[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20200330071655.169823-2-saeedm@mellanox.com>
Date: Mon, 30 Mar 2020 00:16:52 -0700
From: Saeed Mahameed <saeedm@...lanox.com>
To: "David S. Miller" <davem@...emloft.net>, kuba@...nel.org
Cc: netdev@...r.kernel.org, Mark Zhang <markz@...lanox.com>,
Roi Dayan <roid@...lanox.com>,
Saeed Mahameed <saeedm@...lanox.com>
Subject: [net-next 1/4] net/mlx5: Use a separate work queue for fib event handling
From: Mark Zhang <markz@...lanox.com>
In VF lag mode when remove the bonding module without bring down the
bond device first, we could potentially have circular dependency when we
unload IB devices and also handle fib events:
1. The bond work starts first;
2. The "modprobe -rv bonding" process tries to release the bond device,
with the "pernet_ops_rwsem" lock hold;
3. The bond work blocks in unregister_netdevice_notifier() and waits for
the lock because fib event came right before;
4. The kernel fib module tries to free all the fib entries by broadcasting
the "FIB_EVENT_NH_DEL" event;
5. Upon the fib event this lag_mp module holds the fib lock and queue a
fib work.
So:
bond work -> modprobe task -> kernel fib module -> lag_mp -> bond work
Today we either reload IB devices in roce lag in nic mode or either handle
fib events in switchdev mode, but a new feature could change that we'll
need to reload IB devices also in switchdev mode so this is a future proof
fix as one may not notice this later.
Signed-off-by: Mark Zhang <markz@...lanox.com>
Reviewed-by: Roi Dayan <roid@...lanox.com>
Signed-off-by: Saeed Mahameed <saeedm@...lanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/lag_mp.c | 14 ++++++++++----
drivers/net/ethernet/mellanox/mlx5/core/lag_mp.h | 1 +
2 files changed, 11 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.c b/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.c
index 416676c35b1f..e9089a793632 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.c
@@ -93,9 +93,8 @@ static void mlx5_lag_set_port_affinity(struct mlx5_lag *ldev,
static void mlx5_lag_fib_event_flush(struct notifier_block *nb)
{
struct lag_mp *mp = container_of(nb, struct lag_mp, fib_nb);
- struct mlx5_lag *ldev = container_of(mp, struct mlx5_lag, lag_mp);
- flush_workqueue(ldev->wq);
+ flush_workqueue(mp->wq);
}
struct mlx5_fib_event_work {
@@ -293,7 +292,7 @@ static int mlx5_lag_fib_event(struct notifier_block *nb,
return NOTIFY_DONE;
}
- queue_work(ldev->wq, &fib_work->work);
+ queue_work(mp->wq, &fib_work->work);
return NOTIFY_DONE;
}
@@ -306,11 +305,17 @@ int mlx5_lag_mp_init(struct mlx5_lag *ldev)
if (mp->fib_nb.notifier_call)
return 0;
+ mp->wq = create_singlethread_workqueue("mlx5_lag_mp");
+ if (!mp->wq)
+ return -ENOMEM;
+
mp->fib_nb.notifier_call = mlx5_lag_fib_event;
err = register_fib_notifier(&init_net, &mp->fib_nb,
mlx5_lag_fib_event_flush, NULL);
- if (err)
+ if (err) {
+ destroy_workqueue(mp->wq);
mp->fib_nb.notifier_call = NULL;
+ }
return err;
}
@@ -323,5 +328,6 @@ void mlx5_lag_mp_cleanup(struct mlx5_lag *ldev)
return;
unregister_fib_notifier(&init_net, &mp->fib_nb);
+ destroy_workqueue(mp->wq);
mp->fib_nb.notifier_call = NULL;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.h b/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.h
index 79be89e9c7a4..258ac7b2964e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag_mp.h
@@ -16,6 +16,7 @@ enum mlx5_lag_port_affinity {
struct lag_mp {
struct notifier_block fib_nb;
struct fib_info *mfi; /* used in tracking fib events */
+ struct workqueue_struct *wq;
};
#ifdef CONFIG_MLX5_ESWITCH
--
2.25.1
Powered by blists - more mailing lists