linux-kernel - Re: [PATCH 1/7] sched_ext: Implement event counter infrastructure and add an event

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <Z4qEwKNcqIx25cJE@slm.duckdns.org>
Date: Fri, 17 Jan 2025 06:26:40 -1000
From: Tejun Heo <tj@...nel.org>
To: Andrea Righi <arighi@...dia.com>
Cc: Changwoo Min <changwoo@...lia.com>, void@...ifault.com,
	kernel-dev@...lia.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/7] sched_ext: Implement event counter infrastructure
 and add an event

Hello,

On Fri, Jan 17, 2025 at 10:49:55AM +0100, Andrea Righi wrote:
> On Fri, Jan 17, 2025 at 12:15:37AM +0900, Changwoo Min wrote:
> ...
> > +/*
> > + * The event counter is organized by a per-CPU variable to minimize the
> > + * accounting overhead without synchronization. A system-wide view on the
> > + * event counter is constructed when requested by scx_bpf_get_event_stat().
> > + */
> > +static DEFINE_PER_CPU(struct scx_event_stat, event_stats);
> 
> Should we consider tracking these statistics per-scheduler rather than
> globally (like adding scx_event_stat to sched_ext_ops)?
> 
> It's not particularly important for now, but in the future, if we allow
> multiple scx schedulers to be loaded at the same time, tracking separate
> stats per-scheduler would be preferable.

Yeah, eventually, but I don't think it affects anything user visible right
now. Later, we may add an interface to query the stats for a specific
scheduler provided the scheduler is a child of the current one and so on but
we'd still need an interface to query "my stats" which the current interface
can become.

Thanks.

-- 
tejun