linux-kernel - Re: [PATCH 4/4] Workqueue: add WQ_PERCPU to every alloc

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: <aBoCRTWyYkOCN6TE@localhost.localdomain>
Date: Tue, 6 May 2025 14:36:21 +0200
From: Frederic Weisbecker <frederic@...nel.org>
To: Michal Hocko <mhocko@...e.com>
Cc: Marco Crivellari <marco.crivellari@...e.com>,
	linux-kernel@...r.kernel.org, Tejun Heo <tj@...nel.org>,
	Lai Jiangshan <jiangshanlai@...il.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Subject: Re: [PATCH 4/4] Workqueue: add WQ_PERCPU to every alloc_workqueue
 user

Le Mon, May 05, 2025 at 11:22:16AM +0200, Michal Hocko a écrit :
> On Sat 03-05-25 10:28:34, Marco Crivellari wrote:
> > By default, alloc_workqueue() creates bound workqueues
> > (i.e., without WQ_UNBOUND).
> > 
> > With the introduction of the WQ_PERCPU flag (equivalent
> > to !WQ_UNBOUND), any alloc_workqueue() caller that
> > doesn’t explicitly specify WQ_UNBOUND must now use WQ_PERCPU.
> > 
> > All existing users have been updated accordingly.
> 
> I assume that the next step is to drop WQ_PERCPU when it is not strictly
> necessary, right?

Indeed. And also convert the users of schedule_work() and queue_work(system_wq, ...)

> 
> Looking at the huge list this will be a huge undergoing. Is there any
> rule of thumb to tell whether WQ_PERCPU is required or not? Maybe
> something as simple as - if the worker doesn't use any per cpu variables
> then it is highly likely that it doesn't need WQ_PERCPU? Are there other
> useful hints?

None that I know of. But yes it is a very big work. Ideally we
should focus on core users first. And then we can think about
the drivers later...

-- 
Frederic Weisbecker
SUSE Labs