Re: [PATCH net v4 2/4] virtio-net: Ensure hdr_len is not set unless the header is forwarded to the device.
From: Jason Wang <jasowang@redhat.com>
Date: 2025-10-30 02:42:49
Also in:
virtualization
On Wed, Oct 29, 2025 at 11:09 AM Xuan Zhuo [off-list ref] wrote:
Although `virtio_net_hdr_from_skb` is used in several places outside of `virtio-net.c`, the `hdr_len` field is only utilized by the device according to the specification. Therefore, we do not need to set `hdr_len` unless the header is actually passed to the device. Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
I wonder if this will cause any issue consider hdr_len is just a hint. E.g device needs to survive from hdr_len = 0.
quoted hunk ↗ jump to hunk
--- include/linux/virtio_net.h | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-)diff --git a/include/linux/virtio_net.h b/include/linux/virtio_net.h index 4d1780848d0e..710ae0d2d336 100644 --- a/include/linux/virtio_net.h +++ b/include/linux/virtio_net.h@@ -218,9 +218,14 @@ static inline int virtio_net_hdr_from_skb(const struct sk_buff *skb, if (skb_is_gso(skb)) { struct skb_shared_info *sinfo = skb_shinfo(skb); - /* This is a hint as to how much should be linear. */ - hdr->hdr_len = __cpu_to_virtio16(little_endian, - skb_headlen(skb)); + /* In certain code paths (such as the af_packet.c receive path), + * this function may be called without a transport header. + * In this case, we do not need to set the hdr_len. + */ + if (skb_transport_header_was_set(skb)) + hdr->hdr_len = __cpu_to_virtio16(little_endian, + skb_headlen(skb)); + hdr->gso_size = __cpu_to_virtio16(little_endian, sinfo->gso_size); if (sinfo->gso_type & SKB_GSO_TCPV4) --2.32.0.3.g01195cf9f
Thanks