From: Lai Jiangshan <laijs@cn.fujitsu.com>
To: Tejun Heo <tj@kernel.org>
Cc: <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 10/10 V2] workqueue: use generic attach/detach routine for rescuers
Date: Tue, 13 May 2014 14:55:24 +0800 [thread overview]
Message-ID: <5371C1DC.8030102@cn.fujitsu.com> (raw)
In-Reply-To: <20140512220505.GI18959@mtj.dyndns.org>
On 05/13/2014 06:05 AM, Tejun Heo wrote:
> On Mon, May 12, 2014 at 02:56:22PM +0800, Lai Jiangshan wrote:
>> There are several problems with the code that rescuers bind itself to the pool'
>> cpumask
>> 1) It uses a way different from the normal workers to bind to the cpumask
>> So we can't maintain the normal/rescuer workers under the same framework.
>> 2) The the code of cpu-binding for rescuer is complicated
>> 3) If one or more cpuhotplugs happen while the rescuer processes the
>> scheduled works, the rescuer may not be correctly bound to the cpumask of
>> the pool. This is allowed behavior, but is not good. It will be better
>> if the cpumask of the rescuer is always kept coordination with the pool
>> across any cpuhotplugs.
>>
>> Using generic attach/detach routine will solve the above problems,
>> and result much more simple code.
>>
>> Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
>> static struct worker *alloc_worker(void)
>> {
>> struct worker *worker;
>> @@ -2343,8 +2279,9 @@ repeat:
>>
>> spin_unlock_irq(&wq_mayday_lock);
>>
>> - /* migrate to the target cpu if possible */
>> - worker_maybe_bind_and_lock(pool);
>> + worker_attach_to_pool(rescuer, pool);
>> +
>> + spin_lock_irq(&pool->lock);
>> rescuer->pool = pool;
>>
>> /*
>> @@ -2357,6 +2294,11 @@ repeat:
>> move_linked_works(work, scheduled, &n);
>>
>> process_scheduled_works(rescuer);
>> + spin_unlock_irq(&pool->lock);
>> +
>> + worker_detach_from_pool(rescuer, pool);
>> +
>> + spin_lock_irq(&pool->lock);
>
> Ah, right, this is how it's used. Yeah, it makes sense. In a long
> patchset, it usually helps to mention your intentions when structuring
> functions tho. When you're separating out detach_from_pool, just
> mention that the function will later be used to make rescuers use the
In patch3, the intention for separating out detach_from_pool() is
only for aync-destruction and it is used for worker_thread().
next prev parent reply other threads:[~2014-05-13 6:51 UTC|newest]
Thread overview: 78+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-04-27 4:08 [PATCH 00/10] workqueue: async worker destruction and pool-binding synchronization Lai Jiangshan
2014-04-27 4:08 ` [PATCH 01/10] workqueue: use manager lock only to protect worker_idr Lai Jiangshan
2014-05-05 13:01 ` Tejun Heo
2014-05-06 15:43 ` Lai Jiangshan
2014-04-27 4:08 ` [PATCH 02/10] workqueue: destroy_worker() should destroy idle workers only Lai Jiangshan
2014-05-05 13:13 ` Tejun Heo
2014-05-06 15:58 ` Lai Jiangshan
2014-04-27 4:08 ` [PATCH 03/10] workqueue: async worker destruction Lai Jiangshan
2014-05-05 14:31 ` Tejun Heo
2014-05-05 15:02 ` Tejun Heo
2014-05-06 16:27 ` Lai Jiangshan
2014-05-06 16:31 ` Tejun Heo
2014-05-07 7:30 ` Lai Jiangshan
2014-05-07 13:15 ` Tejun Heo
2014-04-27 4:08 ` [PATCH 04/10] workqueue: destroy worker directly in the idle timeout handler Lai Jiangshan
2014-05-05 14:36 ` Tejun Heo
2014-05-07 7:10 ` Lai Jiangshan
2014-05-07 13:12 ` Tejun Heo
2014-05-07 13:38 ` Lai Jiangshan
2014-05-07 13:41 ` Tejun Heo
2014-05-07 15:30 ` Lai Jiangshan
2014-05-07 15:37 ` Tejun Heo
2014-04-27 4:09 ` [PATCH 05/10] workqueue: separate iteration role from worker_idr Lai Jiangshan
2014-05-05 14:57 ` Tejun Heo
2014-04-27 4:09 ` [PATCH 06/10] workqueue: convert worker_idr to worker_ida Lai Jiangshan
2014-05-05 14:59 ` Tejun Heo
2014-05-06 16:33 ` Lai Jiangshan
2014-05-06 16:35 ` Tejun Heo
2014-05-06 16:38 ` Tejun Heo
2014-04-27 4:09 ` [PATCH 07/10] workqueue: narrow the protection range of manager_mutex Lai Jiangshan
2014-04-27 4:09 ` [PATCH 08/10] workqueue: rename manager_mutex to bind_mutex Lai Jiangshan
2014-04-27 4:09 ` [PATCH 09/10] workqueue: separate pool-binding code out from create_worker() Lai Jiangshan
2014-04-27 4:09 ` [PATCH 10/10] workqueue: use generic pool-bind/unbind routine for rescuers Lai Jiangshan
2014-05-05 14:54 ` Tejun Heo
2014-05-05 15:05 ` [PATCH 00/10] workqueue: async worker destruction and pool-binding synchronization Tejun Heo
2014-05-12 6:56 ` [PATCH 00/10 V2] workqueue: async worker destruction and worker attaching/detaching Lai Jiangshan
2014-05-12 6:56 ` [PATCH 01/10 V2] workqueue: use manager lock only to protect worker_idr Lai Jiangshan
2014-05-12 6:56 ` [PATCH 02/10 V2] workqueue: destroy_worker() should destroy idle workers only Lai Jiangshan
2014-05-12 21:08 ` Tejun Heo
2014-05-13 6:08 ` Lai Jiangshan
2014-05-12 6:56 ` [PATCH 03/10 V2] workqueue: async worker destruction Lai Jiangshan
2014-05-12 21:20 ` Tejun Heo
2014-05-13 6:32 ` Lai Jiangshan
2014-05-13 14:14 ` Tejun Heo
2014-05-20 9:54 ` Lai Jiangshan
2014-05-12 6:56 ` [PATCH 04/10 V2] workqueue: destroy worker directly in the idle timeout handler Lai Jiangshan
2014-05-12 21:26 ` Tejun Heo
2014-05-12 6:56 ` [PATCH 05/10 V2] workqueue: separate iteration role from worker_idr Lai Jiangshan
2014-05-12 21:35 ` Tejun Heo
2014-05-12 21:37 ` Tejun Heo
2014-05-12 6:56 ` [PATCH 06/10 V2] workqueue: convert worker_idr to worker_ida Lai Jiangshan
2014-05-12 21:40 ` Tejun Heo
2014-05-13 6:43 ` Lai Jiangshan
2014-05-13 14:17 ` Tejun Heo
2014-05-12 6:56 ` [PATCH 07/10 V2] workqueue: narrow the protection range of manager_mutex Lai Jiangshan
2014-05-12 6:56 ` [PATCH 08/10 V2] workqueue: rename manager_mutex to attach_mutex Lai Jiangshan
2014-05-12 22:01 ` Tejun Heo
2014-05-13 6:45 ` Lai Jiangshan
2014-05-13 14:19 ` Tejun Heo
2014-05-12 6:56 ` [PATCH 09/10 V2] workqueue: separate pool-attaching code out from create_worker() Lai Jiangshan
2014-05-12 6:56 ` [PATCH 10/10 V2] workqueue: use generic attach/detach routine for rescuers Lai Jiangshan
2014-05-12 22:05 ` Tejun Heo
2014-05-13 6:55 ` Lai Jiangshan [this message]
2014-05-20 9:46 ` [PATCH V3 00/10] workqueue: async worker destruction and worker attaching/detaching Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 01/10] workqueue: use manager lock only to protect worker_idr Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 02/10] workqueue: destroy_worker() should destroy idle workers only Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 03/10] workqueue: async worker destruction Lai Jiangshan
2014-05-20 14:22 ` Tejun Heo
2014-05-20 14:23 ` Tejun Heo
2014-05-20 14:24 ` Tejun Heo
2014-05-20 9:46 ` [PATCH V3 04/10] workqueue: destroy worker directly in the idle timeout handler Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 05/10] workqueue: separate iteration role from worker_idr Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 06/10] workqueue: convert worker_idr to worker_ida Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 07/10] workqueue: narrow the protection range of manager_mutex Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 08/10] workqueue: rename manager_mutex to attach_mutex Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 09/10] workqueue: separate pool-attaching code out from create_worker() Lai Jiangshan
2014-05-20 9:46 ` [PATCH V3 10/10] workqueue: use generic attach/detach routine for rescuers Lai Jiangshan
2014-05-20 15:00 ` [PATCH V3 00/10] workqueue: async worker destruction and worker attaching/detaching Tejun Heo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5371C1DC.8030102@cn.fujitsu.com \
--to=laijs@cn.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=tj@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox