Re: [PATCH v13 06/13] sched/core: Try to use a preferred CPU in is_cpu_allowed
flat view
From: Peter Zijlstra <peterz@infradead.org>
Date: 2026-09-25 07:43:34
Also in:
linux-doc, lkml
On Wed, Sep 09, 2026 at 07:26:10PM +0530, Shrikanth Hegde wrote:
quoted hunk ↗ jump to hunk
diff --git a/kernel/sched/core.c b/kernel/sched/core.c index a689a0cea4eb..b4ef2e92d786 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c@@ -2504,6 +2504,23 @@ static inline bool rq_has_pinned_tasks(struct rq *rq) return rq->nr_pinned; } +static inline bool task_can_sched_on_preferred(int cpu, struct task_struct *p)
(somehow those arguments feel backwards)
quoted hunk ↗ jump to hunk
+{ + if (cpu_preferred(cpu)) + return false;
FWIW, this reads very weird. task_can_sched_on_preferred() returning false when task is on a preferred cpu. The question really is more like: can we migrate to a preferred CPU, and in that context it makes more sense. No point in migrating if we are already on one. Perhaps a comment on top of the function can clarify?
quoted hunk ↗ jump to hunk
+ /* Only FAIR tasks honor preferred CPU state */ + if (unlikely(p->sched_class != &fair_sched_class)) + return false; + + /* Ignore preferred state if task affinity is changing */ + if (unlikely(!cpumask_test_cpu(task_cpu(p), p->cpus_ptr))) + return false; + + return cpumask_intersects_and(p->cpus_ptr, cpu_preferred_mask, + task_cpu_possible_mask(p)); +}