Re: [PATCH v5 sched_ext/for-7.3 18/33] sched_ext: Add reject DSQ for cap-rejected dispatches

[Tejun Heo]

Hello, Andrea.

On Tue, Jul 14, 2026 at 08:49:49AM +0200, Andrea Righi wrote:
> With CONFIG_EXT_SUB_SCHED=y, this calls scx_local_or_reject_dsq() for every
> local DSQ insertion even when there are no sub-schedulers attached. It shouldn't
> be too bad, but the hot path still pays for the out-of-line call and the
> CID/capability bookkeeping.
>
> Maybe the same static key that I suggested for the ops.update_idle() case
> (PATCH 28/33) can be used here as well and we could do something like:
>
>     if (!static_branch_unlikely(&scx_has_sub_scheds))
>         return &rq->scx.local_dsq;

Yeah, this and the other sub-sched hot paths will be gated behind the
static key in follow-up patches.

Thanks.

--
tejun