lkml.org 
[lkml]   [2011]   [Jan]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 21/32] fs/aio: aio_wq isn't used in memory reclaim path
Hello,

On Tue, Jan 04, 2011 at 10:56:20AM -0500, Jeff Moyer wrote:
> > aio_wq isn't used during memory reclaim. Convert to alloc_workqueue()
> > without WQ_MEM_RECLAIM. It's possible to use system_wq but given that
> > the number of work items is determined from userland and the work item
> > may block, enforcing strict concurrency limit would be a good idea.
>
> I would think that just given that it may block would be enough to keep
> it off of the system workqueue.

Oh, workqueue now can handle parallel execution. Blocking on system
workqueue is no longer a problem. One of the main reasons for this
whole series.

> > @@ -85,7 +85,7 @@ static int __init aio_setup(void)
> > kiocb_cachep = KMEM_CACHE(kiocb, SLAB_HWCACHE_ALIGN|SLAB_PANIC);
> > kioctx_cachep = KMEM_CACHE(kioctx,SLAB_HWCACHE_ALIGN|SLAB_PANIC);
> >
> > - aio_wq = create_workqueue("aio");
> > + aio_wq = alloc_workqueue("aio", 0, 1); /* used to limit concurrency */
>
> OK, the only difference here is the removal of the WQ_MEM_RECLAIM flag,
> as you noted.

Yeap. Do you agree that the concurrency limit is necessary? If not,
we can just put everything onto system_wq.

> > @@ -569,7 +569,7 @@ static int __aio_put_req(struct kioctx *ctx, struct kiocb *req)
> > spin_lock(&fput_lock);
> > list_add(&req->ki_list, &fput_head);
> > spin_unlock(&fput_lock);
> > - queue_work(aio_wq, &fput_work);
> > + schedule_work(&fput_work);
>
> I'm not sure where this change fits into the patch description. Why did
> you do this?

Yeah, that's me being forgetful. Now that aio_wq is solely used to
throttle the max concurrency of aio work items, I thought it would be
better to push fput_work to system workqueue so that it doesn't
interact with aio work items. I'll update the patch description.

Thanks.

--
tejun


\
 
 \ /
  Last update: 2011-01-05 12:31    [W:0.717 / U:0.352 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site