Re: [PATCH net v4] vmxnet3: correctly report gso type for UDP tunnels

[Paolo Abeni]

On 6/3/25 9:23 AM, Simon Horman wrote:
> On Fri, May 30, 2025 at 03:27:00PM +0000, Ronak Doshi wrote:
>> Commit 3d010c8031e3 ("udp: do not accept non-tunnel GSO skbs landing
>> in a tunnel") added checks in linux stack to not accept non-tunnel
>> GRO packets landing in a tunnel. This exposed an issue in vmxnet3
>> which was not correctly reporting GRO packets for tunnel packets.
>>
>> This patch fixes this issue by setting correct GSO type for the
>> tunnel packets.
>>
>> Currently, vmxnet3 does not support reporting inner fields for LRO
>> tunnel packets. The issue is not seen for egress drivers that do not
>> use skb inner fields. The workaround is to enable tnl-segmentation
>> offload on the egress interfaces if the driver supports it. This
>> problem pre-exists this patch fix and can be addressed as a separate
>> future patch.
>>
>> Fixes: dacce2be3312 ("vmxnet3: add geneve and vxlan tunnel offload support")
>> Signed-off-by: Ronak Doshi <ronak.doshi@xxxxxxxxxxxx>
>> Acked-by: Guolin Yang <guolin.yang@xxxxxxxxxxxx>
>>
>> Changes v1-->v2:
>>   Do not set encapsulation bit as inner fields are not updated
>> Changes v2-->v3:
>>   Update the commit message explaining the next steps to address
>>   segmentation issues that pre-exists this patch fix.
>> Changes v3->v4:
>>   Update the commit message to clarify the workaround.
>> ---
>>  drivers/net/vmxnet3/vmxnet3_drv.c | 26 ++++++++++++++++++++++++++
>>  1 file changed, 26 insertions(+)
>>
>> diff --git a/drivers/net/vmxnet3/vmxnet3_drv.c b/drivers/net/vmxnet3/vmxnet3_drv.c
>> index c676979c7ab9..287b7c20c0d6 100644
>> --- a/drivers/net/vmxnet3/vmxnet3_drv.c
>> +++ b/drivers/net/vmxnet3/vmxnet3_drv.c
>> @@ -1568,6 +1568,30 @@ vmxnet3_get_hdr_len(struct vmxnet3_adapter *adapter, struct sk_buff *skb,
>>  	return (hlen + (hdr.tcp->doff << 2));
>>  }
>>  
>> +static void
>> +vmxnet3_lro_tunnel(struct sk_buff *skb, __be16 ip_proto)
>> +{
>> +	struct udphdr *uh = NULL;
>> +
>> +	if (ip_proto == htons(ETH_P_IP)) {
>> +		struct iphdr *iph = (struct iphdr *)skb->data;
>> +
>> +		if (iph->protocol == IPPROTO_UDP)
>> +			uh = (struct udphdr *)(iph + 1);
>> +	} else {
>> +		struct ipv6hdr *iph = (struct ipv6hdr *)skb->data;
>> +
>> +		if (iph->nexthdr == IPPROTO_UDP)
>> +			uh = (struct udphdr *)(iph + 1);
>> +	}
> 
> Hi Ronak,
> 
> Possibly a naive question, but does skb->data always contain an iphdr
> or ipv6hdr? Or perhaps more to the point, is it safe to assume IPv6
> is ip_proto is not ETH_P_IP?

I think it's safe, or at least the guest can assume that. Otherwise
there is a bug in the hypervisor cooking the descriptor, and the guest
can do little to nothing is such scenario.

/P