linux-kernel - Re: [PATCHSET sched_ext/for-6.11] sched_ext: Allow SCX_DSQ_LOCAL

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

Message-ID: <Zo5OATwDf5uGjO2w@gpd>
Date: Wed, 10 Jul 2024 11:01:53 +0200
From: Andrea Righi <righi.andrea@...il.com>
To: Tejun Heo <tj@...nel.org>
Cc: void@...ifault.com, linux-kernel@...r.kernel.org, kernel-team@...a.com,
	schatzberg.dan@...il.com, mingo@...hat.com, peterz@...radead.org,
	changwoo@...lia.com
Subject: Re: [PATCHSET sched_ext/for-6.11] sched_ext: Allow SCX_DSQ_LOCAL_ON
 for direct dispatches

On Tue, Jul 09, 2024 at 11:21:06AM -1000, Tejun Heo wrote:
> Hello,
> 
> In ops.dispatch(), SCX_DSQ_LOCAL_ON can be used to dispatch the task to the
> local DSQ of any CPU. However, during direct dispatch from ops.select_cpu()
> and ops.enqueue(), this isn't allowed. This is because dispatching to the
> local DSQ of a remote CPU requires locking both the task's current and new
> rq's and such double locking can't be done directly from ops.enqueue(). This
> gap in API forces schedulers into work-arounds which are not straightforward
> or optimal such as skipping direct dispatches in such cases.
> 
> This patchset implements SCX_DSQ_LOCAL_ON support for direct dispatches.
> 
> This patchset contains the following patches:
> 
>  0001-sched-Move-struct-balance_callback-definition-upward.patch
>  0002-sched_ext-Open-code-task_linked_on_dsq.patch
>  0003-sched_ext-Make-rf-optional-for-dispatch_to_local_dsq.patch
>  0004-sched_ext-s-SCX_RQ_BALANCING-SCX_RQ_IN_BALANCE-and-a.patch
>  0005-sched_ext-Allow-SCX_DSQ_LOCAL_ON-for-direct-dispatch.patch
>  0006-sched_ext-scx_qmap-Pick-idle-CPU-for-direct-dispatch.patch
> 
> While 0001 is a scheduler core change, all it does is moving the definition
> of struct balance_callback so that it's visible for struct scx_rq
> definition. If there are no objections, it'd make sense to route it through
> the sched_ext tree with the rest of the changes.

Apart for the minor nits, everything looks good to me. I did a quick
test inside virtme-ng and I haven't found any regression. I'll do more
tests with the new way of using SCX_DSQ_LOCAL_ON, but so far so good.
FWIW:

Tested-by: Andrea Righi <righi.andrea@...il.com>

Thanks,
-Andrea

> 
> and is also available in the following git branch:
> 
>  git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext.git scx-ddsp-local-on
> 
>  kernel/sched/ext.c             |  207 +++++++++++++++++++++++++++++++++++++++++++++++++++------------
>  kernel/sched/sched.h           |   19 +++--
>  tools/sched_ext/scx_qmap.bpf.c |   39 +++++++++--
>  tools/sched_ext/scx_qmap.c     |    5 -
>  4 files changed, 215 insertions(+), 55 deletions(-)
> 
> --
> tejun