Currently, the virtio-net driver will perform a pre-dma-mapping for small or mergeable RX buffer. But for small packets, a mismatched address without VIRTNET_RX_PAD and xdp_headroom is used for unmapping.
That will result in unsynchronized buffers when SWIOTLB is enabled, for example, when running as a TDX guest. This patch unifies the address passed to the virtio core into the address of the virtnet header and fixes the mismatched buffer address. Changes from v2: unify the buf that passed to the virtio core in small and merge mode. Changes from v1: Use ctx to get xdp_headroom. Fixes: 295525e29a5b ("virtio_net: merge dma operations when filling mergeable buffers") Signed-off-by: Wenbo Li <liwenbo.mar...@bytedance.com> Signed-off-by: Jiahui Cen <cenjia...@bytedance.com> Signed-off-by: Ying Fang <fangying.to...@bytedance.com> --- drivers/net/virtio_net.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index 6f4781ec2b36..9446666c84aa 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -1804,9 +1804,15 @@ static struct sk_buff *receive_small(struct net_device *dev, struct virtnet_rq_stats *stats) { unsigned int xdp_headroom = (unsigned long)ctx; - struct page *page = virt_to_head_page(buf); + struct page *page; struct sk_buff *skb; + // We passed the address of virtnet header to virtio-core, + // so truncate the padding. + buf -= VIRTNET_RX_PAD + xdp_headroom; + + page = virt_to_head_page(buf); + len -= vi->hdr_len; u64_stats_add(&stats->bytes, len); @@ -2422,8 +2428,9 @@ static int add_recvbuf_small(struct virtnet_info *vi, struct receive_queue *rq, if (unlikely(!buf)) return -ENOMEM; - virtnet_rq_init_one_sg(rq, buf + VIRTNET_RX_PAD + xdp_headroom, - vi->hdr_len + GOOD_PACKET_LEN); + buf += VIRTNET_RX_PAD + xdp_headroom; + + virtnet_rq_init_one_sg(rq, buf, vi->hdr_len + GOOD_PACKET_LEN); err = virtqueue_add_inbuf_ctx(rq->vq, rq->sg, 1, buf, ctx, gfp); if (err < 0) { -- 2.20.1