From mboxrd@z Thu Jan  1 00:00:00 1970
From: Christoph Hellwig <hch@lst.de>
Date: Thu, 22 Nov 2018 11:40:49 +0100
Subject: [Cluster-devel] [PATCH V11 14/19] block: handle non-cluster bio
 out of blk_bio_segment_split
In-Reply-To: <20181122102616.GC27273@ming.t460p>
References: <20181121032327.8434-1-ming.lei@redhat.com>
	<20181121032327.8434-15-ming.lei@redhat.com>
	<20181121143355.GB2594@lst.de> <20181121153726.GC19111@ming.t460p>
	<20181121174621.GA6961@lst.de> <20181122093259.GA27007@ming.t460p>
	<20181122100427.GA28871@lst.de> <20181122102616.GC27273@ming.t460p>
Message-ID: <20181122104049.GA29654@lst.de>
List-Id: <cluster-devel.redhat.com>
To: cluster-devel.redhat.com
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

On Thu, Nov 22, 2018 at 06:26:17PM +0800, Ming Lei wrote:
> Suppose one bio includes (pg0, 0, 512) and (pg1, 512, 512):
> 
> The split is introduced by the following code in blk_bio_segment_split():
> 
>       if (bvprvp && bvec_gap_to_prev(q, bvprvp, bv.bv_offset))
> 		  	goto split;
> 
> Without this patch, for non-cluster, the two bvecs are just in different
> segment, but still handled by one same bio. Now you convert into two bios.

Oh, true - we create new bios instead of new segments.  So I guess we
need to do something special here if we don't want to pay that overhead.

Or just look into killing the cluster setting..