Re: [PATCH V3 14/19] virtio_ring: determine descriptor flags at one time

[Michael S. Tsirkin]

On Mon, Jun 16, 2025 at 04:25:12PM +0800, Jason Wang wrote:
> Let's determine the last descriptor by counting the number of sg. This
> would be consistent with packed virtqueue implementation and ease the
> future in-order implementation.
> 
> Reviewed-by: Xuan Zhuo <xuanzhuo@xxxxxxxxxxxxxxxxx>
> Signed-off-by: Jason Wang <jasowang@xxxxxxxxxx>
> ---
>  drivers/virtio/virtio_ring.c | 25 +++++++++++++------------
>  1 file changed, 13 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
> index af32d1a1a1db..d5e4d4cd2487 100644
> --- a/drivers/virtio/virtio_ring.c
> +++ b/drivers/virtio/virtio_ring.c
> @@ -570,7 +570,7 @@ static inline int virtqueue_add_split(struct vring_virtqueue *vq,
>  	struct vring_desc_extra *extra;
>  	struct scatterlist *sg;
>  	struct vring_desc *desc;
> -	unsigned int i, n, avail, descs_used, prev, err_idx;
> +	unsigned int i, n, c, avail, descs_used, err_idx;
>  	int head;
>  	bool indirect;
>  
> @@ -626,46 +626,47 @@ static inline int virtqueue_add_split(struct vring_virtqueue *vq,
>  		return -ENOSPC;
>  	}
>  
> +	c = 0;

initialize at point of declaration?

>  	for (n = 0; n < out_sgs; n++) {
> +		sg = sgs[n];
>  		for (sg = sgs[n]; sg; sg = sg_next(sg)) {
>  			dma_addr_t addr;
>  			u32 len;
> +			u16 flags = 0;
>  
>  			if (vring_map_one_sg(vq, sg, DMA_TO_DEVICE, &addr, &len, premapped))
>  				goto unmap_release;
>  
> -			prev = i;
> +			if (++c != total_sg)
> +				flags = VRING_DESC_F_NEXT;
> +

Don't like it how the logic is split.
flags isn't used before that.
So I prefer:

	flags = ++c == total_sg ? 0 : VRING_DESC_F_NEXT;

and at that point, we do not really need flags anymore:


>  			/* Note that we trust indirect descriptor
>  			 * table since it use stream DMA mapping.
>  			 */
>  			i = virtqueue_add_desc_split(vq, desc, extra, i, addr, len,
> -						     VRING_DESC_F_NEXT,
> +						     flags,

So just:


	i = virtqueue_add_desc_split(vq, desc, extra, i, addr, len,
				     ++c == total_sg ? 0 : VRING_DESC_F_NEXT,

here and we are done.


>  						     premapped);
>  		}
>  	}
>  	for (; n < (out_sgs + in_sgs); n++) {
>  		for (sg = sgs[n]; sg; sg = sg_next(sg)) {
> +			u16 flags = VRING_DESC_F_WRITE;
>  			dma_addr_t addr;
>  			u32 len;
>  
>  			if (vring_map_one_sg(vq, sg, DMA_FROM_DEVICE, &addr, &len, premapped))
>  				goto unmap_release;
>  
> -			prev = i;
> +			if (++c != total_sg)
> +				flags |= VRING_DESC_F_NEXT;
> +

Don't like it that above it's "=" here it is "|=".
And flags isn't used before that.
So I prefer:

	flags = ++c == total_sg ? VRING_DESC_F_WRITE : VRING_DESC_F_WRITE | VRING_DESC_F_NEXT;

and again we don't really need the variable:


>  			/* Note that we trust indirect descriptor
>  			 * table since it use stream DMA mapping.
>  			 */
>  			i = virtqueue_add_desc_split(vq, desc, extra, i, addr, len,
> -						     VRING_DESC_F_NEXT |
> -						     VRING_DESC_F_WRITE,


so just:

		i = virtqueue_add_desc_split(vq, desc, extra, i, addr, len,
					     (++c == total_sg ? 0 : VRING_DESC_F_NEXT) |
					     VRING_DESC_F_WRITE,


is clearer, and the patch smaller.



> -						     premapped);
> +						     flags, premapped);
>  		}
>  	}
> -	/* Last one doesn't continue. */
> -	desc[prev].flags &= cpu_to_virtio16(vq->vq.vdev, ~VRING_DESC_F_NEXT);
> -	if (!indirect && vring_need_unmap_buffer(vq, &extra[prev]))
> -		vq->split.desc_extra[prev & (vq->split.vring.num - 1)].flags &=
> -			~VRING_DESC_F_NEXT;
>  
>  	if (indirect) {
>  		/* Now that the indirect table is filled in, map it. */
> -- 
> 2.34.1