On Tue, Sep 24, 2024 at 9:32 AM Xuan Zhuo <xuanzhuo@xxxxxxxxxxxxxxxxx> wrote:
>
> Because the af-xdp will introduce a new xmit type, so I refactor the
> xmit type mechanism first.
>
> In general, pointers are aligned to 4 or 8 bytes.
I think this needs some clarification, the alignment seems to depend
on the lowest common multiple between the alignments of all struct
members. So we know both xdp_frame and sk_buff are at least 4 bytes
aligned.
If we want to reuse the lowest bit of pointers in AF_XDP, the
alignment of the data structure should be at least 4 bytes.
> If it is aligned to 4
> bytes, then only two bits are free for a pointer. But there are 4 types
> here, so we can't use bits to distinguish them. And 2 bits is enough for
> 4 types:
>
> 00 for skb
> 01 for SKB_ORPHAN
> 10 for XDP
> 11 for af-xdp tx
>
> Signed-off-by: Xuan Zhuo <xuanzhuo@xxxxxxxxxxxxxxxxx>
> ---
> drivers/net/virtio_net.c | 90 +++++++++++++++++++++++-----------------
> 1 file changed, 51 insertions(+), 39 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 630e5b21ad69..41a5ea9b788d 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -45,9 +45,6 @@ module_param(napi_tx, bool, 0644);
> #define VIRTIO_XDP_TX BIT(0)
> #define VIRTIO_XDP_REDIR BIT(1)
>
> -#define VIRTIO_XDP_FLAG BIT(0)
> -#define VIRTIO_ORPHAN_FLAG BIT(1)
> -
> /* RX packet size EWMA. The average packet size is used to determine the packet
> * buffer size when refilling RX rings. As the entire RX ring may be refilled
> * at once, the weight is chosen so that the EWMA will be insensitive to short-
> @@ -512,34 +509,35 @@ static struct sk_buff *virtnet_skb_append_frag(struct sk_buff *head_skb,
> struct page *page, void *buf,
> int len, int truesize);
>
> -static bool is_xdp_frame(void *ptr)
> -{
> - return (unsigned long)ptr & VIRTIO_XDP_FLAG;
> -}
> +enum virtnet_xmit_type {
> + VIRTNET_XMIT_TYPE_SKB,
> + VIRTNET_XMIT_TYPE_SKB_ORPHAN,
> + VIRTNET_XMIT_TYPE_XDP,
> +};
>
> -static void *xdp_to_ptr(struct xdp_frame *ptr)
> -{
> - return (void *)((unsigned long)ptr | VIRTIO_XDP_FLAG);
> -}
> +/* We use the last two bits of the pointer to distinguish the xmit type. */
> +#define VIRTNET_XMIT_TYPE_MASK (BIT(0) | BIT(1))
>
> -static struct xdp_frame *ptr_to_xdp(void *ptr)
> +static enum virtnet_xmit_type virtnet_xmit_ptr_strip(void **ptr)
Nit: not a native speaker but I think something like pack/unpack might
be better.
With those changes.
Acked-by: Jason Wang <jasowang@xxxxxxxxxx>
Thanks
