Hi Simon, I appreciate the review, Yes we will look in some of the other ether types and see if it's something we think is needed to prioritize as well.
Thanks, Roi Sent from Nine<http://www.9folders.com/> ________________________________ From: Simon Horman <[email protected]> Sent: Tuesday, June 4, 2024 11:56 To: Roi Dayan Cc: Ilya Maximets; [email protected]; Maor Dickman Subject: Re: [ovs-dev] [PATCH 1/1] netdev-offload-tc: Reserve lower tc prio for vlan ethertype. On Thu, May 30, 2024 at 09:31:06AM +0300, Roi Dayan via dev wrote: > > > On 28/05/2024 20:12, Ilya Maximets wrote: > > On 5/26/24 10:31, Roi Dayan via dev wrote: > >> From: Maor Dickman <[email protected]> > >> > >> The cited commit reserved lower tc priorities for IP ethertypes in order > >> to give IP traffic higher priority than other management traffic. > >> In case of of vlan encap traffic, IP traffic will still get lower > >> priority. > >> > >> Fix it by also reserving low priority tc prio for vlan. > >> > >> Fixes: c230c7579c14 ("netdev-offload-tc: Reserve lower tc prios for ip > >> ethertypes") > >> Signed-off-by: Maor Dickman <[email protected]> > >> Acked-by: Roi Dayan <[email protected]> > >> --- > >> lib/netdev-offload-tc.c | 2 ++ > >> lib/tc.h | 1 + > >> 2 files changed, 3 insertions(+) > >> > >> diff --git a/lib/netdev-offload-tc.c b/lib/netdev-offload-tc.c > >> index 921d5231777e..3be1c08d24f6 100644 > >> --- a/lib/netdev-offload-tc.c > >> +++ b/lib/netdev-offload-tc.c > >> @@ -400,6 +400,8 @@ get_next_available_prio(ovs_be16 protocol) > >> return TC_RESERVED_PRIORITY_IPV4; > >> } else if (protocol == htons(ETH_P_IPV6)) { > >> return TC_RESERVED_PRIORITY_IPV6; > >> + } else if (protocol == htons(ETH_P_8021Q)) { > > > > Should 802.1ad traffic also get the priority? > > What about MPLS? > > > > Best regards, Ilya Maximets. > > Hi Ilya, > > It is correct there could be more types that could benefit from a reserved > prio but from what we saw in the field we didn't notice a real usage > of 802.1ad or MPLS while 8021q was being used actively and we noticed the > performance degradation and improvement after using the reserved prio. > We think this patch can help many active openvswitch users and in the future > if other types would be more common we could add those. Hi Roi and Ilya, FWIIW, I think it is reasonable to update 8021q up-front, i.e. accept this patch. But I also think that, as a follow-up, we should look into other Ether Types and see if they also need this treatment, rather than waiting for problems to materialise in the field. Roi, is this something your team can take on? _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
