Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1667421 > unrolled thread

[PATCH 0/2] Two fixes for loop devices

Back to article view | Back to linux.kernel

Contents

  [PATCH 0/2] Two fixes for loop devices NeilBrown <neilb@suse.com> - 2017-06-16 07:10 +0200
    [PATCH 2/2] loop: Add PF_LESS_THROTTLE to block/loop device thread. NeilBrown <neilb@suse.com> - 2017-06-16 07:10 +0200
      Re: [PATCH 2/2] loop: Add PF_LESS_THROTTLE to block/loop device  thread. Christoph Hellwig <hch@infradead.org> - 2017-06-16 09:40 +0200
    Re: [PATCH 0/2] Two fixes for loop devices NeilBrown <neilb@suse.com> - 2017-06-18 06:40 +0200
      Re: [PATCH 0/2] Two fixes for loop devices Jens Axboe <axboe@kernel.dk> - 2017-06-18 17:10 +0200

#1667421 — [PATCH 0/2] Two fixes for loop devices

Hi Jens,
 one of these is a resend of a patch I sent a while back.
 The other is new - loop closes files differently from close()
 and in a way that can confuse NFS.

Thanks,
NeilBrown

---

NeilBrown (2):
      loop: use filp_close() rather than fput()
      loop: Add PF_LESS_THROTTLE to block/loop device thread.


 drivers/block/loop.c |   14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

--
Signature

[toc] | [next] | [standalone]

#1667423 — [PATCH 2/2] loop: Add PF_LESS_THROTTLE to block/loop device thread.

When a filesystem is mounted from a loop device, writes are
throttled by balance_dirty_pages() twice: once when writing
to the filesystem and once when the loop_handle_cmd() writes
to the backing file.  This double-throttling can trigger
positive feedback loops that create significant delays.  The
throttling at the lower level is seen by the upper level as
a slow device, so it throttles extra hard.

The PF_LESS_THROTTLE flag was created to handle exactly this
circumstance, though with an NFS filesystem mounted from a
local NFS server.  It reduces the throttling on the lower
layer so that it can proceed largely unthrottled.

To demonstrate this, create a filesystem on a loop device
and write (e.g. with dd) several large files which combine
to consume significantly more than the limit set by
/proc/sys/vm/dirty_ratio or dirty_bytes.  Measure the total
time taken.

When I do this directly on a device (no loop device) the
total time for several runs (mkfs, mount, write 200 files,
umount) is fairly stable: 28-35 seconds.
When I do this over a loop device the times are much worse
and less stable.  52-460 seconds.  Half below 100seconds,
half above.
When I apply this patch, the times become stable again,
though not as fast as the no-loop-back case: 53-72 seconds.

There may be room for further improvement as the total overhead still
seems too high, but this is a big improvement.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <tom.leiming@gmail.com>
Suggested-by: Michal Hocko <mhocko@suse.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: NeilBrown <neilb@suse.com>
---
 drivers/block/loop.c |    8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 9c457ca6c55e..6ed7c4506951 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -843,10 +843,16 @@ static void loop_unprepare_queue(struct loop_device *lo)
 	kthread_stop(lo->worker_task);
 }
 
+static int loop_kthread_worker_fn(void *worker_ptr)
+{
+	current->flags |= PF_LESS_THROTTLE;
+	return kthread_worker_fn(worker_ptr);
+}
+
 static int loop_prepare_queue(struct loop_device *lo)
 {
 	kthread_init_worker(&lo->worker);
-	lo->worker_task = kthread_run(kthread_worker_fn,
+	lo->worker_task = kthread_run(loop_kthread_worker_fn,
 			&lo->worker, "loop%d", lo->lo_number);
 	if (IS_ERR(lo->worker_task))
 		return -ENOMEM;

[toc] | [prev] | [next] | [standalone]

#1667499 — Re: [PATCH 2/2] loop: Add PF_LESS_THROTTLE to block/loop device thread.

why isn't loop using kthread_create_worker()?  Why isn't the less
throttle a flag to kthread_create_worker()?  I hate all this open
coding..

[toc] | [prev] | [next] | [standalone]

#1668528

[Multipart message — attachments visible in raw view] — view raw

On Fri, Jun 16 2017, Jens Axboe wrote:

> On 06/15/2017 11:02 PM, NeilBrown wrote:
>> Hi Jens,
>>  one of these is a resend of a patch I sent a while back.
>>  The other is new - loop closes files differently from close()
>>  and in a way that can confuse NFS.
>
> Are you wanting to get these into 4.12, or defer to 4.13?

I'm happy either way.  I just want them to land eventually, so I can
tick them off my list.

The conversation with Al might result in a different fix for the
flip_close() problem, but I think that it is still more correct to call
filp_close(), because what loop is doing is a lot like closing a file
that it has been writing to.

Thanks,
NeilBrown

[toc] | [prev] | [next] | [standalone]

#1668627

On 06/17/2017 10:33 PM, NeilBrown wrote:
> On Fri, Jun 16 2017, Jens Axboe wrote:
> 
>> On 06/15/2017 11:02 PM, NeilBrown wrote:
>>> Hi Jens,
>>>  one of these is a resend of a patch I sent a while back.
>>>  The other is new - loop closes files differently from close()
>>>  and in a way that can confuse NFS.
>>
>> Are you wanting to get these into 4.12, or defer to 4.13?
> 
> I'm happy either way.  I just want them to land eventually, so I can
> tick them off my list.
> 
> The conversation with Al might result in a different fix for the
> flip_close() problem, but I think that it is still more correct to call
> filp_close(), because what loop is doing is a lot like closing a file
> that it has been writing to.

I'll toss 2/2 in for now for 4.13, please resend the other once the issues
have been ironed out.

-- 
Jens Axboe

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web