[PATCH] nvme: fix handling single range discard request

[Ming Lei]

On Sat, Mar 04, 2023 at 09:00:28AM +0100, Hannes Reinecke wrote:
> On 3/4/23 00:13, Ming Lei wrote:
> > When investigating one customer report on warning in nvme_setup_discard,
> > we observed the controller(nvme/tcp) actually exposes
> > queue_max_discard_segments(req->q) == 1.
> > 
> > Obviously the current code can't handle this situation, since contiguity
> > merge like normal RW request is taken.
> > 
> > Fix the issue by building range from request sector/nr_sectors directly.
> > 
> > Fixes: b35ba01ea697 ("nvme: support ranged discard requests")
> > Signed-off-by: Ming Lei <ming.lei at redhat.com>
> > ---
> >   drivers/nvme/host/core.c | 28 +++++++++++++++++++---------
> >   1 file changed, 19 insertions(+), 9 deletions(-)
> > 
> > diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
> > index c2730b116dc6..d4be525f8100 100644
> > --- a/drivers/nvme/host/core.c
> > +++ b/drivers/nvme/host/core.c
> > @@ -781,16 +781,26 @@ static blk_status_t nvme_setup_discard(struct nvme_ns *ns, struct request *req,
> >   		range = page_address(ns->ctrl->discard_page);
> >   	}
> > -	__rq_for_each_bio(bio, req) {
> > -		u64 slba = nvme_sect_to_lba(ns, bio->bi_iter.bi_sector);
> > -		u32 nlb = bio->bi_iter.bi_size >> ns->lba_shift;
> > -
> > -		if (n < segments) {
> > -			range[n].cattr = cpu_to_le32(0);
> > -			range[n].nlb = cpu_to_le32(nlb);
> > -			range[n].slba = cpu_to_le64(slba);
> > +	if (queue_max_discard_segments(req->q) == 1) {
> > +		u64 slba = nvme_sect_to_lba(ns, blk_rq_pos(req));
> > +		u32 nlb = blk_rq_sectors(req) >> (ns->lba_shift - 9);
> > +
> > +		range[0].cattr = cpu_to_le32(0);
> > +		range[0].nlb = cpu_to_le32(nlb);
> > +		range[0].slba = cpu_to_le64(slba);
> > +		n = 1;
> > +	} else { > +		__rq_for_each_bio(bio, req) {
> > +			u64 slba = nvme_sect_to_lba(ns, bio->bi_iter.bi_sector);
> > +			u32 nlb = bio->bi_iter.bi_size >> ns->lba_shift;
> > +
> > +			if (n < segments) {
> > +				range[n].cattr = cpu_to_le32(0);
> > +				range[n].nlb = cpu_to_le32(nlb);
> > +				range[n].slba = cpu_to_le64(slba);
> > +			}
> > +			n++;
> >   		}
> > -		n++;
> >   	}
> >   	if (WARN_ON_ONCE(n != segments)) {
> Now _that_ is odd.
> Looks like 'req' is not formatted according to the 'max_discard_sectors'
> setting.
> But if that's the case, then this 'fix' would fail whenever
> 'max_discard_sectors' < 'max_hw_sectors', right?

No, it isn't the case.

> Shouldn't we rather modify the merge algorithm to check for
> max_discard_sectors for DISCARD requests, such that we never _have_
> mis-matched requests and this patch would be pointless?

But it is related with discard merge.

If queue_max_discard_segments() is 1, block layer merges discard
request/bios just like normal RW IO.

However, if queue_max_discard_segments() is > 1, block layer simply
'merges' all bios into one request, no matter if the LBA is adjacent
or not, and treat each bio as one discard segment, that is called
multi range discard too.

That is the reason why we have to treat queue_max_discard_segments() == 1
specially, and you can see same handling in virtio_blk.


Thanks,
Ming