[PATCHv2 2/2] nvme: remove virtual boundary for sgl capable devices

[Christoph Hellwig]

> +	lim->max_segment_size = UINT_MAX;
> +	if (!nvme_ctrl_sgl_supported(ctrl) || admin ||
> +	    ctrl->ops->flags & NVME_F_FABRICS)
> +		lim->virt_boundary_mask = NVME_CTRL_PAGE_SIZE - 1;
> +	else
> +		lim->virt_boundary_mask = 0;

We start out with a virt_boundary_mask by default, so the else branch
here can go away.  This would also benefit from a comment explaining that
we only need the virt_boundary_mask for PRPs, and that as soon as SGLs
are supported for a given queue we don't set it because we don't want to
pay for the overhead it generates.

>  	if (nvmeq->qid && nvme_ctrl_sgl_supported(&dev->ctrl)) {
> -		if (nvme_req(req)->flags & NVME_REQ_USERCMD)
> -			return SGL_FORCED;
> -		if (req->nr_integrity_segments > 1)
> +		if (blk_rq_page_gaps(req) & (NVME_CTRL_PAGE_SIZE - 1) ||
> +		    nvme_req(req)->flags & NVME_REQ_USERCMD ||
> +		    req->nr_integrity_segments > 1)

And this would also really benefit from a comment explaining the high
level rationale behind the checks.