netdev - [net-next 12/13] net/mlx5e: Rely on rcu instead of rtnl lock when getting upper dev

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20190729234934.23595-13-saeedm@mellanox.com>
Date:   Mon, 29 Jul 2019 23:50:34 +0000
From:   Saeed Mahameed <saeedm@...lanox.com>
To:     "David S. Miller" <davem@...emloft.net>
CC:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        Vlad Buslov <vladbu@...lanox.com>,
        Jianbo Liu <jianbol@...lanox.com>,
        Roi Dayan <roid@...lanox.com>,
        Saeed Mahameed <saeedm@...lanox.com>
Subject: [net-next 12/13] net/mlx5e: Rely on rcu instead of rtnl lock when
 getting upper dev

From: Vlad Buslov <vladbu@...lanox.com>

Function netdev_master_upper_dev_get() generates warning if caller doesn't
hold rtnl lock. Modify rules update path to use rcu version of that
function.

Signed-off-by: Vlad Buslov <vladbu@...lanox.com>
Reviewed-by: Jianbo Liu <jianbol@...lanox.com>
Reviewed-by: Roi Dayan <roid@...lanox.com>
Signed-off-by: Saeed Mahameed <saeedm@...lanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/en/tc_tun.c    | 14 +++++++++++++-
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c    |  6 +++++-
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
index ae439d95f5a3..4c4620db3d31 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_tun.c
@@ -31,11 +31,23 @@ static int get_route_and_out_devs(struct mlx5e_priv *priv,
 
 	real_dev = is_vlan_dev(dev) ? vlan_dev_real_dev(dev) : dev;
 	uplink_dev = mlx5_eswitch_uplink_get_proto_dev(esw, REP_ETH);
-	uplink_upper = netdev_master_upper_dev_get(uplink_dev);
+
+	rcu_read_lock();
+	uplink_upper = netdev_master_upper_dev_get_rcu(uplink_dev);
+	/* mlx5_lag_is_sriov() is a blocking function which can't be called
+	 * while holding rcu read lock. Take the net_device for correctness
+	 * sake.
+	 */
+	if (uplink_upper)
+		dev_hold(uplink_upper);
+	rcu_read_unlock();
+
 	dst_is_lag_dev = (uplink_upper &&
 			  netif_is_lag_master(uplink_upper) &&
 			  real_dev == uplink_upper &&
 			  mlx5_lag_is_sriov(priv->mdev));
+	if (uplink_upper)
+		dev_put(uplink_upper);
 
 	/* if the egress device isn't on the same HW e-switch or
 	 * it's a LAG device, use the uplink
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 714aa9d7180b..595a4c5667ea 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -2978,12 +2978,16 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
 			if (netdev_port_same_parent_id(priv->netdev, out_dev)) {
 				struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
 				struct net_device *uplink_dev = mlx5_eswitch_uplink_get_proto_dev(esw, REP_ETH);
-				struct net_device *uplink_upper = netdev_master_upper_dev_get(uplink_dev);
+				struct net_device *uplink_upper;
 
+				rcu_read_lock();
+				uplink_upper =
+					netdev_master_upper_dev_get_rcu(uplink_dev);
 				if (uplink_upper &&
 				    netif_is_lag_master(uplink_upper) &&
 				    uplink_upper == out_dev)
 					out_dev = uplink_dev;
+				rcu_read_unlock();
 
 				if (is_vlan_dev(out_dev)) {
 					err = add_vlan_push_action(priv, attr,
-- 
2.21.0