On 08/10/2024 16:52, Eelco Chaudron wrote:
> This patch adds support for TC to match on tunnel flags,
> if supported by the kernel. If the kernel does not support this,
> it will fall back to the previous behavior, ignoring the DF and
> CSUM flags.
>
> Signed-off-by: Eelco Chaudron <[email protected]>
> ---
> lib/netdev-offload-tc.c | 74 +++++++++++++++++++++++++++++++++++++----
> 1 file changed, 68 insertions(+), 6 deletions(-)
>
> diff --git a/lib/netdev-offload-tc.c b/lib/netdev-offload-tc.c
> index fcd206f2c..53c4245b9 100644
> --- a/lib/netdev-offload-tc.c
> +++ b/lib/netdev-offload-tc.c
> @@ -737,6 +737,36 @@ flower_tun_opt_to_match(struct match *match, struct
> tc_flower *flower)
> }
> }
>
> +static void
> +flower_tun_enc_flags_to_match(struct match *match, struct tc_flower *flower)
> +{
> + uint32_t tc_flags = flower->key.tunnel.tc_enc_flags;
> + uint32_t tc_mask = flower->mask.tunnel.tc_enc_flags;
> + uint16_t *m_flags = &match->flow.tunnel.flags;
> + uint16_t *m_mask = &match->wc.masks.tunnel.flags;
> +
> + if (tc_mask & TCA_FLOWER_KEY_FLAGS_TUNNEL_OAM) {
> + if (tc_flags & TCA_FLOWER_KEY_FLAGS_TUNNEL_OAM) {
> + *m_flags |= FLOW_TNL_F_OAM;
> + }
> + *m_mask |= FLOW_TNL_F_OAM;
> + }
> +
> + if (tc_mask & TCA_FLOWER_KEY_FLAGS_TUNNEL_DONT_FRAGMENT) {
> + if (tc_flags & TCA_FLOWER_KEY_FLAGS_TUNNEL_DONT_FRAGMENT) {
> + *m_flags |= FLOW_TNL_F_DONT_FRAGMENT;
> + }
> + *m_mask |= FLOW_TNL_F_DONT_FRAGMENT;
> + }
> +
> + if (tc_mask & TCA_FLOWER_KEY_FLAGS_TUNNEL_CSUM) {
> + if (tc_flags & TCA_FLOWER_KEY_FLAGS_TUNNEL_CSUM) {
> + *m_flags |= FLOW_TNL_F_CSUM;
> + }
> + *m_mask |= FLOW_TNL_F_CSUM;
> + }
> +}
> +
> static void
> parse_tc_flower_to_stats(struct tc_flower *flower,
> struct dpif_flow_stats *stats)
> @@ -1272,6 +1302,9 @@ parse_tc_flower_to_match(const struct netdev *netdev,
> flower->key.tunnel.gbp.flags,
> flower->mask.tunnel.gbp.flags);
> }
> + if (flower->mask.tunnel.tc_enc_flags) {
> + flower_tun_enc_flags_to_match(match, flower);
> + }
>
> if (!strcmp(netdev_get_type(netdev), "geneve")) {
> flower_tun_opt_to_match(match, flower);
> @@ -2293,12 +2326,41 @@ netdev_tc_flow_put(struct netdev *netdev, struct
> match *match,
> memset(&tnl_mask->gbp_flags, 0, sizeof tnl_mask->gbp_flags);
> tnl_mask->flags &= ~FLOW_TNL_F_KEY;
>
> - /* XXX: This is wrong! We're ignoring DF and CSUM flags
> configuration
> - * requested by the user. However, TC for now has no way to pass
> - * these flags in a flower key and their masks are set by default,
> - * meaning tunnel offloading will not work at all if not cleared.
> - * Keeping incorrect behavior for now. */
> - tnl_mask->flags &= ~(FLOW_TNL_F_DONT_FRAGMENT | FLOW_TNL_F_CSUM);
> + if (enc_flags_support) {
> + if (tnl_mask->flags & FLOW_TNL_F_OAM) {
> + if (tnl->flags & FLOW_TNL_F_OAM) {
> + flower.key.tunnel.tc_enc_flags |=
> + TCA_FLOWER_KEY_FLAGS_TUNNEL_OAM;
> + }
> + flower.mask.tunnel.tc_enc_flags |=
> + TCA_FLOWER_KEY_FLAGS_TUNNEL_OAM;
> + tnl_mask->flags &= ~FLOW_TNL_F_OAM;
> + }
> + if (tnl_mask->flags & FLOW_TNL_F_DONT_FRAGMENT) {
> + if (tnl->flags & FLOW_TNL_F_DONT_FRAGMENT) {
> + flower.key.tunnel.tc_enc_flags |=
> + TCA_FLOWER_KEY_FLAGS_TUNNEL_DONT_FRAGMENT;
> + }
> + flower.mask.tunnel.tc_enc_flags |=
> + TCA_FLOWER_KEY_FLAGS_TUNNEL_DONT_FRAGMENT;
> + tnl_mask->flags &= ~FLOW_TNL_F_DONT_FRAGMENT;
> + }
> + if (tnl_mask->flags & FLOW_TNL_F_CSUM) {
> + if (tnl->flags & FLOW_TNL_F_CSUM) {
> + flower.key.tunnel.tc_enc_flags |=
> + TCA_FLOWER_KEY_FLAGS_TUNNEL_CSUM;
> + }
> + flower.mask.tunnel.tc_enc_flags |=
> + TCA_FLOWER_KEY_FLAGS_TUNNEL_CSUM;
> + tnl_mask->flags &= ~FLOW_TNL_F_CSUM;
> + }
> + } else {
> + /* For kernels not supporting the encapsulation flags we're
> + * ignoring DF and CSUM flags configuration requested by the
> user.
> + * This to keep the old, incorrect behaviour, and allow tunnels
> to
> + * be offloaded by TC with these kernels. */
> + tnl_mask->flags &= ~(FLOW_TNL_F_DONT_FRAGMENT | FLOW_TNL_F_CSUM);
> + }
>
> if (!strcmp(netdev_get_type(netdev), "geneve")) {
> err = flower_match_to_tun_opt(&flower, tnl, tnl_mask);
Acked-by: Roi Dayan <[email protected]>
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
