Re: [PATCH 10/16] nvmet: add metadata characteristics for a namespace

[James Smart <james.smart@broadcom.com>]

On 5/4/2020 8:57 AM, Max Gurtovoy wrote:
> From: Israel Rukshin <israelr@mellanox.com>
>
> Fill those namespace fields from the block device format for adding
> metadata (T10-PI) over fabric support with block devices.
>
> Signed-off-by: Israel Rukshin <israelr@mellanox.com>
> Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> ---
>   drivers/nvme/target/io-cmd-bdev.c | 22 ++++++++++++++++++++++
>   drivers/nvme/target/nvmet.h       |  3 +++
>   2 files changed, 25 insertions(+)
>
> diff --git a/drivers/nvme/target/io-cmd-bdev.c b/drivers/nvme/target/io-cmd-bdev.c
> index 0427e04..45dc644 100644
> --- a/drivers/nvme/target/io-cmd-bdev.c
> +++ b/drivers/nvme/target/io-cmd-bdev.c
> @@ -47,6 +47,22 @@ void nvmet_bdev_set_limits(struct block_device *bdev, struct nvme_id_ns *id)
>   	id->nows = to0based(ql->io_opt / ql->logical_block_size);
>   }
>   
> +static void nvmet_bdev_ns_enable_integrity(struct nvmet_ns *ns)
> +{
> +	struct blk_integrity *bi = bdev_get_integrity(ns->bdev);
> +
> +	if (bi) {
> +		ns->metadata_size = bi->tuple_size;
> +		if (bi->profile == &t10_pi_type1_crc)
> +			ns->pi_type = NVME_NS_DPS_PI_TYPE1;
> +		else if (bi->profile == &t10_pi_type3_crc)
> +			ns->pi_type = NVME_NS_DPS_PI_TYPE3;
> +		else
> +			/* Unsupported metadata type */
> +			ns->metadata_size = 0;
> +	}
> +}
> +
>   int nvmet_bdev_ns_enable(struct nvmet_ns *ns)
>   {
>   	int ret;
> @@ -64,6 +80,12 @@ int nvmet_bdev_ns_enable(struct nvmet_ns *ns)
>   	}
>   	ns->size = i_size_read(ns->bdev->bd_inode);
>   	ns->blksize_shift = blksize_bits(bdev_logical_block_size(ns->bdev));
> +
> +	ns->pi_type = 0;
> +	ns->metadata_size = 0;
> +	if (IS_ENABLED(CONFIG_BLK_DEV_INTEGRITY))
> +		nvmet_bdev_ns_enable_integrity(ns);
> +

Ok. In the future we will want to consider a case of a transport 
stripping/inserting PI data, even if not implemented by the controller's 
bdev. Although it only checks fabric transmission, it's better than 0 
checking.

I assume, all bdevs that support an integrity profile will report a 
blocksize that does not contain the integrity data ?

And we're going to fail as badly now as before if the bdev didn't have a 
power-of-2 blocksize, or supported metadata (a blk-integrity buffer) and 
metadata_size isn't set ?

-- james


_______________________________________________
linux-nvme mailing list
linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme