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

Groups > linux.kernel > #1200550 > unrolled thread

[PATCH] block: fix update first/last mergeable segment sizes

Back to article view | Back to linux.kernel

Contents

  [PATCH] block: fix update first/last mergeable segment sizes Wanpeng Li <wanpeng.li@hotmail.com> - 2015-08-05 11:00 +0200
    Re: [PATCH] block: fix update first/last mergeable segment sizes Jeff Moyer <jmoyer@redhat.com> - 2015-08-07 18:00 +0200

#1200550 — [PATCH] block: fix update first/last mergeable segment sizes

To keep track of the max segment size, the sizes of the first and last 
mergeable segments in the bio are accounted, prev->bio->bi_seg_front_size
and next->biotail->bi_seg_back_size will be updated if blk_phys_contig_segment 
is true, however, there are still some scenarios such like nonconsistent
blk integrity which leads to merge fail, in addtion, futher merge fail is 
incurred due to bio->bi_seg_back_size + next->bi_seg_front_size > 
queue_max_segment_size(q) check in function blk_phys_contig_segment(). This 
patch fix it by updating first/last mergeable segments size iff all the 
conditions of merging are meet.

Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
---
 block/blk-merge.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/block/blk-merge.c b/block/blk-merge.c
index 30a0d9f..7e7bda4 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -367,6 +367,7 @@ static int req_gap_to_prev(struct request *req, struct request *next)
 static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
 				struct request *next)
 {
+	bool update_segment_size = false;
 	int total_phys_segments;
 	unsigned int seg_size =
 		req->biotail->bi_seg_back_size + next->bio->bi_seg_front_size;
@@ -391,10 +392,7 @@ static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
 
 	total_phys_segments = req->nr_phys_segments + next->nr_phys_segments;
 	if (blk_phys_contig_segment(q, req->biotail, next->bio)) {
-		if (req->nr_phys_segments == 1)
-			req->bio->bi_seg_front_size = seg_size;
-		if (next->nr_phys_segments == 1)
-			next->biotail->bi_seg_back_size = seg_size;
+		update_segment_size = true;
 		total_phys_segments--;
 	}
 
@@ -405,6 +403,12 @@ static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
 		return 0;
 
 	/* Merge is OK... */
+	if (update_segment_size) {
+		if (req->nr_phys_segments == 1)
+			req->bio->bi_seg_front_size = seg_size;
+		if (next->nr_phys_segments == 1)
+			next->biotail->bi_seg_back_size = seg_size;
+	}
 	req->nr_phys_segments = total_phys_segments;
 	return 1;
 }
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1202845

Wanpeng Li <wanpeng.li@hotmail.com> writes:

> To keep track of the max segment size, the sizes of the first and last 
> mergeable segments in the bio are accounted, prev->bio->bi_seg_front_size
> and next->biotail->bi_seg_back_size will be updated if blk_phys_contig_segment 
> is true, however, there are still some scenarios such like nonconsistent
> blk integrity which leads to merge fail, in addtion, futher merge fail is 
> incurred due to bio->bi_seg_back_size + next->bi_seg_front_size > 
> queue_max_segment_size(q) check in function blk_phys_contig_segment(). This 
> patch fix it by updating first/last mergeable segments size iff all the 
> conditions of merging are meet.

Does it matter?  Did you see any incorrect behaviour as a result of the
current code?

Cheers,
Jeff

>
> Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
> ---
>  block/blk-merge.c | 12 ++++++++----
>  1 file changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/block/blk-merge.c b/block/blk-merge.c
> index 30a0d9f..7e7bda4 100644
> --- a/block/blk-merge.c
> +++ b/block/blk-merge.c
> @@ -367,6 +367,7 @@ static int req_gap_to_prev(struct request *req, struct request *next)
>  static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
>  				struct request *next)
>  {
> +	bool update_segment_size = false;
>  	int total_phys_segments;
>  	unsigned int seg_size =
>  		req->biotail->bi_seg_back_size + next->bio->bi_seg_front_size;
> @@ -391,10 +392,7 @@ static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
>  
>  	total_phys_segments = req->nr_phys_segments + next->nr_phys_segments;
>  	if (blk_phys_contig_segment(q, req->biotail, next->bio)) {
> -		if (req->nr_phys_segments == 1)
> -			req->bio->bi_seg_front_size = seg_size;
> -		if (next->nr_phys_segments == 1)
> -			next->biotail->bi_seg_back_size = seg_size;
> +		update_segment_size = true;
>  		total_phys_segments--;
>  	}
>  
> @@ -405,6 +403,12 @@ static int ll_merge_requests_fn(struct request_queue *q, struct request *req,
>  		return 0;
>  
>  	/* Merge is OK... */
> +	if (update_segment_size) {
> +		if (req->nr_phys_segments == 1)
> +			req->bio->bi_seg_front_size = seg_size;
> +		if (next->nr_phys_segments == 1)
> +			next->biotail->bi_seg_back_size = seg_size;
> +	}
>  	req->nr_phys_segments = total_phys_segments;
>  	return 1;
>  }
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web