Re: [PATCH] blk: clean up plug

[Jeff Moyer <jmoyer@redhat.com>]

Shaohua Li <shli@fb.com> writes:

> Current code looks like inner plug gets flushed with a
> blk_finish_plug(). Actually it's a nop. All requests/callbacks are added
> to current->plug, while only outmost plug is assigned to current->plug.
> So inner plug always has empty request/callback list, which makes
> blk_flush_plug_list() a nop. This tries to make the code more clear.
>
> Signed-off-by: Shaohua Li <shli@fb.com>

Hi, Shaohua,

I agree that this looks like a clean-up with no behavioral change, and
it looks good to me.  However,it does make me scratch my head about the
numbers I was seeing.  Here's the table from that other email thread[1]:

device|  vanilla   |    patch1      |   dio-noplug  |  noflush-nested
------+------------+----------------+---------------+-----------------
rssda | 701,684    |    1,168,527   |   1,342,177   |   1,297,612
      |     100%   |         +66%   |        +91%   |        +85%
vdb0  | 358,264    |      902,913   |     906,850   |     922,327
      |     100%   |        +152%   |       +153%   |       +157%

Patch1 refers to the first patch in this series, which fixes the merge
logic for single-queue blk-mq devices.  Each column after that includes
that first patch.  In dio-noplug, I removed the blk_plug from the
direct-io code path (so there is no nesting at all).  This is a control,
since it is what I expect the outcome of the noflush-nested column to
actually be.  Then, the noflush-nested column leaves the blk_plug in
place in the dio code, but includes the patch that prevents nested
blk_plug's from being flushed.  All numbers are the average of 5 runs.
With the exception of the vanilla run on rssda (the first run was
faster, causing the average to go up), the standard deviation is very
small.

For the dio-noplug column, if the inner plug really was a noop, then why
would we see any change in performance?  Like I said, I agree with your
reading of the code and the patch.  Color me confused.  I'll poke at it
more next week.  For now, I think your patch is fine.

Reviewed-by: Jeff Moyer <jmoyer@redhat.com>

Also, Jens or Shaohua or anyone, please review my blk-mq plug fix (patch
1/2 of aforementioned thread).  ;)

-Jeff

[1] https://lkml.org/lkml/2015/4/16/366

> ---
>  block/blk-core.c | 24 ++++++++++++------------
>  1 file changed, 12 insertions(+), 12 deletions(-)
>
> diff --git a/block/blk-core.c b/block/blk-core.c
> index 794c3e7..d3161f3 100644
> --- a/block/blk-core.c
> +++ b/block/blk-core.c
> @@ -3018,21 +3018,20 @@ void blk_start_plug(struct blk_plug *plug)
>  {
>  	struct task_struct *tsk = current;
>  
> +	/*
> +	 * If this is a nested plug, don't actually assign it.
> +	 */
> +	if (tsk->plug)
> +		return;
> +
>  	INIT_LIST_HEAD(&plug->list);
>  	INIT_LIST_HEAD(&plug->mq_list);
>  	INIT_LIST_HEAD(&plug->cb_list);
> -
>  	/*
> -	 * If this is a nested plug, don't actually assign it. It will be
> -	 * flushed on its own.
> +	 * Store ordering should not be needed here, since a potential
> +	 * preempt will imply a full memory barrier
>  	 */
> -	if (!tsk->plug) {
> -		/*
> -		 * Store ordering should not be needed here, since a potential
> -		 * preempt will imply a full memory barrier
> -		 */
> -		tsk->plug = plug;
> -	}
> +	tsk->plug = plug;
>  }
>  EXPORT_SYMBOL(blk_start_plug);
>  
> @@ -3179,10 +3178,11 @@ void blk_flush_plug_list(struct blk_plug *plug, bool from_schedule)
>  
>  void blk_finish_plug(struct blk_plug *plug)
>  {
> +	if (plug != current->plug)
> +		return;
>  	blk_flush_plug_list(plug, false);
>  
> -	if (plug == current->plug)
> -		current->plug = NULL;
> +	current->plug = NULL;
>  }
>  EXPORT_SYMBOL(blk_finish_plug);