linux-kernel - Re: [workqueue/driver-core PATCH 1/5] workqueue: Provide queue_work_near to queue work near a given NUMA node

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]

Date:   Wed, 10 Oct 2018 08:37:02 -0700
From:   Tejun Heo <tj@...nel.org>
To:     Alexander Duyck <alexander.h.duyck@...ux.intel.com>
Cc:     gregkh@...uxfoundation.org, akpm@...ux-foundation.org,
        linux-kernel@...r.kernel.org, len.brown@...el.com,
        rafael@...nel.org, linux-pm@...r.kernel.org,
        jiangshanlai@...il.com, pavel@....cz, zwisler@...nel.org
Subject: Re: [workqueue/driver-core PATCH 1/5] workqueue: Provide
 queue_work_near to queue work near a given NUMA node

Hello,

On Wed, Oct 10, 2018 at 08:34:06AM -0700, Alexander Duyck wrote:
> >Maybe we wanna round-robin within the node?
> 
> I had thought about it. It adds extra complexity to do it and for
> unbound work queues it doesn't really add anything since it gets
> converted back into a node anyway.

Yeap, it only makes sense for per-cpu workqueues.

> If you wanted I could look at adding extra logic that would do the
> round-robin for non-unbound workqueues. I just wasn't sure if it was
> worth the effort since the current users are all unbound workqueues.

It doesn't have to be complex, just a per-node unsynchronized int
which gets incremented and %'s would work fine.  Alternatively, we can
trigger WARN_ON_ONCE() if the function is used on a per-cpu workqueue.

Thanks.

-- 
tejun