Re: [patch 1/2]block: avoid unnecessary plug list flush

[Namhyung Kim <namhyung@gmail.com>]

Shaohua Li <shaohua.li@intel.com> writes:

> get_request_wait() could sleep and flush the plug list. If the list
> is already flushed, don't flush again.
>
> Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Reviewed-by: Namhyung Kim <namhyung@gmail.com>

Good catch! If get_request_wait() updated request_count after sleeping
we could live with the original code, but your change makes more sense IMHO.

Thanks.


>
> ---
>  block/blk-core.c |   16 +++++++++-------
>  1 file changed, 9 insertions(+), 7 deletions(-)
>
> Index: linux/block/blk-core.c
> ===================================================================
> --- linux.orig/block/blk-core.c	2011-09-28 11:05:06.000000000 +0800
> +++ linux/block/blk-core.c	2011-09-28 11:05:22.000000000 +0800
> @@ -1299,15 +1299,17 @@ get_rq:
>  		 */
>  		if (list_empty(&plug->list))
>  			trace_block_plug(q);
> -		else if (!plug->should_sort) {
> -			struct request *__rq;
> +		else {
> +			if (!plug->should_sort) {
> +				struct request *__rq;
>  
> -			__rq = list_entry_rq(plug->list.prev);
> -			if (__rq->q != q)
> -				plug->should_sort = 1;
> +				__rq = list_entry_rq(plug->list.prev);
> +				if (__rq->q != q)
> +					plug->should_sort = 1;
> +			}
> +			if (request_count >= BLK_MAX_REQUEST_COUNT)
> +				blk_flush_plug_list(plug, false);
>  		}
> -		if (request_count >= BLK_MAX_REQUEST_COUNT)
> -			blk_flush_plug_list(plug, false);
>  		list_add_tail(&req->queuelist, &plug->list);
>  		drive_stat_acct(req, 1);
>  	} else {