Re: [PATCH v7 02/15] sched/core: uclamp: Enforce last task UCLAMP_MAX

On Wed, Mar 13, 2019 at 04:20:51PM +0000, Patrick Bellasi wrote:

On 13-Mar 15:10, Peter Zijlstra wrote:

quoted

On Fri, Feb 08, 2019 at 10:05:41AM +0000, Patrick Bellasi wrote:

quoted

+uclamp_idle_value(struct rq *rq, unsigned int clamp_id, unsigned int clamp_value)
+{
+	/*
+	 * Avoid blocked utilization pushing up the frequency when we go
+	 * idle (which drops the max-clamp) by retaining the last known
+	 * max-clamp.
+	 */
+	if (clamp_id == UCLAMP_MAX) {
+		rq->uclamp_flags |= UCLAMP_FLAG_IDLE;
+		return clamp_value;
+	}
+
+	return uclamp_none(UCLAMP_MIN);

That's a very complicated way or writing: return 0, right?

In my mind it's just a simple way to hardcode values in just one place.

In the current implementation uclamp_none(UCLAMP_MIN) is 0 and the
compiler is not in trubles to inline a 0 there.

Is it really so disgusting ?

Not disguisting per se, just complicated. It had me go back and check
wth uclamp_none() did again.