mlxsw: spectrum: Add FDB lock to prevent session interleaving

Dumping the FDB (invoked with a process context) or handling FDB
notifications (polled periodicly in delayed work) might each entail
multiple EMAD transcations due to the number of entries.

While we only allow one EMAD transaction at a time, there is nothing
stopping the dump and notification processing sessions from
interleaving. However, this is forbidden by the hardware, so we need to
make sure only one of these sessions can run at a time.

Solve this by adding a mutex ('fdb_lock'), as both kernel threads can
sleep while waiting for the response EMAD.

Fixes: 56ade8fe3f ("mlxsw: spectrum: Add initial support for Spectrum ASIC")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 4365c8bccc6d291f6f8f60a03fe8726f20ad93b5..69281ca534b198104a2fd01ecff5ece71cb7b67a 100644 (file)
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -63,6 +63,7 @@ struct mlxsw_sp {
        } fdb_notify;
 #define MLXSW_SP_DEFAULT_AGEING_TIME 300
        u32 ageing_time;
+       struct mutex fdb_lock;  /* Make sure FDB sessions are atomic. */
        struct {
                struct net_device *dev;
                unsigned int ref_count;

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
index 617fb22b5d81e75e34a96a73a3c9918e060d3cb1..80e266063aee1c7fa362a9c7b5d4b864bae0089a 100644 (file)
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c
@@ -650,6 +650,7 @@ static int mlxsw_sp_port_fdb_dump(struct mlxsw_sp_port *mlxsw_sp_port,
        if (!sfd_pl)
                return -ENOMEM;
 
+       mutex_lock(&mlxsw_sp_port->mlxsw_sp->fdb_lock);
        mlxsw_reg_sfd_pack(sfd_pl, MLXSW_REG_SFD_OP_QUERY_DUMP, 0);
        do {
                mlxsw_reg_sfd_num_rec_set(sfd_pl, MLXSW_REG_SFD_REC_MAX_COUNT);
@@ -684,6 +685,7 @@ static int mlxsw_sp_port_fdb_dump(struct mlxsw_sp_port *mlxsw_sp_port,
        } while (num_rec == MLXSW_REG_SFD_REC_MAX_COUNT);
 
 out:
+       mutex_unlock(&mlxsw_sp_port->mlxsw_sp->fdb_lock);
        kfree(sfd_pl);
        return stored_err ? stored_err : err;
 }
@@ -812,6 +814,7 @@ static void mlxsw_sp_fdb_notify_work(struct work_struct *work)
 
        mlxsw_sp = container_of(work, struct mlxsw_sp, fdb_notify.dw.work);
 
+       mutex_lock(&mlxsw_sp->fdb_lock);
        do {
                mlxsw_reg_sfn_pack(sfn_pl);
                err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(sfn), sfn_pl);
@@ -824,6 +827,7 @@ static void mlxsw_sp_fdb_notify_work(struct work_struct *work)
                        mlxsw_sp_fdb_notify_rec_process(mlxsw_sp, sfn_pl, i);
 
        } while (num_rec);
+       mutex_unlock(&mlxsw_sp->fdb_lock);
 
        kfree(sfn_pl);
        mlxsw_sp_fdb_notify_work_schedule(mlxsw_sp);
@@ -838,6 +842,7 @@ static int mlxsw_sp_fdb_init(struct mlxsw_sp *mlxsw_sp)
                dev_err(mlxsw_sp->bus_info->dev, "Failed to set default ageing time\n");
                return err;
        }
+       mutex_init(&mlxsw_sp->fdb_lock);
        INIT_DELAYED_WORK(&mlxsw_sp->fdb_notify.dw, mlxsw_sp_fdb_notify_work);
        mlxsw_sp->fdb_notify.interval = MLXSW_SP_DEFAULT_LEARNING_INTERVAL;
        mlxsw_sp_fdb_notify_work_schedule(mlxsw_sp);