linux-kernel - Re: [PATCH] workqueue: rust: add creation of workqueues

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: <aAALmSjLyWqrcQ45@slm.duckdns.org>
Date: Wed, 16 Apr 2025 09:57:13 -1000
From: Tejun Heo <tj@...nel.org>
To: Alice Ryhl <aliceryhl@...gle.com>
Cc: Danilo Krummrich <dakr@...nel.org>, Philipp Stanner <phasta@...nel.org>,
	Miguel Ojeda <ojeda@...nel.org>,
	Lai Jiangshan <jiangshanlai@...il.com>,
	Boqun Feng <boqun.feng@...il.com>, Gary Guo <gary@...yguo.net>,
	Björn Roy Baron <bjorn3_gh@...tonmail.com>,
	Benno Lossin <benno.lossin@...ton.me>,
	Andreas Hindborg <a.hindborg@...nel.org>,
	Trevor Gross <tmgross@...ch.edu>,
	Daniel Almeida <daniel.almeida@...labora.com>,
	Tamir Duberstein <tamird@...il.com>, rust-for-linux@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] workqueue: rust: add creation of workqueues

Hello, Alice.

On Wed, Apr 16, 2025 at 09:41:21PM +0200, Alice Ryhl wrote:
...
> I thought about implementation approaches. The first thought that
> sprang to mind is add a list of all delayed work items, but now I
> think we can do better. We can have an atomic counter tracking the
> number of delayed work items, and have destroy_workqueue() do this:
> 
> retry:
> drain_workqueue(wq);
> if (has_delayed_work_items(wq)) {
>     wait_for_delayed_to_be_scheduled(wq);
>     goto retry;
> }
> 
> where wait_for_delayed_to_be_scheduled() either waits for the counter
> to hit zero, or waits for at least waits for one of them to be
> scheduled. For example, maybe wait_for_delayed_to_be_scheduled() could
> add a dummy work item *without* waking up the worker threads, and then
> wait for that work item to get executed, which would effectively mean
> that it sleeps until something else wakes up a worker.

I suppose that can work too but the delays can be pretty long, so while
correct, I'm not sure it'd be very pleasant to use. If we per-cpu lists, I
don't think the overhead would be all that noticeable, so may as well do
that?

Thanks.

-- 
tejun