Re: [PATCH v4 sched_ext/for-7.3 32/40] sched_ext: Add the SCX_CAP_ENQ cap

[Tejun Heo]

On Wed, 08 Jul 2026 22:00:09 +0000, sashiko-bot@xxxxxxxxxx wrote:
> - [High] Tasks with carried-in IMMED state are spuriously rejected
> during internal movements (like RESTORE) if the scheduler lacks
> SCX_CAP_ENQ.

Correct - an in-place RESTORE requeue of a running task re-runs cid
admission requiring SCX_CAP_ENQ while the enqueue carries no
SCX_ENQ_IMMED, so a sub holding only ENQ_IMMED bounces a task it's
already running. The restore isn't a real scheduling event
(set_next_task_scx() follows immediately) and shouldn't re-run admission
at all. Fixed in v5: the restore path marks the enqueue with an internal
IGNORE_CAPS flag which scx_caps_for_enq() maps to no required caps, split
out as a prep patch.

Thanks.

-- 
tejun