Re: [PATCH] block: recalculate segment count for multi-segment discard requests correctly

[Laurence Oberman]

On Wed, 2021-02-03 at 08:50 -0500, Laurence Oberman wrote:
> On Wed, 2021-02-03 at 03:15 +0000, Chaitanya Kulkarni wrote:
> > On 2/2/21 18:39, Ming Lei wrote:
> > > +		/* fall through */
> > >  	case REQ_OP_WRITE_ZEROES:
> > > -		return 0;
> > 
> > I don't think returning 1 for write-zeroes is right,
> > did you test this patch with write-zeores enabled controller with
> > the right fs that triggers this behavior ?
> > 
> > 
> 
> I tested the first iteration of the patch fully mounting an XFS file
> system with -o discard and creating and deleting files.
> That was our specific RHEL8 failure we were handling here with
> David's
> first submission.
> 
> I can test his most recent, I have not done that yet.
> Again, please follow up with exactly what you want based against
> David's patch and I can test that.
> 
> Regards
> Laurence 

So if I understand what it is you were wanting I will test this now.

diff --git a/block/blk-merge.c b/block/blk-merge.c
index 808768f6b174..a9bd958c07c4 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -383,6 +383,13 @@ unsigned int blk_recalc_rq_segments(struct request
*rq)
        switch (bio_op(rq->bio)) {
        case REQ_OP_DISCARD:
        case REQ_OP_SECURE_ERASE:
+                if (queue_max_discard_segments(rq->q) > 1) {
+                        struct bio *bio = rq->bio;

+                        for_each_bio(bio)
+                                nr_phys_segs++;
+                        return nr_phys_segs;
+                }
+                /* fall through */
        case REQ_OP_WRITE_ZEROES:
                return 0;
        case REQ_OP_WRITE_SAME: