linux-kernel - Re: [PATCH v4 1/2] workqueue: add new schedule_on_cpu

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: <520514FB.8060502@tilera.com>
Date:	Fri, 9 Aug 2013 12:12:43 -0400
From:	Chris Metcalf <cmetcalf@...era.com>
To:	Tejun Heo <tj@...nel.org>
CC:	<linux-kernel@...r.kernel.org>, <linux-mm@...ck.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Frederic Weisbecker <fweisbec@...il.com>,
	Cody P Schafer <cody@...ux.vnet.ibm.com>
Subject: Re: [PATCH v4 1/2] workqueue: add new schedule_on_cpu_mask() API

On 8/9/2013 11:02 AM, Tejun Heo wrote:
> Hello, Chris.
>
> On Wed, Aug 07, 2013 at 04:49:44PM -0400, Chris Metcalf wrote:
>> This primitive allows scheduling work to run on a particular set of
>> cpus described by a "struct cpumask".  This can be useful, for example,
>> if you have a per-cpu variable that requires code execution only if the
>> per-cpu variable has a certain value (for example, is a non-empty list).
> So, this allows scheduling work items on !online CPUs.  Workqueue does
> allow scheduling per-cpu work items on offline CPUs if the CPU has
> ever been online, but the behavior when scheduling work items on cpu
> which has never been online is undefined.  I think the interface at
> least needs to verify that the the target cpus have been online,
> trigger warning and mask off invalid CPUs otherwise.

I could certainly make schedule_on_cpu_mask() do sanity checking, perhaps via a WARN_ON_ONCE() if offline cpus were specified, and otherwise just have it create a local struct cpumask that it and's with cpu_online_mask, suitably wrapping with get_online_cpus()/put_online_cpus().  (I'm not sure how to test if a cpu has ever been online, vs whether it's online right now.)  I don't want to unnecessarily slow down the existing schedule_on_each_cpu(), so perhaps the implementation should have a static schedule_on_cpu_mask_internal() function that is the same as my previous schedule_on_cpu_mask(), allowing schedule_on_each_cpu() to call it directly to bypass the checking.

That said... I wonder if it might make sense to treat this API the same as other APIs that already take a cpu?  schedule_work_on(), schedule_delayed_work_on(), and queue_delayed_work_on() all take a cpu parameter without API comment or validity checking; queue_work_on() just says "the caller must ensure [the cpu] can't go away".  Does it make sense to just add a similar comment to schedule_on_cpu_mask() rather than make this API the first to actually do cpu validity checking?

Let me know; I'm happy to respin it either way.

-- 
Chris Metcalf, Tilera Corp.
http://www.tilera.com

--
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/