Re: [PATCH] sched_ext: Count cap-rejected local DSQ inserts in SCX_EV_SUB_REJECT

[luoliang]

From: Liang Luo <luoliang@xxxxxxxxxx>

Hello Tejun,

Thanks for the review.

On Mon, Sep 28, 2026 at 09:06:21AM -1000, Tejun Heo wrote:
> I have nothing against adding the counter, but we don't count every path,
> so an uncounted branch arm isn't a reason by itself. What's the rationale
> for counting this one?

That's fair - a counter should be justified on its own, not by symmetry
with its siblings. The rationale here is that of the three outcomes, the
reject diversion is the one without preconditions: the other two
require a draining rq, a task that cannot migrate anywhere else, or an
explicit SCX_ENQ_RESCUE opt-in, while every other cap-missing insert
lands in the reject DSQ. With this counter, the events cover all three
outcomes of a cap-missing local insert.

> This doesn't fix a bug. Can you drop the Fixes: tag?

Done in v2.

Thanks.

Liang Luo