netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Saeed Mahameed <saeed@kernel.org>
To: "David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>
Cc: netdev@vger.kernel.org, Dima Chumak <dchumak@nvidia.com>,
	Paul Blakey <paulb@nvidia.com>,
	Saeed Mahameed <saeedm@nvidia.com>
Subject: [net 3/6] net/mlx5e: Offload tuple rewrite for non-CT flows
Date: Mon, 22 Mar 2021 13:25:21 -0700	[thread overview]
Message-ID: <20210322202524.68886-4-saeed@kernel.org> (raw)
In-Reply-To: <20210322202524.68886-1-saeed@kernel.org>

From: Dima Chumak <dchumak@nvidia.com>

Setting connection tracking OVS flows and then setting non-CT flows that
use tuple rewrite action (e.g. mod_tp_dst), causes the latter flows not
being offloaded.

Fix by using a stricter condition in modify_header_match_supported() to
check tuple rewrite support only for flows with CT action. The check is
factored out into standalone modify_tuple_supported() function to aid
readability.

Fixes: 7e36feeb0467 ("net/mlx5e: CT: Don't offload tuple rewrites for established tuples")
Signed-off-by: Dima Chumak <dchumak@nvidia.com>
Reviewed-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../ethernet/mellanox/mlx5/core/en/tc_ct.c    |  3 +-
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   | 44 ++++++++++++++-----
 2 files changed, 35 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
index f3f6eb081948..b2cd29847a37 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
@@ -1181,7 +1181,8 @@ int mlx5_tc_ct_add_no_trk_match(struct mlx5_flow_spec *spec)
 
 	mlx5e_tc_match_to_reg_get_match(spec, CTSTATE_TO_REG,
 					&ctstate, &ctstate_mask);
-	if (ctstate_mask)
+
+	if ((ctstate & ctstate_mask) == MLX5_CT_STATE_TRK_BIT)
 		return -EOPNOTSUPP;
 
 	ctstate_mask |= MLX5_CT_STATE_TRK_BIT;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 3359098c51d4..df2a0af854bb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -2909,6 +2909,37 @@ static int is_action_keys_supported(const struct flow_action_entry *act,
 	return 0;
 }
 
+static bool modify_tuple_supported(bool modify_tuple, bool ct_clear,
+				   bool ct_flow, struct netlink_ext_ack *extack,
+				   struct mlx5e_priv *priv,
+				   struct mlx5_flow_spec *spec)
+{
+	if (!modify_tuple || ct_clear)
+		return true;
+
+	if (ct_flow) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "can't offload tuple modification with non-clear ct()");
+		netdev_info(priv->netdev,
+			    "can't offload tuple modification with non-clear ct()");
+		return false;
+	}
+
+	/* Add ct_state=-trk match so it will be offloaded for non ct flows
+	 * (or after clear action), as otherwise, since the tuple is changed,
+	 * we can't restore ct state
+	 */
+	if (mlx5_tc_ct_add_no_trk_match(spec)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "can't offload tuple modification with ct matches and no ct(clear) action");
+		netdev_info(priv->netdev,
+			    "can't offload tuple modification with ct matches and no ct(clear) action");
+		return false;
+	}
+
+	return true;
+}
+
 static bool modify_header_match_supported(struct mlx5e_priv *priv,
 					  struct mlx5_flow_spec *spec,
 					  struct flow_action *flow_action,
@@ -2947,18 +2978,9 @@ static bool modify_header_match_supported(struct mlx5e_priv *priv,
 			return err;
 	}
 
-	/* Add ct_state=-trk match so it will be offloaded for non ct flows
-	 * (or after clear action), as otherwise, since the tuple is changed,
-	 *  we can't restore ct state
-	 */
-	if (!ct_clear && modify_tuple &&
-	    mlx5_tc_ct_add_no_trk_match(spec)) {
-		NL_SET_ERR_MSG_MOD(extack,
-				   "can't offload tuple modify header with ct matches");
-		netdev_info(priv->netdev,
-			    "can't offload tuple modify header with ct matches");
+	if (!modify_tuple_supported(modify_tuple, ct_clear, ct_flow, extack,
+				    priv, spec))
 		return false;
-	}
 
 	ip_proto = MLX5_GET(fte_match_set_lyr_2_4, headers_v, ip_protocol);
 	if (modify_ip_header && ip_proto != IPPROTO_TCP &&
-- 
2.30.2


  parent reply	other threads:[~2021-03-22 20:26 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-22 20:25 [pull request][net 0/6] mlx5 fixes 2021-03-22 Saeed Mahameed
2021-03-22 20:25 ` [net 1/6] net/mlx5: Add back multicast stats for uplink representor Saeed Mahameed
2021-03-23  0:30   ` patchwork-bot+netdevbpf
2021-03-22 20:25 ` [net 2/6] net/mlx5e: Allow to match on MPLS parameters only for MPLS over UDP Saeed Mahameed
2021-03-22 20:25 ` Saeed Mahameed [this message]
2021-03-22 20:25 ` [net 4/6] net/mlx5e: Fix error path for ethtool set-priv-flag Saeed Mahameed
2021-03-22 20:25 ` [net 5/6] net/mlx5e: Fix division by 0 in mlx5e_select_queue Saeed Mahameed
2021-03-22 20:25 ` [net 6/6] net/mlx5: SF, do not use ecpu bit for vhca state processing Saeed Mahameed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210322202524.68886-4-saeed@kernel.org \
    --to=saeed@kernel.org \
    --cc=davem@davemloft.net \
    --cc=dchumak@nvidia.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=paulb@nvidia.com \
    --cc=saeedm@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).