[v2] loop: Add PF_LESS_THROTTLE to block/loop device thread.

Commit Message

NeilBrown April 6, 2017, 2:23 a.m. UTC

On Wed, Apr 05 2017, Michal Hocko wrote:

> On Wed 05-04-17 09:19:27, Michal Hocko wrote:
>> On Wed 05-04-17 14:33:50, NeilBrown wrote:
> [...]
>> > diff --git a/drivers/block/loop.c b/drivers/block/loop.c
>> > index 0ecb6461ed81..44b3506fd086 100644
>> > --- a/drivers/block/loop.c
>> > +++ b/drivers/block/loop.c
>> > @@ -852,6 +852,7 @@ static int loop_prepare_queue(struct loop_device *lo)
>> >  	if (IS_ERR(lo->worker_task))
>> >  		return -ENOMEM;
>> >  	set_user_nice(lo->worker_task, MIN_NICE);
>> > +	lo->worker_task->flags |= PF_LESS_THROTTLE;
>> >  	return 0;
>> 
>> As mentioned elsewhere, PF flags should be updated only on the current
>> task otherwise there is potential rmw race. Is this safe? The code runs
>> concurrently with the worker thread.
>
> I believe you need something like this instead
> ---
> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
> index f347285c67ec..07b2a909e4fb 100644
> --- a/drivers/block/loop.c
> +++ b/drivers/block/loop.c
> @@ -844,10 +844,16 @@ static void loop_unprepare_queue(struct loop_device *lo)
>  	kthread_stop(lo->worker_task);
>  }
>  
> +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;

Arg - of course.
How about we just split the kthread_create from the wake_up?

Thanks,
NeilBrown


From: NeilBrown <[e-mail]>
Subject: [PATCH] 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.

Signed-off-by: NeilBrown <[e-mail]>
---
 drivers/block/loop.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Comments

Michal Hocko April 6, 2017, 6:53 a.m. UTC | #1

On Thu 06-04-17 12:23:51, NeilBrown wrote:
[...]
> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
> index 0ecb6461ed81..95679d988725 100644
> --- a/drivers/block/loop.c
> +++ b/drivers/block/loop.c
> @@ -847,10 +847,12 @@ static void loop_unprepare_queue(struct loop_device *lo)
>  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_create(kthread_worker_fn,
>  			&lo->worker, "loop%d", lo->lo_number);
>  	if (IS_ERR(lo->worker_task))
>  		return -ENOMEM;
> +	lo->worker_task->flags |= PF_LESS_THROTTLE;
> +	wake_up_process(lo->worker_task);
>  	set_user_nice(lo->worker_task, MIN_NICE);
>  	return 0;

This should work for the current implementation because kthread_create
will return only after the full initialization has been done. No idea
whether we can rely on that in future. I also think it would be cleaner
to set the flag on current and keep the current semantic that only
current changes its flags.

So while I do not have a strong opinion on this I think defining loop
specific thread function which set PF_LESS_THROTTLE as the first thing
is more elegant and less error prone longerm. A short comment explaining
why we use the flag there would be also preferred.

I will leave the decision to you.

Thanks.

Patch

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 0ecb6461ed81..95679d988725 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -847,10 +847,12 @@  static void loop_unprepare_queue(struct loop_device *lo)
 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_create(kthread_worker_fn,
 			&lo->worker, "loop%d", lo->lo_number);
 	if (IS_ERR(lo->worker_task))
 		return -ENOMEM;
+	lo->worker_task->flags |= PF_LESS_THROTTLE;
+	wake_up_process(lo->worker_task);
 	set_user_nice(lo->worker_task, MIN_NICE);
 	return 0;
 }