On 9/5/25 2:31 PM, David Marchand wrote:
> On Fri, 29 Aug 2025 at 11:41, David Marchand via dev
> <ovs-dev@openvswitch.org> wrote:
>>
>> TSO packets were incorrectly treated as too big by the check_pkt_len
>> action with the userspace datapath.
>> Adjust the check by looking at the requested segment size.
>>
>> Fixes: 29cf9c1b3b9c ("userspace: Add TCP Segmentation Offload support")
>> Reported-at: https://issues.redhat.com/browse/FDP-1631
>> Signed-off-by: David Marchand <david.march...@redhat.com>
>
> Some additional thought.. this issue seems to affect
> xlate_check_pkt_larger() when the datapath does not support
> check_pkt_len.
> It is not clear to me if it is worth fixing: our kernel and userspace
> datapath have an implementation for the check_pkt_len action.
> Maybe logging an error or adding a coverage counter for gso packets is enough.
In case of kernel datapath, gso packet will be segmented before sending
to userspace, so there shouldn't be any issues. There would be many
more issues with PMTUD if we allowed GSO packets to go to userspace.
In case of userspace datapath, it's part of the same codebase, and we
know that the action is supported. Ideally, we should segment packets
before doing an "upcall", but our current dp-packet infrastructure does
not allow that.
Best regards, Ilya Maximets.
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
