On 6/16/25 5:20 AM, Jason Wang wrote:

On Thu, Jun 12, 2025 at 6:18 PM Paolo Abeni [off-list ref] wrote:

quoted

On 6/12/25 6:05 AM, Jason Wang wrote:

quoted

I wonder why virtio_net_chk_data_valid() is not part of the
virtio_net_hdr_tnl_to_skb().

It can't be part of virtio_net_hdr_tnl_to_skb(), as hdr to skb
conversion is actually not symmetric with respect to the checksum - only
the driver handles DATA_VALID.

Tun must not call virtio_net_chk_data_valid()  (or whatever different
name will use).

Note that we've already dealt with this via introducing a boolean
has_data_valid:

static inline int virtio_net_hdr_from_skb(const struct sk_buff *skb,
                                          struct virtio_net_hdr *hdr,
                                          bool little_endian,
                                          bool has_data_valid,
                                          int vlan_hlen)

I'll keep the csum offload function separated, to avoid adding even more
argument to the common helper.

/P