg_cap formula in a comment

[Zhan Xusheng]

The comment documents the initial util_avg cap for a new task as:

    util_avg_cap = (cpu_scale - cfs_rq->avg.util_avg) / 2^n

but the code caps it at half of the remaining budget:

    cap = (long)(cpu_scale - cfs_rq->avg.util_avg) / 2;

The "/ 2^n" form matches neither the code nor the comment's own worked
example: for the 2nd task with a cfs_rq util_avg of 512 on a
1024-capacity CPU it would give (1024 - 512) / 2^2 = 128, whereas both
the code and the tabulated example give (1024 - 512) / 2 = 256.

The cpu_scale / 2^n progression shown in the example instead emerges
from applying the per-task "/ 2" cap repeatedly, since
cfs_rq->avg.util_avg accumulates the previously created tasks. Correct
the formula to match the code and note where the 2^n series comes from.

Fixes: 2b8c41daba32 ("sched/fair: Fix initial util_avg calculation")
Signed-off-by: Zhan Xusheng <zhanxusheng@xxxxxxxxxx>
---
 kernel/sched/fair.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index d78467ec6ee1..df57e32e36ad 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -1299,9 +1299,11 @@ void init_entity_runnable_average(struct sched_entity *se)
  * To solve this problem, we also cap the util_avg of successive tasks to
  * only 1/2 of the left utilization budget:
  *
- *   util_avg_cap = (cpu_scale - cfs_rq->avg.util_avg) / 2^n
+ *   util_avg_cap = (cpu_scale - cfs_rq->avg.util_avg) / 2
  *
- * where n denotes the nth task and cpu_scale the CPU capacity.
+ * where cpu_scale is the CPU capacity.  Since cfs_rq->avg.util_avg
+ * accumulates the previously created tasks, applying this cap to each new
+ * task gives the nth task a util_avg_cap of cpu_scale / 2^n.
  *
  * For example, for a CPU with 1024 of capacity, a simplest series from
  * the beginning would be like:
-- 
2.43.0