linux-kernel - Re: workqueue_set_max

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: <20111215193133.GF32002@google.com>
Date:	Thu, 15 Dec 2011 11:31:33 -0800
From:	Tejun Heo <tj@...nel.org>
To:	Johannes Berg <johannes@...solutions.net>
Cc:	LKML <linux-kernel@...r.kernel.org>
Subject: Re: workqueue_set_max_active(wq, 0)?

Hello,

On Thu, Dec 15, 2011 at 08:26:01PM +0100, Johannes Berg wrote:
> Ah. So fundamentally, the freeze code does:
> 
>  * set each gcwq frozen
>  * set max_active=0 for each CWQ in each WQ

Yeap and then iterate over them waiting for all nr_actives to drop to
zero.

> but it interleaves the two loops. I guess this would have to be
> untangled if we want to share it so it sets all gcwq frozen and then
> iterates the workqueues and their CWQs. Locking seems a bit hairy
> though, why does the current code keep the GCWQ lock over CWQ changes? I
> guess that's so nothing can work on the CWQ?

All CWQ's are protected by the corresponding GCWQ lock, so all CWQs on
the same CPU are protected by single gcwq->lock on that CPU.  It's
actually rather simple.  The reason the loop there is interleaved is
to avoid releasing and grabbing different gcwq->lock's for each
iteration.  I don't think that would really matter either way.

Thanks.

-- 
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/