Re: [PATCH sched_ext/for-6.15 v3 3/5] sched_ext: Add scx_kfunc_ids_ops_context_sensitive for unified filtering of context-sensitive SCX kfuncs

[Tejun Heo <tj@kernel.org>]

Hello,

On Thu, Feb 27, 2025 at 06:38:32PM -0800, Alexei Starovoitov wrote:
...
> > > Not from this change but these can probably be allowed from TRACING too.
> > >
> >
> > Not sure if it is safe to make these kfuncs available in TRACING.
> > If Alexei sees this email, could you please leave a comment?
> 
> Hold on, you want to enable all of scx_kfunc_ids_unlocked[] set
> to all of TRACING ? What is the use case ?

I thought it may be useful to be able to iterate DSQs and trigger
dispatching from there but

> Maybe it's safe, but without in-depth analysis we shouldn't.
> Currently sched-ext allows scx_kfunc_set_any[] for tracing.
> I would stick to that in this patch set.

I haven't thought through about safety at all and was mostly naively
thinking that if _any is safe _unlocked should too. Right now, unlocked has
two groups of kfuncs - the ones that should be able to sleep and the ones
that can be called from within scx_dsq_iter iterations. The former is
excluded from TRACING through KF_SLEEPABLE, I think. I don't know whether
dsq iteration is allowed in TRACING.

Anyways, not a real issue either way.

Thanks.

-- 
tejun