[PATCH 3/5] nvme: refactor namespace probing

[Joel Granados]

On Tue, Jul 19, 2022 at 12:38:45PM +0530, Kanchan Joshi wrote:
> On Mon, Jul 18, 2022 at 07:25:01AM +0200, Christoph Hellwig wrote:
> 
> > -static int nvme_update_ns_info(struct nvme_ns *ns, struct nvme_id_ns *id)
> > +static int nvme_update_ns_info_block(struct nvme_ns *ns,
> > +		struct nvme_ns_info *info)
> > {
> > -	unsigned lbaf = nvme_lbaf_index(id->flbas);
> > +	struct nvme_id_ns *id;
> > +	unsigned lbaf;
> > 	int ret;
> > 
> > +	ret = nvme_identify_ns(ns->ctrl, info->nsid, &id);
> > +	if (ret)
> > +		return ret;
> > +
> > 	blk_mq_freeze_queue(ns->disk->queue);
> > +	lbaf = nvme_lbaf_index(id->flbas);
> > 	ns->lba_shift = id->lbaf[lbaf].ds;
> > 	nvme_set_queue_limits(ns->ctrl, ns->queue);
> > 
> > @@ -1967,6 +1995,7 @@ static int nvme_update_ns_info(struct nvme_ns *ns, struct nvme_id_ns *id)
> > 		disk_update_readahead(ns->head->disk);
> > 		blk_mq_unfreeze_queue(ns->head->disk->queue);
> > 	}
> > +	kfree(id);
> > 	return 0;
> > 
> > out_unfreeze:
> > @@ -1980,9 +2009,30 @@ static int nvme_update_ns_info(struct nvme_ns *ns, struct nvme_id_ns *id)
> > 		ret = 0;
> > 	}
> > 	blk_mq_unfreeze_queue(ns->disk->queue);
> > +	kfree(id);
> > 	return ret;
> 
> Seems this new function (nvme_update_ns_info_block) requires kfree at one more place?
> This one:
> 
> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
> index bbdf0f95331f..5a917211e377 100644
> --- a/drivers/nvme/host/core.c
> +++ b/drivers/nvme/host/core.c
> @@ -2011,8 +2011,10 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,
> 
>        if (blk_queue_is_zoned(ns->queue)) {
>                ret = nvme_revalidate_zones(ns);
> -               if (ret && !nvme_first_scan(ns->disk))
> +               if (ret && !nvme_first_scan(ns->disk)) {
> +                       kfree(id);
>                        return ret;
> +               }
>        }
> 	

Should we also take care to hide the device?
As I read it the call to nvme_revalidate_zones can return a -ENODEV
(through blk_revalidate_disk_zones) which is the original condition to
hide the block device.

> With that-
> Reviewed-by: Kanchan Joshi <joshi.k at samsung.com>


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-nvme/attachments/20220719/ce9b01db/attachment.sig>