Re: [PATCH] nvme: fix Write Zeroes limitations

[Himanshu Madhani <himanshu.madhani@oracle.com>]

> On Mar 15, 2021, at 4:34 AM, Christoph Hellwig <hch@lst.de> wrote:
> 
> We voluntarily limit the Write Zeroes sizes to the MDTS value provided by
> the hardware, but currently get the units wrong, so fix that.
> 
> Fixes: 6e02318eaea5 ("nvme: add support for the Write Zeroes command")
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
> drivers/nvme/host/core.c | 36 ++++++++++++------------------------
> 1 file changed, 12 insertions(+), 24 deletions(-)
> 
> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
> index 82ad5eef9d0c30..1bd18d7f53b316 100644
> --- a/drivers/nvme/host/core.c
> +++ b/drivers/nvme/host/core.c
> @@ -1964,30 +1964,18 @@ static void nvme_config_discard(struct gendisk *disk, struct nvme_ns *ns)
> 		blk_queue_max_write_zeroes_sectors(queue, UINT_MAX);
> }
> 
> -static void nvme_config_write_zeroes(struct gendisk *disk, struct nvme_ns *ns)
> +/*
> + * Even though NVMe spec explicitly states that MDTS is not applicable to the
> + * write-zeroes, we are cautious and limit the size to the controllers
> + * max_hw_sectors value, which is based on the MDTS field and possibly other
> + * limiting factors.
> + */
> +static void nvme_config_write_zeroes(struct request_queue *q,
> +		struct nvme_ctrl *ctrl)
> {
> -	u64 max_blocks;
> -
> -	if (!(ns->ctrl->oncs & NVME_CTRL_ONCS_WRITE_ZEROES) ||
> -	    (ns->ctrl->quirks & NVME_QUIRK_DISABLE_WRITE_ZEROES))
> -		return;
> -	/*
> -	 * Even though NVMe spec explicitly states that MDTS is not
> -	 * applicable to the write-zeroes:- "The restriction does not apply to
> -	 * commands that do not transfer data between the host and the
> -	 * controller (e.g., Write Uncorrectable ro Write Zeroes command).".
> -	 * In order to be more cautious use controller's max_hw_sectors value
> -	 * to configure the maximum sectors for the write-zeroes which is
> -	 * configured based on the controller's MDTS field in the
> -	 * nvme_init_identify() if available.
> -	 */
> -	if (ns->ctrl->max_hw_sectors == UINT_MAX)
> -		max_blocks = (u64)USHRT_MAX + 1;
> -	else
> -		max_blocks = ns->ctrl->max_hw_sectors + 1;
> -
> -	blk_queue_max_write_zeroes_sectors(disk->queue,
> -					   nvme_lba_to_sect(ns, max_blocks));
> +	if ((ctrl->oncs & NVME_CTRL_ONCS_WRITE_ZEROES) &&
> +	    !(ctrl->quirks & NVME_QUIRK_DISABLE_WRITE_ZEROES))
> +		blk_queue_max_write_zeroes_sectors(q, ctrl->max_hw_sectors);
> }
> 
> static bool nvme_ns_ids_valid(struct nvme_ns_ids *ids)
> @@ -2159,7 +2147,7 @@ static void nvme_update_disk_info(struct gendisk *disk,
> 	set_capacity_and_notify(disk, capacity);
> 
> 	nvme_config_discard(disk, ns);
> -	nvme_config_write_zeroes(disk, ns);
> +	nvme_config_write_zeroes(disk->queue, ns->ctrl);
> 
> 	set_disk_ro(disk, (id->nsattr & NVME_NS_ATTR_RO) ||
> 		test_bit(NVME_NS_FORCE_RO, &ns->flags));
> -- 
> 2.30.1
> 
> 
> _______________________________________________
> Linux-nvme mailing list
> Linux-nvme@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-nvme

Looks Good. 

Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>

--
Himanshu Madhani	 Oracle Linux Engineering


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