* [PATCH mlx5-next 01/12] net/mlx5: Add device cap abs_native_port_num
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic Tariq Toukan
` (10 subsequent siblings)
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Rongwei Liu,
Shay Drory, Mark Bloch, Tariq Toukan
From: Rongwei Liu <rongweil@nvidia.com>
When the abs_native_port_num is set, the native_port_num reported
by the device may not be continuous and bigger than the num_lag_ports.
Signed-off-by: Rongwei Liu <rongweil@nvidia.com>
Reviewed-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
include/linux/mlx5/mlx5_ifc.h | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index 5451ff1d4356..43b3cb4bf8d1 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -1599,7 +1599,8 @@ enum {
struct mlx5_ifc_cmd_hca_cap_bits {
u8 reserved_at_0[0x6];
u8 page_request_disable[0x1];
- u8 reserved_at_7[0x9];
+ u8 abs_native_port_num[0x1];
+ u8 reserved_at_8[0x8];
u8 shared_object_to_user_object_allowed[0x1];
u8 reserved_at_13[0xe];
u8 vhca_resource_manager[0x1];
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
2024-12-11 13:42 ` [PATCH mlx5-next 01/12] net/mlx5: Add device cap abs_native_port_num Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-16 17:55 ` Alexander Lobakin
2024-12-11 13:42 ` [PATCH net-next 03/12] net/mlx5: LAG, Support LAG over Multi-Host NICs Tariq Toukan
` (9 subsequent siblings)
11 siblings, 1 reply; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Rongwei Liu,
Mark Bloch, Tariq Toukan
From: Rongwei Liu <rongweil@nvidia.com>
Wrap the lag pf access into two new macros:
1. ldev_for_each()
2. ldev_for_each_reverse()
The maximum number of lag ports and the index to `natvie_port_num`
mapping will be handled by the two new macros.
Users shouldn't use the for loop anymore.
Signed-off-by: Rongwei Liu <rongweil@nvidia.com>
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
.../ethernet/mellanox/mlx5/core/lag/debugfs.c | 13 +-
.../net/ethernet/mellanox/mlx5/core/lag/lag.c | 181 +++++++++---------
.../net/ethernet/mellanox/mlx5/core/lag/lag.h | 14 +-
.../net/ethernet/mellanox/mlx5/core/lag/mp.c | 24 ++-
.../ethernet/mellanox/mlx5/core/lag/mpesw.c | 10 +-
.../mellanox/mlx5/core/lag/port_sel.c | 16 +-
6 files changed, 137 insertions(+), 121 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
index f4b777d4e108..798fb414d932 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/debugfs.c
@@ -105,20 +105,20 @@ static int mapping_show(struct seq_file *file, void *priv)
struct mlx5_lag *ldev;
bool hash = false;
bool lag_active;
+ int i, idx = 0;
int num_ports;
- int i;
ldev = mlx5_lag_dev(dev);
mutex_lock(&ldev->lock);
lag_active = __mlx5_lag_is_active(ldev);
if (lag_active) {
if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &ldev->mode_flags)) {
- mlx5_infer_tx_enabled(&ldev->tracker, ldev->ports, ports,
+ mlx5_infer_tx_enabled(&ldev->tracker, ldev, ports,
&num_ports);
hash = true;
} else {
- for (i = 0; i < ldev->ports; i++)
- ports[i] = ldev->v2p_map[i];
+ ldev_for_each(i, 0, ldev)
+ ports[idx++] = ldev->v2p_map[i];
num_ports = ldev->ports;
}
}
@@ -144,11 +144,8 @@ static int members_show(struct seq_file *file, void *priv)
ldev = mlx5_lag_dev(dev);
mutex_lock(&ldev->lock);
- for (i = 0; i < ldev->ports; i++) {
- if (!ldev->pf[i].dev)
- continue;
+ ldev_for_each(i, 0, ldev)
seq_printf(file, "%s\n", dev_name(ldev->pf[i].dev->device));
- }
mutex_unlock(&ldev->lock);
return 0;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
index 7f68468c2e75..73fd3f747f1a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
@@ -43,10 +43,6 @@
#include "mp.h"
#include "mpesw.h"
-enum {
- MLX5_LAG_EGRESS_PORT_1 = 1,
- MLX5_LAG_EGRESS_PORT_2,
-};
/* General purpose, use for short periods of time.
* Beware of lock dependencies (preferably, no locks should be acquired
@@ -72,7 +68,7 @@ static u8 lag_active_port_bits(struct mlx5_lag *ldev)
int num_enabled;
int idx;
- mlx5_infer_tx_enabled(&ldev->tracker, ldev->ports, enabled_ports,
+ mlx5_infer_tx_enabled(&ldev->tracker, ldev, enabled_ports,
&num_enabled);
for (idx = 0; idx < num_enabled; idx++)
active_port |= BIT_MASK(enabled_ports[idx]);
@@ -113,7 +109,7 @@ static int mlx5_cmd_create_lag(struct mlx5_core_dev *dev, u8 *ports, int mode,
return mlx5_cmd_exec_in(dev, create_lag, in);
}
-static int mlx5_cmd_modify_lag(struct mlx5_core_dev *dev, u8 num_ports,
+static int mlx5_cmd_modify_lag(struct mlx5_core_dev *dev, struct mlx5_lag *ldev,
u8 *ports)
{
u32 in[MLX5_ST_SZ_DW(modify_lag_in)] = {};
@@ -148,33 +144,31 @@ int mlx5_cmd_destroy_vport_lag(struct mlx5_core_dev *dev)
}
EXPORT_SYMBOL(mlx5_cmd_destroy_vport_lag);
-static void mlx5_infer_tx_disabled(struct lag_tracker *tracker, u8 num_ports,
+static void mlx5_infer_tx_disabled(struct lag_tracker *tracker, struct mlx5_lag *ldev,
u8 *ports, int *num_disabled)
{
int i;
*num_disabled = 0;
- for (i = 0; i < num_ports; i++) {
+ ldev_for_each(i, 0, ldev)
if (!tracker->netdev_state[i].tx_enabled ||
!tracker->netdev_state[i].link_up)
ports[(*num_disabled)++] = i;
- }
}
-void mlx5_infer_tx_enabled(struct lag_tracker *tracker, u8 num_ports,
+void mlx5_infer_tx_enabled(struct lag_tracker *tracker, struct mlx5_lag *ldev,
u8 *ports, int *num_enabled)
{
int i;
*num_enabled = 0;
- for (i = 0; i < num_ports; i++) {
+ ldev_for_each(i, 0, ldev)
if (tracker->netdev_state[i].tx_enabled &&
tracker->netdev_state[i].link_up)
ports[(*num_enabled)++] = i;
- }
if (*num_enabled == 0)
- mlx5_infer_tx_disabled(tracker, num_ports, ports, num_enabled);
+ mlx5_infer_tx_disabled(tracker, ldev, ports, num_enabled);
}
static void mlx5_lag_print_mapping(struct mlx5_core_dev *dev,
@@ -192,7 +186,7 @@ static void mlx5_lag_print_mapping(struct mlx5_core_dev *dev,
int j;
if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags)) {
- mlx5_infer_tx_enabled(tracker, ldev->ports, enabled_ports,
+ mlx5_infer_tx_enabled(tracker, ldev, enabled_ports,
&num_enabled);
for (i = 0; i < num_enabled; i++) {
err = scnprintf(buf + written, 4, "%d, ", enabled_ports[i] + 1);
@@ -203,7 +197,7 @@ static void mlx5_lag_print_mapping(struct mlx5_core_dev *dev,
buf[written - 2] = 0;
mlx5_core_info(dev, "lag map active ports: %s\n", buf);
} else {
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
for (j = 0; j < ldev->buckets; j++) {
idx = i * ldev->buckets + j;
err = scnprintf(buf + written, 10,
@@ -286,7 +280,7 @@ int mlx5_lag_dev_get_netdev_idx(struct mlx5_lag *ldev,
{
int i;
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
if (ldev->pf[i].netdev == ndev)
return i;
@@ -310,7 +304,7 @@ static bool __mlx5_lag_is_sriov(struct mlx5_lag *ldev)
* with mapping that points to active ports.
*/
static void mlx5_infer_tx_affinity_mapping(struct lag_tracker *tracker,
- u8 num_ports,
+ struct mlx5_lag *ldev,
u8 buckets,
u8 *ports)
{
@@ -323,7 +317,7 @@ static void mlx5_infer_tx_affinity_mapping(struct lag_tracker *tracker,
int i;
int j;
- for (i = 0; i < num_ports; i++) {
+ ldev_for_each(i, 0, ldev) {
if (tracker->netdev_state[i].tx_enabled &&
tracker->netdev_state[i].link_up)
enabled[enabled_ports_num++] = i;
@@ -334,15 +328,16 @@ static void mlx5_infer_tx_affinity_mapping(struct lag_tracker *tracker,
/* Use native mapping by default where each port's buckets
* point the native port: 1 1 1 .. 1 2 2 2 ... 2 3 3 3 ... 3 etc
*/
- for (i = 0; i < num_ports; i++)
+ ldev_for_each(i, 0, ldev) {
for (j = 0; j < buckets; j++) {
idx = i * buckets + j;
- ports[idx] = MLX5_LAG_EGRESS_PORT_1 + i;
+ ports[idx] = i + 1;
}
+ }
/* If all ports are disabled/enabled keep native mapping */
- if (enabled_ports_num == num_ports ||
- disabled_ports_num == num_ports)
+ if (enabled_ports_num == ldev->ports ||
+ disabled_ports_num == ldev->ports)
return;
/* Go over the disabled ports and for each assign a random active port */
@@ -358,7 +353,7 @@ static bool mlx5_lag_has_drop_rule(struct mlx5_lag *ldev)
{
int i;
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
if (ldev->pf[i].has_drop)
return true;
return false;
@@ -368,7 +363,7 @@ static void mlx5_lag_drop_rule_cleanup(struct mlx5_lag *ldev)
{
int i;
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
if (!ldev->pf[i].has_drop)
continue;
@@ -396,7 +391,7 @@ static void mlx5_lag_drop_rule_setup(struct mlx5_lag *ldev,
if (!ldev->tracker.has_inactive)
return;
- mlx5_infer_tx_disabled(tracker, ldev->ports, disabled_ports, &num_disabled);
+ mlx5_infer_tx_disabled(tracker, ldev, disabled_ports, &num_disabled);
for (i = 0; i < num_disabled; i++) {
disabled_index = disabled_ports[i];
@@ -442,7 +437,7 @@ static int _mlx5_modify_lag(struct mlx5_lag *ldev, u8 *ports)
return mlx5_cmd_modify_active_port(dev0, active_ports);
}
- return mlx5_cmd_modify_lag(dev0, ldev->ports, ports);
+ return mlx5_cmd_modify_lag(dev0, ldev, ports);
}
static struct net_device *mlx5_lag_active_backup_get_netdev(struct mlx5_core_dev *dev)
@@ -458,7 +453,7 @@ static struct net_device *mlx5_lag_active_backup_get_netdev(struct mlx5_core_dev
if (!ldev)
goto unlock;
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
if (ldev->tracker.netdev_state[i].tx_enabled)
ndev = ldev->pf[i].netdev;
if (!ndev)
@@ -483,9 +478,9 @@ void mlx5_modify_lag(struct mlx5_lag *ldev,
int i;
int j;
- mlx5_infer_tx_affinity_mapping(tracker, ldev->ports, ldev->buckets, ports);
+ mlx5_infer_tx_affinity_mapping(tracker, ldev, ldev->buckets, ports);
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
for (j = 0; j < ldev->buckets; j++) {
idx = i * ldev->buckets + j;
if (ports[idx] == ldev->v2p_map[idx])
@@ -596,9 +591,9 @@ static int mlx5_lag_create_single_fdb(struct mlx5_lag *ldev)
struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
struct mlx5_eswitch *master_esw = dev0->priv.eswitch;
int err;
- int i;
+ int i, j;
- for (i = MLX5_LAG_P1 + 1; i < ldev->ports; i++) {
+ ldev_for_each(i, 1, ldev) {
struct mlx5_eswitch *slave_esw = ldev->pf[i].dev->priv.eswitch;
err = mlx5_eswitch_offloads_single_fdb_add_one(master_esw,
@@ -608,9 +603,9 @@ static int mlx5_lag_create_single_fdb(struct mlx5_lag *ldev)
}
return 0;
err:
- for (; i > MLX5_LAG_P1; i--)
+ ldev_for_each_reverse(j, i, 1, ldev)
mlx5_eswitch_offloads_single_fdb_del_one(master_esw,
- ldev->pf[i].dev->priv.eswitch);
+ ldev->pf[j].dev->priv.eswitch);
return err;
}
@@ -671,7 +666,7 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
return err;
if (mode != MLX5_LAG_MODE_MPESW) {
- mlx5_infer_tx_affinity_mapping(tracker, ldev->ports, ldev->buckets, ldev->v2p_map);
+ mlx5_infer_tx_affinity_mapping(tracker, ldev, ldev->buckets, ldev->v2p_map);
if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &flags)) {
err = mlx5_lag_port_sel_create(ldev, tracker->hash_type,
ldev->v2p_map);
@@ -722,7 +717,7 @@ int mlx5_deactivate_lag(struct mlx5_lag *ldev)
mlx5_lag_mp_reset(ldev);
if (test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags)) {
- for (i = MLX5_LAG_P1 + 1; i < ldev->ports; i++)
+ ldev_for_each(i, 1, ldev)
mlx5_eswitch_offloads_single_fdb_del_one(master_esw,
ldev->pf[i].dev->priv.eswitch);
clear_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags);
@@ -766,7 +761,7 @@ bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
return false;
#ifdef CONFIG_MLX5_ESWITCH
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
dev = ldev->pf[i].dev;
if (mlx5_eswitch_num_vfs(dev->priv.eswitch) && !is_mdev_switchdev_mode(dev))
return false;
@@ -774,17 +769,17 @@ bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
dev = ldev->pf[MLX5_LAG_P1].dev;
mode = mlx5_eswitch_mode(dev);
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
if (mlx5_eswitch_mode(ldev->pf[i].dev) != mode)
return false;
#else
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
if (mlx5_sriov_is_enabled(ldev->pf[i].dev))
return false;
#endif
roce_support = mlx5_get_roce_state(ldev->pf[MLX5_LAG_P1].dev);
- for (i = 1; i < ldev->ports; i++)
+ ldev_for_each(i, MLX5_LAG_P2, ldev)
if (mlx5_get_roce_state(ldev->pf[i].dev) != roce_support)
return false;
@@ -795,10 +790,7 @@ void mlx5_lag_add_devices(struct mlx5_lag *ldev)
{
int i;
- for (i = 0; i < ldev->ports; i++) {
- if (!ldev->pf[i].dev)
- continue;
-
+ ldev_for_each(i, 0, ldev) {
if (ldev->pf[i].dev->priv.flags &
MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV)
continue;
@@ -812,10 +804,7 @@ void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
{
int i;
- for (i = 0; i < ldev->ports; i++) {
- if (!ldev->pf[i].dev)
- continue;
-
+ ldev_for_each(i, 0, ldev) {
if (ldev->pf[i].dev->priv.flags &
MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV)
continue;
@@ -842,7 +831,7 @@ void mlx5_disable_lag(struct mlx5_lag *ldev)
dev0->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
mlx5_rescan_drivers_locked(dev0);
}
- for (i = 1; i < ldev->ports; i++)
+ ldev_for_each(i, MLX5_LAG_P2, ldev)
mlx5_nic_vport_disable_roce(ldev->pf[i].dev);
}
@@ -854,7 +843,7 @@ void mlx5_disable_lag(struct mlx5_lag *ldev)
mlx5_lag_add_devices(ldev);
if (shared_fdb)
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
if (!(ldev->pf[i].dev->priv.flags & MLX5_PRIV_FLAGS_DISABLE_ALL_ADEV))
mlx5_eswitch_reload_ib_reps(ldev->pf[i].dev->priv.eswitch);
}
@@ -864,7 +853,7 @@ static bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev)
struct mlx5_core_dev *dev;
int i;
- for (i = MLX5_LAG_P1 + 1; i < ldev->ports; i++) {
+ ldev_for_each(i, MLX5_LAG_P1 + 1, ldev) {
dev = ldev->pf[i].dev;
if (is_mdev_switchdev_mode(dev) &&
mlx5_eswitch_vport_match_metadata_enabled(dev->priv.eswitch) &&
@@ -892,11 +881,11 @@ static bool mlx5_lag_is_roce_lag(struct mlx5_lag *ldev)
bool roce_lag = true;
int i;
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
roce_lag = roce_lag && !mlx5_sriov_is_enabled(ldev->pf[i].dev);
#ifdef CONFIG_MLX5_ESWITCH
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
roce_lag = roce_lag && is_mdev_legacy_mode(ldev->pf[i].dev);
#endif
@@ -956,7 +945,7 @@ static void mlx5_do_bond(struct mlx5_lag *ldev)
} else if (roce_lag) {
dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
mlx5_rescan_drivers_locked(dev0);
- for (i = 1; i < ldev->ports; i++) {
+ ldev_for_each(i, MLX5_LAG_P2, ldev) {
if (mlx5_get_roce_state(ldev->pf[i].dev))
mlx5_nic_vport_enable_roce(ldev->pf[i].dev);
}
@@ -966,7 +955,7 @@ static void mlx5_do_bond(struct mlx5_lag *ldev)
dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
mlx5_rescan_drivers_locked(dev0);
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
err = mlx5_eswitch_reload_ib_reps(ldev->pf[i].dev->priv.eswitch);
if (err)
break;
@@ -977,7 +966,7 @@ static void mlx5_do_bond(struct mlx5_lag *ldev)
mlx5_rescan_drivers_locked(dev0);
mlx5_deactivate_lag(ldev);
mlx5_lag_add_devices(ldev);
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
mlx5_eswitch_reload_ib_reps(ldev->pf[i].dev->priv.eswitch);
mlx5_core_err(dev0, "Failed to enable lag\n");
return;
@@ -1010,12 +999,9 @@ struct mlx5_devcom_comp_dev *mlx5_lag_get_devcom_comp(struct mlx5_lag *ldev)
int i;
mutex_lock(&ldev->lock);
- for (i = 0; i < ldev->ports; i++) {
- if (ldev->pf[i].dev) {
- devcom = ldev->pf[i].dev->priv.hca_devcom_comp;
- break;
- }
- }
+ i = get_next_ldev_func(ldev, 0);
+ if (i < MLX5_MAX_PORTS)
+ devcom = ldev->pf[i].dev->priv.hca_devcom_comp;
mutex_unlock(&ldev->lock);
return devcom;
}
@@ -1068,7 +1054,7 @@ static int mlx5_handle_changeupper_event(struct mlx5_lag *ldev,
u8 bond_status = 0;
int num_slaves = 0;
int changed = 0;
- int idx;
+ int i, idx = -1;
if (!netif_is_lag_master(upper))
return 0;
@@ -1083,8 +1069,13 @@ static int mlx5_handle_changeupper_event(struct mlx5_lag *ldev,
*/
rcu_read_lock();
for_each_netdev_in_bond_rcu(upper, ndev_tmp) {
- idx = mlx5_lag_dev_get_netdev_idx(ldev, ndev_tmp);
- if (idx >= 0) {
+ ldev_for_each(i, 0, ldev) {
+ if (ldev->pf[i].netdev == ndev_tmp) {
+ idx++;
+ break;
+ }
+ }
+ if (i < MLX5_MAX_PORTS) {
slave = bond_slave_get_rcu(ndev_tmp);
if (slave)
has_inactive |= bond_is_slave_inactive(slave);
@@ -1234,15 +1225,12 @@ static int mlx5_lag_netdev_event(struct notifier_block *this,
}
static void mlx5_ldev_add_netdev(struct mlx5_lag *ldev,
- struct mlx5_core_dev *dev,
- struct net_device *netdev)
+ struct mlx5_core_dev *dev,
+ struct net_device *netdev)
{
unsigned int fn = mlx5_get_dev_index(dev);
unsigned long flags;
- if (fn >= ldev->ports)
- return;
-
spin_lock_irqsave(&lag_lock, flags);
ldev->pf[fn].netdev = netdev;
ldev->tracker.netdev_state[fn].link_up = 0;
@@ -1257,7 +1245,7 @@ static void mlx5_ldev_remove_netdev(struct mlx5_lag *ldev,
int i;
spin_lock_irqsave(&lag_lock, flags);
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
if (ldev->pf[i].netdev == netdev) {
ldev->pf[i].netdev = NULL;
break;
@@ -1267,13 +1255,10 @@ static void mlx5_ldev_remove_netdev(struct mlx5_lag *ldev,
}
static void mlx5_ldev_add_mdev(struct mlx5_lag *ldev,
- struct mlx5_core_dev *dev)
+ struct mlx5_core_dev *dev)
{
unsigned int fn = mlx5_get_dev_index(dev);
- if (fn >= ldev->ports)
- return;
-
ldev->pf[fn].dev = dev;
dev->priv.lag = ldev;
}
@@ -1281,16 +1266,13 @@ static void mlx5_ldev_add_mdev(struct mlx5_lag *ldev,
static void mlx5_ldev_remove_mdev(struct mlx5_lag *ldev,
struct mlx5_core_dev *dev)
{
- int i;
+ int fn;
- for (i = 0; i < ldev->ports; i++)
- if (ldev->pf[i].dev == dev)
- break;
-
- if (i == ldev->ports)
+ fn = mlx5_get_dev_index(dev);
+ if (ldev->pf[fn].dev != dev)
return;
- ldev->pf[i].dev = NULL;
+ ldev->pf[fn].dev = NULL;
dev->priv.lag = NULL;
}
@@ -1406,7 +1388,6 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
mutex_lock(&ldev->lock);
mlx5_ldev_add_netdev(ldev, dev, netdev);
-
for (i = 0; i < ldev->ports; i++)
if (!ldev->pf[i].netdev)
break;
@@ -1417,6 +1398,26 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
mlx5_queue_bond_work(ldev, 0);
}
+int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx)
+{
+ int i;
+
+ for (i = start_idx; i >= end_idx; i--)
+ if (ldev->pf[i].dev)
+ return i;
+ return -1;
+}
+
+int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx)
+{
+ int i;
+
+ for (i = start_idx; i < MLX5_MAX_PORTS; i++)
+ if (ldev->pf[i].dev)
+ return i;
+ return MLX5_MAX_PORTS;
+}
+
bool mlx5_lag_is_roce(struct mlx5_core_dev *dev)
{
struct mlx5_lag *ldev;
@@ -1467,7 +1468,7 @@ bool mlx5_lag_is_master(struct mlx5_core_dev *dev)
{
struct mlx5_lag *ldev;
unsigned long flags;
- bool res;
+ bool res = false;
spin_lock_irqsave(&lag_lock, flags);
ldev = mlx5_lag_dev(dev);
@@ -1555,7 +1556,7 @@ u8 mlx5_lag_get_slave_port(struct mlx5_core_dev *dev,
if (!(ldev && __mlx5_lag_is_roce(ldev)))
goto unlock;
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
if (ldev->pf[i].netdev == slave) {
port = i;
break;
@@ -1594,13 +1595,13 @@ struct mlx5_core_dev *mlx5_lag_get_next_peer_mdev(struct mlx5_core_dev *dev, int
if (!ldev)
goto unlock;
- if (*i == ldev->ports)
+ if (*i == MLX5_MAX_PORTS)
goto unlock;
- for (idx = *i; idx < ldev->ports; idx++)
+ ldev_for_each(idx, *i, ldev)
if (ldev->pf[idx].dev != dev)
break;
- if (idx == ldev->ports) {
+ if (idx == MLX5_MAX_PORTS) {
*i = idx;
goto unlock;
}
@@ -1621,10 +1622,10 @@ int mlx5_lag_query_cong_counters(struct mlx5_core_dev *dev,
{
int outlen = MLX5_ST_SZ_BYTES(query_cong_statistics_out);
struct mlx5_core_dev **mdev;
+ int ret = 0, i, j, idx = 0;
struct mlx5_lag *ldev;
unsigned long flags;
int num_ports;
- int ret, i, j;
void *out;
out = kvzalloc(outlen, GFP_KERNEL);
@@ -1643,8 +1644,8 @@ int mlx5_lag_query_cong_counters(struct mlx5_core_dev *dev,
ldev = mlx5_lag_dev(dev);
if (ldev && __mlx5_lag_is_active(ldev)) {
num_ports = ldev->ports;
- for (i = 0; i < ldev->ports; i++)
- mdev[i] = ldev->pf[i].dev;
+ ldev_for_each(i, 0, ldev)
+ mdev[idx++] = ldev->pf[i].dev;
} else {
num_ports = 1;
mdev[MLX5_LAG_P1] = dev;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
index 50fcb1eee574..1be34eb43723 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
@@ -103,7 +103,7 @@ int mlx5_lag_dev_get_netdev_idx(struct mlx5_lag *ldev,
struct net_device *ndev);
char *mlx5_get_str_port_sel_mode(enum mlx5_lag_mode mode, unsigned long flags);
-void mlx5_infer_tx_enabled(struct lag_tracker *tracker, u8 num_ports,
+void mlx5_infer_tx_enabled(struct lag_tracker *tracker, struct mlx5_lag *ldev,
u8 *ports, int *num_enabled);
void mlx5_ldev_add_debugfs(struct mlx5_core_dev *dev);
@@ -119,9 +119,21 @@ static inline bool mlx5_lag_is_supported(struct mlx5_core_dev *dev)
if (!MLX5_CAP_GEN(dev, vport_group_manager) ||
!MLX5_CAP_GEN(dev, lag_master) ||
MLX5_CAP_GEN(dev, num_lag_ports) < 2 ||
+ mlx5_get_dev_index(dev) >= MLX5_MAX_PORTS ||
MLX5_CAP_GEN(dev, num_lag_ports) > MLX5_MAX_PORTS)
return false;
return true;
}
+#define ldev_for_each(i, start_index, ldev) \
+ for (int tmp = start_index; tmp = get_next_ldev_func(ldev, tmp), \
+ i = tmp, tmp < MLX5_MAX_PORTS; tmp++)
+
+#define ldev_for_each_reverse(i, start_index, end_index, ldev) \
+ for (int tmp = start_index, tmp1 = end_index; \
+ tmp = get_pre_ldev_func(ldev, tmp, tmp1), \
+ i = tmp, tmp >= tmp1; tmp--)
+
+int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx);
+int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx);
#endif /* __MLX5_LAG_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
index b1aa494c76ba..9596cf433815 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
@@ -153,6 +153,7 @@ static void mlx5_lag_fib_route_event(struct mlx5_lag *ldev, unsigned long event,
struct net_device *nh_dev0, *nh_dev1;
struct fib_info *fi = fen_info->fi;
struct lag_mp *mp = &ldev->lag_mp;
+ int i, dev_idx = 0;
/* Handle delete event */
if (event == FIB_EVENT_ENTRY_DEL) {
@@ -186,10 +187,12 @@ static void mlx5_lag_fib_route_event(struct mlx5_lag *ldev, unsigned long event,
if (!nh_dev1) {
if (__mlx5_lag_is_active(ldev)) {
- int i = mlx5_lag_dev_get_netdev_idx(ldev, nh_dev0);
-
- i++;
- mlx5_lag_set_port_affinity(ldev, i);
+ ldev_for_each(i, 0, ldev) {
+ dev_idx++;
+ if (ldev->pf[i].netdev == nh_dev0)
+ break;
+ }
+ mlx5_lag_set_port_affinity(ldev, dev_idx);
mlx5_lag_fib_set(mp, fi, fen_info->dst, fen_info->dst_len);
}
@@ -214,6 +217,7 @@ static void mlx5_lag_fib_nexthop_event(struct mlx5_lag *ldev,
struct fib_info *fi)
{
struct lag_mp *mp = &ldev->lag_mp;
+ int i, dev_idx = 0;
/* Check the nh event is related to the route */
if (!mp->fib.mfi || mp->fib.mfi != fi)
@@ -221,11 +225,15 @@ static void mlx5_lag_fib_nexthop_event(struct mlx5_lag *ldev,
/* nh added/removed */
if (event == FIB_EVENT_NH_DEL) {
- int i = mlx5_lag_dev_get_netdev_idx(ldev, fib_nh->fib_nh_dev);
+ ldev_for_each(i, 0, ldev) {
+ if (ldev->pf[i].netdev == fib_nh->fib_nh_dev)
+ break;
+ dev_idx++;
+ }
- if (i >= 0) {
- i = (i + 1) % 2 + 1; /* peer port */
- mlx5_lag_set_port_affinity(ldev, i);
+ if (dev_idx >= 0) {
+ dev_idx = (dev_idx + 1) % 2 + 1; /* peer port */
+ mlx5_lag_set_port_affinity(ldev, dev_idx);
}
} else if (event == FIB_EVENT_NH_ADD &&
fib_info_num_path(fi) == 2) {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index 571ea26edd0c..dd1b2caa0182 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -15,7 +15,7 @@ static void mlx5_mpesw_metadata_cleanup(struct mlx5_lag *ldev)
u32 pf_metadata;
int i;
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
dev = ldev->pf[i].dev;
esw = dev->priv.eswitch;
pf_metadata = ldev->lag_mpesw.pf_metadata[i];
@@ -36,7 +36,7 @@ static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
u32 pf_metadata;
int i, err;
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
dev = ldev->pf[i].dev;
esw = dev->priv.eswitch;
pf_metadata = mlx5_esw_match_metadata_alloc(esw);
@@ -52,7 +52,7 @@ static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
goto err_metadata;
}
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
dev = ldev->pf[i].dev;
mlx5_notifier_call_chain(dev->priv.events, MLX5_DEV_EVENT_MULTIPORT_ESW,
(void *)0);
@@ -98,7 +98,7 @@ static int enable_mpesw(struct mlx5_lag *ldev)
dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
mlx5_rescan_drivers_locked(dev0);
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
err = mlx5_eswitch_reload_ib_reps(ldev->pf[i].dev->priv.eswitch);
if (err)
goto err_rescan_drivers;
@@ -112,7 +112,7 @@ static int enable_mpesw(struct mlx5_lag *ldev)
mlx5_deactivate_lag(ldev);
err_add_devices:
mlx5_lag_add_devices(ldev);
- for (i = 0; i < ldev->ports; i++)
+ ldev_for_each(i, 0, ldev)
mlx5_eswitch_reload_ib_reps(ldev->pf[i].dev->priv.eswitch);
mlx5_mpesw_metadata_cleanup(ldev);
return err;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
index ab2717012b79..6b52b09ffc40 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
@@ -44,9 +44,7 @@ static int mlx5_lag_create_port_sel_table(struct mlx5_lag *ldev,
struct mlx5_flow_destination dest = {};
MLX5_DECLARE_FLOW_ACT(flow_act);
struct mlx5_flow_namespace *ns;
- int err, i;
- int idx;
- int j;
+ int err, i, j, k, idx;
ft_attr.max_fte = ldev->ports * ldev->buckets;
ft_attr.level = MLX5_LAG_FT_LEVEL_DEFINER;
@@ -74,7 +72,7 @@ static int mlx5_lag_create_port_sel_table(struct mlx5_lag *ldev,
dest.type = MLX5_FLOW_DESTINATION_TYPE_UPLINK;
dest.vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
flow_act.flags |= FLOW_ACT_NO_APPEND;
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
for (j = 0; j < ldev->buckets; j++) {
u8 affinity;
@@ -88,13 +86,13 @@ static int mlx5_lag_create_port_sel_table(struct mlx5_lag *ldev,
&dest, 1);
if (IS_ERR(lag_definer->rules[idx])) {
err = PTR_ERR(lag_definer->rules[idx]);
- do {
+ ldev_for_each_reverse(k, i, 0, ldev) {
while (j--) {
- idx = i * ldev->buckets + j;
+ idx = k * ldev->buckets + j;
mlx5_del_flow_rules(lag_definer->rules[idx]);
}
j = ldev->buckets;
- } while (i--);
+ };
goto destroy_fg;
}
}
@@ -346,7 +344,7 @@ static void mlx5_lag_destroy_definer(struct mlx5_lag *ldev,
int i;
int j;
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
for (j = 0; j < ldev->buckets; j++) {
idx = i * ldev->buckets + j;
mlx5_del_flow_rules(lag_definer->rules[idx]);
@@ -565,7 +563,7 @@ static int __mlx5_lag_modify_definers_destinations(struct mlx5_lag *ldev,
dest.type = MLX5_FLOW_DESTINATION_TYPE_UPLINK;
dest.vport.flags |= MLX5_FLOW_DEST_VPORT_VHCA_ID;
- for (i = 0; i < ldev->ports; i++) {
+ ldev_for_each(i, 0, ldev) {
for (j = 0; j < ldev->buckets; j++) {
idx = i * ldev->buckets + j;
if (ldev->v2p_map[idx] == ports[idx])
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-11 13:42 ` [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic Tariq Toukan
@ 2024-12-16 17:55 ` Alexander Lobakin
2024-12-17 5:44 ` rongwei liu
0 siblings, 1 reply; 34+ messages in thread
From: Alexander Lobakin @ 2024-12-16 17:55 UTC (permalink / raw)
To: Tariq Toukan
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Rongwei Liu, Mark Bloch
From: Tariq Toukan <tariqt@nvidia.com>
Date: Wed, 11 Dec 2024 15:42:13 +0200
> From: Rongwei Liu <rongweil@nvidia.com>
>
> Wrap the lag pf access into two new macros:
> 1. ldev_for_each()
> 2. ldev_for_each_reverse()
> The maximum number of lag ports and the index to `natvie_port_num`
> mapping will be handled by the two new macros.
> Users shouldn't use the for loop anymore.
[...]
> @@ -1417,6 +1398,26 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
> mlx5_queue_bond_work(ldev, 0);
> }
>
> +int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx)
> +{
> + int i;
> +
> + for (i = start_idx; i >= end_idx; i--)
> + if (ldev->pf[i].dev)
> + return i;
> + return -1;
> +}
> +
> +int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx)
> +{
> + int i;
> +
> + for (i = start_idx; i < MLX5_MAX_PORTS; i++)
> + if (ldev->pf[i].dev)
> + return i;
> + return MLX5_MAX_PORTS;
> +}
Why aren't these two prefixed with mlx5?
> +
> bool mlx5_lag_is_roce(struct mlx5_core_dev *dev)
> {
> struct mlx5_lag *ldev;
[...]
>
> +#define ldev_for_each(i, start_index, ldev) \
> + for (int tmp = start_index; tmp = get_next_ldev_func(ldev, tmp), \
> + i = tmp, tmp < MLX5_MAX_PORTS; tmp++)
> +
> +#define ldev_for_each_reverse(i, start_index, end_index, ldev) \
> + for (int tmp = start_index, tmp1 = end_index; \
> + tmp = get_pre_ldev_func(ldev, tmp, tmp1), \
> + i = tmp, tmp >= tmp1; tmp--)
Same?
Thanks,
Olek
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-16 17:55 ` Alexander Lobakin
@ 2024-12-17 5:44 ` rongwei liu
2024-12-17 11:32 ` Alexander Lobakin
0 siblings, 1 reply; 34+ messages in thread
From: rongwei liu @ 2024-12-17 5:44 UTC (permalink / raw)
To: Alexander Lobakin, Tariq Toukan
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Mark Bloch
On 2024/12/17 01:55, Alexander Lobakin wrote:
> From: Tariq Toukan <tariqt@nvidia.com>
> Date: Wed, 11 Dec 2024 15:42:13 +0200
>
>> From: Rongwei Liu <rongweil@nvidia.com>
>>
>> Wrap the lag pf access into two new macros:
>> 1. ldev_for_each()
>> 2. ldev_for_each_reverse()
>> The maximum number of lag ports and the index to `natvie_port_num`
>> mapping will be handled by the two new macros.
>> Users shouldn't use the for loop anymore.
>
> [...]
>
>> @@ -1417,6 +1398,26 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
>> mlx5_queue_bond_work(ldev, 0);
>> }
>>
>> +int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx)
>> +{
>> + int i;
>> +
>> + for (i = start_idx; i >= end_idx; i--)
>> + if (ldev->pf[i].dev)
>> + return i;
>> + return -1;
>> +}
>> +
>> +int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx)
>> +{
>> + int i;
>> +
>> + for (i = start_idx; i < MLX5_MAX_PORTS; i++)
>> + if (ldev->pf[i].dev)
>> + return i;
>> + return MLX5_MAX_PORTS;
>> +}
>
> Why aren't these two prefixed with mlx5?
> We can have. No mlx5 prefix aligns with "ldev_for_each/ldev_for_each_reverse()", simple, short and meaningful.
>> +
>> bool mlx5_lag_is_roce(struct mlx5_core_dev *dev)
>> {
>> struct mlx5_lag *ldev;
>
> [...]
>
>>
>> +#define ldev_for_each(i, start_index, ldev) \
>> + for (int tmp = start_index; tmp = get_next_ldev_func(ldev, tmp), \
>> + i = tmp, tmp < MLX5_MAX_PORTS; tmp++)
>> +
>> +#define ldev_for_each_reverse(i, start_index, end_index, ldev) \
>> + for (int tmp = start_index, tmp1 = end_index; \
>> + tmp = get_pre_ldev_func(ldev, tmp, tmp1), \
>> + i = tmp, tmp >= tmp1; tmp--)
>
> Same?
Reverse is used to the error handling. Add end index is more convenient.
Of course, we can remove the end_index.
But all the logic need to add:
if (i < end_index)
break;
If no strong comments, I would like to keep as now.
>
> Thanks,
> Olek
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-17 5:44 ` rongwei liu
@ 2024-12-17 11:32 ` Alexander Lobakin
2024-12-17 11:42 ` rongwei liu
2024-12-17 12:52 ` Mark Bloch
0 siblings, 2 replies; 34+ messages in thread
From: Alexander Lobakin @ 2024-12-17 11:32 UTC (permalink / raw)
To: rongwei liu
Cc: Tariq Toukan, David S. Miller, Jakub Kicinski, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Mark Bloch
From: Rongwei Liu <rongweil@nvidia.com>
Date: Tue, 17 Dec 2024 13:44:07 +0800
>
>
> On 2024/12/17 01:55, Alexander Lobakin wrote:
>> From: Tariq Toukan <tariqt@nvidia.com>
>> Date: Wed, 11 Dec 2024 15:42:13 +0200
>>
>>> From: Rongwei Liu <rongweil@nvidia.com>
>>>
>>> Wrap the lag pf access into two new macros:
>>> 1. ldev_for_each()
>>> 2. ldev_for_each_reverse()
>>> The maximum number of lag ports and the index to `natvie_port_num`
>>> mapping will be handled by the two new macros.
>>> Users shouldn't use the for loop anymore.
>>
>> [...]
>>
>>> @@ -1417,6 +1398,26 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
>>> mlx5_queue_bond_work(ldev, 0);
>>> }
>>>
>>> +int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx)
>>> +{
>>> + int i;
>>> +
>>> + for (i = start_idx; i >= end_idx; i--)
>>> + if (ldev->pf[i].dev)
>>> + return i;
>>> + return -1;
>>> +}
>>> +
>>> +int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx)
>>> +{
>>> + int i;
>>> +
>>> + for (i = start_idx; i < MLX5_MAX_PORTS; i++)
>>> + if (ldev->pf[i].dev)
>>> + return i;
>>> + return MLX5_MAX_PORTS;
>>> +}
>>
>> Why aren't these two prefixed with mlx5?
>> We can have. No mlx5 prefix aligns with "ldev_for_each/ldev_for_each_reverse()", simple, short and meaningful.
All drivers must have its symbols prefixed, otherwise there might be
name conflicts at anytime and also it's not clear where a definition
comes from if it's not prefixed.
>>> +
>>> bool mlx5_lag_is_roce(struct mlx5_core_dev *dev)
>>> {
>>> struct mlx5_lag *ldev;
>>
>> [...]
>>
>>>
>>> +#define ldev_for_each(i, start_index, ldev) \
>>> + for (int tmp = start_index; tmp = get_next_ldev_func(ldev, tmp), \
>>> + i = tmp, tmp < MLX5_MAX_PORTS; tmp++)
>>> +
>>> +#define ldev_for_each_reverse(i, start_index, end_index, ldev) \
>>> + for (int tmp = start_index, tmp1 = end_index; \
>>> + tmp = get_pre_ldev_func(ldev, tmp, tmp1), \
>>> + i = tmp, tmp >= tmp1; tmp--)
>>
>> Same?
> Reverse is used to the error handling. Add end index is more convenient.
> Of course, we can remove the end_index.
> But all the logic need to add:
> if (i < end_index)
> break;
> If no strong comments, I would like to keep as now.
By "same?" I meant that there two are also not prefixed with mlx5_, the
same as the two above.
Thanks,
Olek
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-17 11:32 ` Alexander Lobakin
@ 2024-12-17 11:42 ` rongwei liu
2024-12-17 12:52 ` Mark Bloch
1 sibling, 0 replies; 34+ messages in thread
From: rongwei liu @ 2024-12-17 11:42 UTC (permalink / raw)
To: Alexander Lobakin
Cc: Tariq Toukan, David S. Miller, Jakub Kicinski, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Mark Bloch
On 2024/12/17 19:32, Alexander Lobakin wrote:
> From: Rongwei Liu <rongweil@nvidia.com>
> Date: Tue, 17 Dec 2024 13:44:07 +0800
>
>>
>>
>> On 2024/12/17 01:55, Alexander Lobakin wrote:
>>> From: Tariq Toukan <tariqt@nvidia.com>
>>> Date: Wed, 11 Dec 2024 15:42:13 +0200
>>>
>>>> From: Rongwei Liu <rongweil@nvidia.com>
>>>>
>>>> Wrap the lag pf access into two new macros:
>>>> 1. ldev_for_each()
>>>> 2. ldev_for_each_reverse()
>>>> The maximum number of lag ports and the index to `natvie_port_num`
>>>> mapping will be handled by the two new macros.
>>>> Users shouldn't use the for loop anymore.
>>>
>>> [...]
>>>
>>>> @@ -1417,6 +1398,26 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
>>>> mlx5_queue_bond_work(ldev, 0);
>>>> }
>>>>
>>>> +int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx)
>>>> +{
>>>> + int i;
>>>> +
>>>> + for (i = start_idx; i >= end_idx; i--)
>>>> + if (ldev->pf[i].dev)
>>>> + return i;
>>>> + return -1;
>>>> +}
>>>> +
>>>> +int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx)
>>>> +{
>>>> + int i;
>>>> +
>>>> + for (i = start_idx; i < MLX5_MAX_PORTS; i++)
>>>> + if (ldev->pf[i].dev)
>>>> + return i;
>>>> + return MLX5_MAX_PORTS;
>>>> +}
>>>
>>> Why aren't these two prefixed with mlx5?
>>> We can have. No mlx5 prefix aligns with "ldev_for_each/ldev_for_each_reverse()", simple, short and meaningful.
>
> All drivers must have its symbols prefixed, otherwise there might be
> name conflicts at anytime and also it's not clear where a definition
> comes from if it's not prefixed.
>
ACK
>>>> +
>>>> bool mlx5_lag_is_roce(struct mlx5_core_dev *dev)
>>>> {
>>>> struct mlx5_lag *ldev;
>>>
>>> [...]
>>>
>>>>
>>>> +#define ldev_for_each(i, start_index, ldev) \
>>>> + for (int tmp = start_index; tmp = get_next_ldev_func(ldev, tmp), \
>>>> + i = tmp, tmp < MLX5_MAX_PORTS; tmp++)
>>>> +
>>>> +#define ldev_for_each_reverse(i, start_index, end_index, ldev) \
>>>> + for (int tmp = start_index, tmp1 = end_index; \
>>>> + tmp = get_pre_ldev_func(ldev, tmp, tmp1), \
>>>> + i = tmp, tmp >= tmp1; tmp--)
>>>
>>> Same?
>> Reverse is used to the error handling. Add end index is more convenient.
>> Of course, we can remove the end_index.
>> But all the logic need to add:
>> if (i < end_index)
>> break;
>> If no strong comments, I would like to keep as now.
>
> By "same?" I meant that there two are also not prefixed with mlx5_, the
> same as the two above.
ACK
>
> Thanks,
> Olek
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-17 11:32 ` Alexander Lobakin
2024-12-17 11:42 ` rongwei liu
@ 2024-12-17 12:52 ` Mark Bloch
2024-12-17 14:58 ` Jakub Kicinski
2024-12-17 15:03 ` Alexander Lobakin
1 sibling, 2 replies; 34+ messages in thread
From: Mark Bloch @ 2024-12-17 12:52 UTC (permalink / raw)
To: Alexander Lobakin, rongwei liu
Cc: Tariq Toukan, David S. Miller, Jakub Kicinski, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma
On 17/12/2024 13:32, Alexander Lobakin wrote:
> From: Rongwei Liu <rongweil@nvidia.com>
> Date: Tue, 17 Dec 2024 13:44:07 +0800
>
>>
>>
>> On 2024/12/17 01:55, Alexander Lobakin wrote:
>>> From: Tariq Toukan <tariqt@nvidia.com>
>>> Date: Wed, 11 Dec 2024 15:42:13 +0200
>>>
>>>> From: Rongwei Liu <rongweil@nvidia.com>
>>>>
>>>> Wrap the lag pf access into two new macros:
>>>> 1. ldev_for_each()
>>>> 2. ldev_for_each_reverse()
>>>> The maximum number of lag ports and the index to `natvie_port_num`
>>>> mapping will be handled by the two new macros.
>>>> Users shouldn't use the for loop anymore.
>>>
>>> [...]
>>>
>>>> @@ -1417,6 +1398,26 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
>>>> mlx5_queue_bond_work(ldev, 0);
>>>> }
>>>>
>>>> +int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx)
>>>> +{
>>>> + int i;
>>>> +
>>>> + for (i = start_idx; i >= end_idx; i--)
>>>> + if (ldev->pf[i].dev)
>>>> + return i;
>>>> + return -1;
>>>> +}
>>>> +
>>>> +int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx)
>>>> +{
>>>> + int i;
>>>> +
>>>> + for (i = start_idx; i < MLX5_MAX_PORTS; i++)
>>>> + if (ldev->pf[i].dev)
>>>> + return i;
>>>> + return MLX5_MAX_PORTS;
>>>> +}
>>>
>>> Why aren't these two prefixed with mlx5?
>>> We can have. No mlx5 prefix aligns with "ldev_for_each/ldev_for_each_reverse()", simple, short and meaningful.
>
> All drivers must have its symbols prefixed, otherwise there might be
> name conflicts at anytime and also it's not clear where a definition
> comes from if it's not prefixed.
>
However, those aren't exported symbols, they are used exclusively by the mlx5 lag code.
I don't see any added value in prefixing internal functions with mlx5 unless it adds
context to the logic.
Here it's very clear we are going over the members that are stored inside the ldev struct.
Mark
>>>> +
>>>> bool mlx5_lag_is_roce(struct mlx5_core_dev *dev)
>>>> {
>>>> struct mlx5_lag *ldev;
>>>
>>> [...]
>>>
>>>>
>>>> +#define ldev_for_each(i, start_index, ldev) \
>>>> + for (int tmp = start_index; tmp = get_next_ldev_func(ldev, tmp), \
>>>> + i = tmp, tmp < MLX5_MAX_PORTS; tmp++)
>>>> +
>>>> +#define ldev_for_each_reverse(i, start_index, end_index, ldev) \
>>>> + for (int tmp = start_index, tmp1 = end_index; \
>>>> + tmp = get_pre_ldev_func(ldev, tmp, tmp1), \
>>>> + i = tmp, tmp >= tmp1; tmp--)
>>>
>>> Same?
>> Reverse is used to the error handling. Add end index is more convenient.
>> Of course, we can remove the end_index.
>> But all the logic need to add:
>> if (i < end_index)
>> break;
>> If no strong comments, I would like to keep as now.
>
> By "same?" I meant that there two are also not prefixed with mlx5_, the
> same as the two above.
>
> Thanks,
> Olek
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-17 12:52 ` Mark Bloch
@ 2024-12-17 14:58 ` Jakub Kicinski
2024-12-17 15:47 ` Mark Bloch
2024-12-17 15:03 ` Alexander Lobakin
1 sibling, 1 reply; 34+ messages in thread
From: Jakub Kicinski @ 2024-12-17 14:58 UTC (permalink / raw)
To: Mark Bloch
Cc: Alexander Lobakin, rongwei liu, Tariq Toukan, David S. Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman
On Tue, 17 Dec 2024 14:52:55 +0200 Mark Bloch wrote:
> > All drivers must have its symbols prefixed, otherwise there might be
> > name conflicts at anytime and also it's not clear where a definition
> > comes from if it's not prefixed.
>
> However, those aren't exported symbols, they are used exclusively by the mlx5 lag code.
> I don't see any added value in prefixing internal functions with mlx5 unless it adds
> context to the logic.
> Here it's very clear we are going over the members that are stored inside the ldev struct.
Prefixing the symbols makes it easier to read your code for people
who aren't exclusively working on mlx5. Also useful when reading
mlx5-originated stack traces.
^ permalink raw reply [flat|nested] 34+ messages in thread
* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-17 14:58 ` Jakub Kicinski
@ 2024-12-17 15:47 ` Mark Bloch
0 siblings, 0 replies; 34+ messages in thread
From: Mark Bloch @ 2024-12-17 15:47 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Alexander Lobakin, rongwei liu, Tariq Toukan, David S. Miller,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman
On 17/12/2024 16:58, Jakub Kicinski wrote:
> On Tue, 17 Dec 2024 14:52:55 +0200 Mark Bloch wrote:
>>> All drivers must have its symbols prefixed, otherwise there might be
>>> name conflicts at anytime and also it's not clear where a definition
>>> comes from if it's not prefixed.
>>
>> However, those aren't exported symbols, they are used exclusively by the mlx5 lag code.
>> I don't see any added value in prefixing internal functions with mlx5 unless it adds
>> context to the logic.
>> Here it's very clear we are going over the members that are stored inside the ldev struct.
>
> Prefixing the symbols makes it easier to read your code for people
> who aren't exclusively working on mlx5. Also useful when reading
> mlx5-originated stack traces.
I agree with that, but since every entry point into mlx5 lag
already uses a prefixed function, any stack trace will clearly
show where the issue originates. I’m not strongly opposed to
this change, there’s already an internal version that addresses
these comments.
Just to note for example, a function name like txring_txq()
is fairly common with a generic name, and several drivers
use the exact same name, with fbnic being one such case.
I don't see a clear rule about this stuff.
Mark
^ permalink raw reply [flat|nested] 34+ messages in thread
* Re: [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic
2024-12-17 12:52 ` Mark Bloch
2024-12-17 14:58 ` Jakub Kicinski
@ 2024-12-17 15:03 ` Alexander Lobakin
1 sibling, 0 replies; 34+ messages in thread
From: Alexander Lobakin @ 2024-12-17 15:03 UTC (permalink / raw)
To: Mark Bloch
Cc: rongwei liu, Tariq Toukan, David S. Miller, Jakub Kicinski,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma
From: Mark Bloch <mbloch@nvidia.com>
Date: Tue, 17 Dec 2024 14:52:55 +0200
>
>
> On 17/12/2024 13:32, Alexander Lobakin wrote:
>> From: Rongwei Liu <rongweil@nvidia.com>
>> Date: Tue, 17 Dec 2024 13:44:07 +0800
>>
>>>
>>>
>>> On 2024/12/17 01:55, Alexander Lobakin wrote:
>>>> From: Tariq Toukan <tariqt@nvidia.com>
>>>> Date: Wed, 11 Dec 2024 15:42:13 +0200
>>>>
>>>>> From: Rongwei Liu <rongweil@nvidia.com>
>>>>>
>>>>> Wrap the lag pf access into two new macros:
>>>>> 1. ldev_for_each()
>>>>> 2. ldev_for_each_reverse()
>>>>> The maximum number of lag ports and the index to `natvie_port_num`
>>>>> mapping will be handled by the two new macros.
>>>>> Users shouldn't use the for loop anymore.
>>>>
>>>> [...]
>>>>
>>>>> @@ -1417,6 +1398,26 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
>>>>> mlx5_queue_bond_work(ldev, 0);
>>>>> }
>>>>>
>>>>> +int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx)
>>>>> +{
>>>>> + int i;
>>>>> +
>>>>> + for (i = start_idx; i >= end_idx; i--)
>>>>> + if (ldev->pf[i].dev)
>>>>> + return i;
>>>>> + return -1;
>>>>> +}
>>>>> +
>>>>> +int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx)
>>>>> +{
>>>>> + int i;
>>>>> +
>>>>> + for (i = start_idx; i < MLX5_MAX_PORTS; i++)
>>>>> + if (ldev->pf[i].dev)
>>>>> + return i;
>>>>> + return MLX5_MAX_PORTS;
>>>>> +}
>>>>
>>>> Why aren't these two prefixed with mlx5?
>>>> We can have. No mlx5 prefix aligns with "ldev_for_each/ldev_for_each_reverse()", simple, short and meaningful.
>>
>> All drivers must have its symbols prefixed, otherwise there might be
>> name conflicts at anytime and also it's not clear where a definition
>> comes from if it's not prefixed.
>>
>
> However, those aren't exported symbols, they are used exclusively by the mlx5 lag code.
> I don't see any added value in prefixing internal functions with mlx5 unless it adds
> context to the logic.
> Here it's very clear we are going over the members that are stored inside the ldev struct.
Tomorrow someone will add a function with the same name to the core
kernel code and get a compilation failure. OTOH nobody would add a
generic symbol prefixed with mlx5_ to the kernel. That's the value.
(+ what Jakub wrote)
>
> Mark
Thanks,
Olek
^ permalink raw reply [flat|nested] 34+ messages in thread
* [PATCH net-next 03/12] net/mlx5: LAG, Support LAG over Multi-Host NICs
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
2024-12-11 13:42 ` [PATCH mlx5-next 01/12] net/mlx5: Add device cap abs_native_port_num Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 02/12] net/mlx5: LAG, Refactor lag logic Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 04/12] net/mlx5: fs, add counter object to flow destination Tariq Toukan
` (8 subsequent siblings)
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Tariq Toukan,
Rongwei Liu
New multi-host NICs provide each host with partial ports,
allowing each host to maintain its unique LAG configuration.
On these multi-host NICs, the 'native_port_num' capability
is no longer continuous on each host and can exceed the
'num_lag_ports' capability. Therefore, it is necessary to
skip the PFs with ldev->pf[i].dev == NULL when querying/modifying
the lag devices' information.
There is no need to check dev.native_port_num against ldev->ports.
Signed-off-by: Rongwei Liu <rongweil@nvidia.com>
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/lag/lag.c | 200 ++++++++++++++----
.../net/ethernet/mellanox/mlx5/core/lag/lag.h | 3 +
.../net/ethernet/mellanox/mlx5/core/lag/mp.c | 55 +++--
.../ethernet/mellanox/mlx5/core/lag/mpesw.c | 6 +-
.../mellanox/mlx5/core/lag/port_sel.c | 41 +++-
.../net/ethernet/mellanox/mlx5/core/main.c | 4 +
6 files changed, 230 insertions(+), 79 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
index 73fd3f747f1a..85e773856c0a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
@@ -76,23 +76,30 @@ static u8 lag_active_port_bits(struct mlx5_lag *ldev)
return active_port;
}
-static int mlx5_cmd_create_lag(struct mlx5_core_dev *dev, u8 *ports, int mode,
- unsigned long flags)
+static int mlx5_cmd_create_lag(struct mlx5_core_dev *dev, struct mlx5_lag *ldev,
+ int mode, unsigned long flags)
{
bool fdb_sel_mode = test_bit(MLX5_LAG_MODE_FLAG_FDB_SEL_MODE_NATIVE,
&flags);
int port_sel_mode = get_port_sel_mode(mode, flags);
u32 in[MLX5_ST_SZ_DW(create_lag_in)] = {};
+ u8 *ports = ldev->v2p_map;
+ int idx0, idx1;
void *lag_ctx;
lag_ctx = MLX5_ADDR_OF(create_lag_in, in, ctx);
MLX5_SET(create_lag_in, in, opcode, MLX5_CMD_OP_CREATE_LAG);
MLX5_SET(lagc, lag_ctx, fdb_selection_mode, fdb_sel_mode);
+ idx0 = mlx5_lag_get_dev_index_by_seq(ldev, 0);
+ idx1 = mlx5_lag_get_dev_index_by_seq(ldev, 1);
+
+ if (idx0 < 0 || idx1 < 0)
+ return -EINVAL;
switch (port_sel_mode) {
case MLX5_LAG_PORT_SELECT_MODE_QUEUE_AFFINITY:
- MLX5_SET(lagc, lag_ctx, tx_remap_affinity_1, ports[0]);
- MLX5_SET(lagc, lag_ctx, tx_remap_affinity_2, ports[1]);
+ MLX5_SET(lagc, lag_ctx, tx_remap_affinity_1, ports[idx0]);
+ MLX5_SET(lagc, lag_ctx, tx_remap_affinity_2, ports[idx1]);
break;
case MLX5_LAG_PORT_SELECT_MODE_PORT_SELECT_FT:
if (!MLX5_CAP_PORT_SELECTION(dev, port_select_flow_table_bypass))
@@ -114,12 +121,18 @@ static int mlx5_cmd_modify_lag(struct mlx5_core_dev *dev, struct mlx5_lag *ldev,
{
u32 in[MLX5_ST_SZ_DW(modify_lag_in)] = {};
void *lag_ctx = MLX5_ADDR_OF(modify_lag_in, in, ctx);
+ int idx0, idx1;
+
+ idx0 = mlx5_lag_get_dev_index_by_seq(ldev, 0);
+ idx1 = mlx5_lag_get_dev_index_by_seq(ldev, 1);
+ if (idx0 < 0 || idx1 < 0)
+ return -EINVAL;
MLX5_SET(modify_lag_in, in, opcode, MLX5_CMD_OP_MODIFY_LAG);
MLX5_SET(modify_lag_in, in, field_select, 0x1);
- MLX5_SET(lagc, lag_ctx, tx_remap_affinity_1, ports[0]);
- MLX5_SET(lagc, lag_ctx, tx_remap_affinity_2, ports[1]);
+ MLX5_SET(lagc, lag_ctx, tx_remap_affinity_1, ports[idx0]);
+ MLX5_SET(lagc, lag_ctx, tx_remap_affinity_2, ports[idx1]);
return mlx5_cmd_exec_in(dev, modify_lag, in);
}
@@ -287,6 +300,48 @@ int mlx5_lag_dev_get_netdev_idx(struct mlx5_lag *ldev,
return -ENOENT;
}
+int mlx5_lag_get_dev_index_by_seq(struct mlx5_lag *ldev, int seq)
+{
+ int i, num = 0;
+
+ if (!ldev)
+ return -ENOENT;
+
+ ldev_for_each(i, 0, ldev) {
+ if (num == seq)
+ return i;
+ num++;
+ }
+ return -ENOENT;
+}
+
+int mlx5_lag_num_devs(struct mlx5_lag *ldev)
+{
+ int i, num = 0;
+
+ if (!ldev)
+ return 0;
+
+ ldev_for_each(i, 0, ldev) {
+ (void)i;
+ num++;
+ }
+ return num;
+}
+
+int mlx5_lag_num_netdevs(struct mlx5_lag *ldev)
+{
+ int i, num = 0;
+
+ if (!ldev)
+ return 0;
+
+ ldev_for_each(i, 0, ldev)
+ if (ldev->pf[i].netdev)
+ num++;
+ return num;
+}
+
static bool __mlx5_lag_is_roce(struct mlx5_lag *ldev)
{
return ldev->mode == MLX5_LAG_MODE_ROCE;
@@ -423,10 +478,15 @@ static int mlx5_cmd_modify_active_port(struct mlx5_core_dev *dev, u8 ports)
static int _mlx5_modify_lag(struct mlx5_lag *ldev, u8 *ports)
{
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ int idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ struct mlx5_core_dev *dev0;
u8 active_ports;
int ret;
+ if (idx < 0)
+ return -EINVAL;
+
+ dev0 = ldev->pf[idx].dev;
if (test_bit(MLX5_LAG_MODE_FLAG_HASH_BASED, &ldev->mode_flags)) {
ret = mlx5_lag_port_sel_modify(ldev, ports);
if (ret ||
@@ -445,7 +505,7 @@ static struct net_device *mlx5_lag_active_backup_get_netdev(struct mlx5_core_dev
struct net_device *ndev = NULL;
struct mlx5_lag *ldev;
unsigned long flags;
- int i;
+ int i, last_idx;
spin_lock_irqsave(&lag_lock, flags);
ldev = mlx5_lag_dev(dev);
@@ -456,8 +516,12 @@ static struct net_device *mlx5_lag_active_backup_get_netdev(struct mlx5_core_dev
ldev_for_each(i, 0, ldev)
if (ldev->tracker.netdev_state[i].tx_enabled)
ndev = ldev->pf[i].netdev;
- if (!ndev)
- ndev = ldev->pf[ldev->ports - 1].netdev;
+ if (!ndev) {
+ last_idx = mlx5_lag_get_dev_index_by_seq(ldev, ldev->ports - 1);
+ if (last_idx < 0)
+ goto unlock;
+ ndev = ldev->pf[last_idx].netdev;
+ }
if (ndev)
dev_hold(ndev);
@@ -471,13 +535,18 @@ static struct net_device *mlx5_lag_active_backup_get_netdev(struct mlx5_core_dev
void mlx5_modify_lag(struct mlx5_lag *ldev,
struct lag_tracker *tracker)
{
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
u8 ports[MLX5_MAX_PORTS * MLX5_LAG_MAX_HASH_BUCKETS] = {};
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ struct mlx5_core_dev *dev0;
int idx;
int err;
int i;
int j;
+ if (first_idx < 0)
+ return;
+
+ dev0 = ldev->pf[first_idx].dev;
mlx5_infer_tx_affinity_mapping(tracker, ldev, ldev->buckets, ports);
ldev_for_each(i, 0, ldev) {
@@ -518,8 +587,13 @@ void mlx5_modify_lag(struct mlx5_lag *ldev,
static int mlx5_lag_set_port_sel_mode_roce(struct mlx5_lag *ldev,
unsigned long *flags)
{
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ struct mlx5_core_dev *dev0;
+ if (first_idx < 0)
+ return -EINVAL;
+
+ dev0 = ldev->pf[first_idx].dev;
if (!MLX5_CAP_PORT_SELECTION(dev0, port_select_flow_table)) {
if (ldev->ports > 2)
return -EINVAL;
@@ -539,11 +613,13 @@ static void mlx5_lag_set_port_sel_mode_offloads(struct mlx5_lag *ldev,
enum mlx5_lag_mode mode,
unsigned long *flags)
{
- struct lag_func *dev0 = &ldev->pf[MLX5_LAG_P1];
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ struct lag_func *dev0;
- if (mode == MLX5_LAG_MODE_MPESW)
+ if (first_idx < 0 || mode == MLX5_LAG_MODE_MPESW)
return;
+ dev0 = &ldev->pf[first_idx];
if (MLX5_CAP_PORT_SELECTION(dev0->dev, port_select_flow_table) &&
tracker->tx_type == NETDEV_LAG_TX_TYPE_HASH) {
if (ldev->ports > 2)
@@ -588,12 +664,18 @@ char *mlx5_get_str_port_sel_mode(enum mlx5_lag_mode mode, unsigned long flags)
static int mlx5_lag_create_single_fdb(struct mlx5_lag *ldev)
{
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
- struct mlx5_eswitch *master_esw = dev0->priv.eswitch;
- int err;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ struct mlx5_eswitch *master_esw;
+ struct mlx5_core_dev *dev0;
int i, j;
+ int err;
- ldev_for_each(i, 1, ldev) {
+ if (first_idx < 0)
+ return -EINVAL;
+
+ dev0 = ldev->pf[first_idx].dev;
+ master_esw = dev0->priv.eswitch;
+ ldev_for_each(i, first_idx + 1, ldev) {
struct mlx5_eswitch *slave_esw = ldev->pf[i].dev->priv.eswitch;
err = mlx5_eswitch_offloads_single_fdb_add_one(master_esw,
@@ -603,7 +685,7 @@ static int mlx5_lag_create_single_fdb(struct mlx5_lag *ldev)
}
return 0;
err:
- ldev_for_each_reverse(j, i, 1, ldev)
+ ldev_for_each_reverse(j, i, first_idx + 1, ldev)
mlx5_eswitch_offloads_single_fdb_del_one(master_esw,
ldev->pf[j].dev->priv.eswitch);
return err;
@@ -615,16 +697,21 @@ static int mlx5_create_lag(struct mlx5_lag *ldev,
unsigned long flags)
{
bool shared_fdb = test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags);
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
u32 in[MLX5_ST_SZ_DW(destroy_lag_in)] = {};
+ struct mlx5_core_dev *dev0;
int err;
+ if (first_idx < 0)
+ return -EINVAL;
+
+ dev0 = ldev->pf[first_idx].dev;
if (tracker)
mlx5_lag_print_mapping(dev0, ldev, tracker, flags);
mlx5_core_info(dev0, "shared_fdb:%d mode:%s\n",
shared_fdb, mlx5_get_str_port_sel_mode(mode, flags));
- err = mlx5_cmd_create_lag(dev0, ldev->v2p_map, mode, flags);
+ err = mlx5_cmd_create_lag(dev0, ldev, mode, flags);
if (err) {
mlx5_core_err(dev0,
"Failed to create LAG (%d)\n",
@@ -656,11 +743,16 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
enum mlx5_lag_mode mode,
bool shared_fdb)
{
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
bool roce_lag = mode == MLX5_LAG_MODE_ROCE;
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ struct mlx5_core_dev *dev0;
unsigned long flags = 0;
int err;
+ if (first_idx < 0)
+ return -EINVAL;
+
+ dev0 = ldev->pf[first_idx].dev;
err = mlx5_lag_set_flags(ldev, mode, tracker, shared_fdb, &flags);
if (err)
return err;
@@ -704,20 +796,26 @@ int mlx5_activate_lag(struct mlx5_lag *ldev,
int mlx5_deactivate_lag(struct mlx5_lag *ldev)
{
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
- struct mlx5_eswitch *master_esw = dev0->priv.eswitch;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
u32 in[MLX5_ST_SZ_DW(destroy_lag_in)] = {};
bool roce_lag = __mlx5_lag_is_roce(ldev);
unsigned long flags = ldev->mode_flags;
+ struct mlx5_eswitch *master_esw;
+ struct mlx5_core_dev *dev0;
int err;
int i;
+ if (first_idx < 0)
+ return -EINVAL;
+
+ dev0 = ldev->pf[first_idx].dev;
+ master_esw = dev0->priv.eswitch;
ldev->mode = MLX5_LAG_MODE_NONE;
ldev->mode_flags = 0;
mlx5_lag_mp_reset(ldev);
if (test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags)) {
- ldev_for_each(i, 1, ldev)
+ ldev_for_each(i, first_idx + 1, ldev)
mlx5_eswitch_offloads_single_fdb_del_one(master_esw,
ldev->pf[i].dev->priv.eswitch);
clear_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &flags);
@@ -749,6 +847,7 @@ int mlx5_deactivate_lag(struct mlx5_lag *ldev)
bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
{
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
#ifdef CONFIG_MLX5_ESWITCH
struct mlx5_core_dev *dev;
u8 mode;
@@ -756,9 +855,8 @@ bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
bool roce_support;
int i;
- for (i = 0; i < ldev->ports; i++)
- if (!ldev->pf[i].dev)
- return false;
+ if (first_idx < 0 || mlx5_lag_num_devs(ldev) != ldev->ports)
+ return false;
#ifdef CONFIG_MLX5_ESWITCH
ldev_for_each(i, 0, ldev) {
@@ -767,7 +865,7 @@ bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
return false;
}
- dev = ldev->pf[MLX5_LAG_P1].dev;
+ dev = ldev->pf[first_idx].dev;
mode = mlx5_eswitch_mode(dev);
ldev_for_each(i, 0, ldev)
if (mlx5_eswitch_mode(ldev->pf[i].dev) != mode)
@@ -778,8 +876,8 @@ bool mlx5_lag_check_prereq(struct mlx5_lag *ldev)
if (mlx5_sriov_is_enabled(ldev->pf[i].dev))
return false;
#endif
- roce_support = mlx5_get_roce_state(ldev->pf[MLX5_LAG_P1].dev);
- ldev_for_each(i, MLX5_LAG_P2, ldev)
+ roce_support = mlx5_get_roce_state(ldev->pf[first_idx].dev);
+ ldev_for_each(i, first_idx + 1, ldev)
if (mlx5_get_roce_state(ldev->pf[i].dev) != roce_support)
return false;
@@ -817,11 +915,16 @@ void mlx5_lag_remove_devices(struct mlx5_lag *ldev)
void mlx5_disable_lag(struct mlx5_lag *ldev)
{
bool shared_fdb = test_bit(MLX5_LAG_MODE_FLAG_SHARED_FDB, &ldev->mode_flags);
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ int idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ struct mlx5_core_dev *dev0;
bool roce_lag;
int err;
int i;
+ if (idx < 0)
+ return;
+
+ dev0 = ldev->pf[idx].dev;
roce_lag = __mlx5_lag_is_roce(ldev);
if (shared_fdb) {
@@ -831,7 +934,7 @@ void mlx5_disable_lag(struct mlx5_lag *ldev)
dev0->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
mlx5_rescan_drivers_locked(dev0);
}
- ldev_for_each(i, MLX5_LAG_P2, ldev)
+ ldev_for_each(i, idx + 1, ldev)
mlx5_nic_vport_disable_roce(ldev->pf[i].dev);
}
@@ -850,10 +953,14 @@ void mlx5_disable_lag(struct mlx5_lag *ldev)
static bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev)
{
+ int idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
struct mlx5_core_dev *dev;
int i;
- ldev_for_each(i, MLX5_LAG_P1 + 1, ldev) {
+ if (idx < 0)
+ return false;
+
+ ldev_for_each(i, idx + 1, ldev) {
dev = ldev->pf[i].dev;
if (is_mdev_switchdev_mode(dev) &&
mlx5_eswitch_vport_match_metadata_enabled(dev->priv.eswitch) &&
@@ -865,7 +972,7 @@ static bool mlx5_shared_fdb_supported(struct mlx5_lag *ldev)
return false;
}
- dev = ldev->pf[MLX5_LAG_P1].dev;
+ dev = ldev->pf[idx].dev;
if (is_mdev_switchdev_mode(dev) &&
mlx5_eswitch_vport_match_metadata_enabled(dev->priv.eswitch) &&
mlx5_esw_offloads_devcom_is_ready(dev->priv.eswitch) &&
@@ -906,13 +1013,18 @@ static bool mlx5_lag_should_disable_lag(struct mlx5_lag *ldev, bool do_bond)
static void mlx5_do_bond(struct mlx5_lag *ldev)
{
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ int idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
struct lag_tracker tracker = { };
+ struct mlx5_core_dev *dev0;
struct net_device *ndev;
bool do_bond, roce_lag;
int err;
int i;
+ if (idx < 0)
+ return;
+
+ dev0 = ldev->pf[idx].dev;
if (!mlx5_lag_is_ready(ldev)) {
do_bond = false;
} else {
@@ -945,7 +1057,7 @@ static void mlx5_do_bond(struct mlx5_lag *ldev)
} else if (roce_lag) {
dev0->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
mlx5_rescan_drivers_locked(dev0);
- ldev_for_each(i, MLX5_LAG_P2, ldev) {
+ ldev_for_each(i, idx + 1, ldev) {
if (mlx5_get_roce_state(ldev->pf[i].dev))
mlx5_nic_vport_enable_roce(ldev->pf[i].dev);
}
@@ -1380,7 +1492,7 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
struct net_device *netdev)
{
struct mlx5_lag *ldev;
- int i;
+ int num = 0;
ldev = mlx5_lag_dev(dev);
if (!ldev)
@@ -1388,11 +1500,8 @@ void mlx5_lag_add_netdev(struct mlx5_core_dev *dev,
mutex_lock(&ldev->lock);
mlx5_ldev_add_netdev(ldev, dev, netdev);
- for (i = 0; i < ldev->ports; i++)
- if (!ldev->pf[i].netdev)
- break;
-
- if (i >= ldev->ports)
+ num = mlx5_lag_num_netdevs(ldev);
+ if (num >= ldev->ports)
set_bit(MLX5_LAG_FLAG_NDEVS_READY, &ldev->state_flags);
mutex_unlock(&ldev->lock);
mlx5_queue_bond_work(ldev, 0);
@@ -1469,11 +1578,12 @@ bool mlx5_lag_is_master(struct mlx5_core_dev *dev)
struct mlx5_lag *ldev;
unsigned long flags;
bool res = false;
+ int idx;
spin_lock_irqsave(&lag_lock, flags);
ldev = mlx5_lag_dev(dev);
- res = ldev && __mlx5_lag_is_active(ldev) &&
- dev == ldev->pf[MLX5_LAG_P1].dev;
+ idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ res = ldev && __mlx5_lag_is_active(ldev) && idx >= 0 && dev == ldev->pf[idx].dev;
spin_unlock_irqrestore(&lag_lock, flags);
return res;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
index 1be34eb43723..345f29aff77a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.h
@@ -136,4 +136,7 @@ static inline bool mlx5_lag_is_supported(struct mlx5_core_dev *dev)
int get_pre_ldev_func(struct mlx5_lag *ldev, int start_idx, int end_idx);
int get_next_ldev_func(struct mlx5_lag *ldev, int start_idx);
+int mlx5_lag_get_dev_index_by_seq(struct mlx5_lag *ldev, int seq);
+int mlx5_lag_num_devs(struct mlx5_lag *ldev);
+int mlx5_lag_num_netdevs(struct mlx5_lag *ldev);
#endif /* __MLX5_LAG_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
index 9596cf433815..9b278c5be5e9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mp.c
@@ -17,7 +17,10 @@ static bool __mlx5_lag_is_multipath(struct mlx5_lag *ldev)
#define MLX5_LAG_MULTIPATH_OFFLOADS_SUPPORTED_PORTS 2
static bool mlx5_lag_multipath_check_prereq(struct mlx5_lag *ldev)
{
- if (!mlx5_lag_is_ready(ldev))
+ int idx0 = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ int idx1 = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P2);
+
+ if (idx0 < 0 || idx1 < 0 || !mlx5_lag_is_ready(ldev))
return false;
if (__mlx5_lag_is_active(ldev) && !__mlx5_lag_is_multipath(ldev))
@@ -26,8 +29,8 @@ static bool mlx5_lag_multipath_check_prereq(struct mlx5_lag *ldev)
if (ldev->ports > MLX5_LAG_MULTIPATH_OFFLOADS_SUPPORTED_PORTS)
return false;
- return mlx5_esw_multipath_prereq(ldev->pf[MLX5_LAG_P1].dev,
- ldev->pf[MLX5_LAG_P2].dev);
+ return mlx5_esw_multipath_prereq(ldev->pf[idx0].dev,
+ ldev->pf[idx1].dev);
}
bool mlx5_lag_is_multipath(struct mlx5_core_dev *dev)
@@ -50,43 +53,45 @@ bool mlx5_lag_is_multipath(struct mlx5_core_dev *dev)
static void mlx5_lag_set_port_affinity(struct mlx5_lag *ldev,
enum mlx5_lag_port_affinity port)
{
+ int idx0 = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ int idx1 = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P2);
struct lag_tracker tracker = {};
- if (!__mlx5_lag_is_multipath(ldev))
+ if (idx0 < 0 || idx1 < 0 || !__mlx5_lag_is_multipath(ldev))
return;
switch (port) {
case MLX5_LAG_NORMAL_AFFINITY:
- tracker.netdev_state[MLX5_LAG_P1].tx_enabled = true;
- tracker.netdev_state[MLX5_LAG_P2].tx_enabled = true;
- tracker.netdev_state[MLX5_LAG_P1].link_up = true;
- tracker.netdev_state[MLX5_LAG_P2].link_up = true;
+ tracker.netdev_state[idx0].tx_enabled = true;
+ tracker.netdev_state[idx1].tx_enabled = true;
+ tracker.netdev_state[idx0].link_up = true;
+ tracker.netdev_state[idx1].link_up = true;
break;
case MLX5_LAG_P1_AFFINITY:
- tracker.netdev_state[MLX5_LAG_P1].tx_enabled = true;
- tracker.netdev_state[MLX5_LAG_P1].link_up = true;
- tracker.netdev_state[MLX5_LAG_P2].tx_enabled = false;
- tracker.netdev_state[MLX5_LAG_P2].link_up = false;
+ tracker.netdev_state[idx0].tx_enabled = true;
+ tracker.netdev_state[idx0].link_up = true;
+ tracker.netdev_state[idx1].tx_enabled = false;
+ tracker.netdev_state[idx1].link_up = false;
break;
case MLX5_LAG_P2_AFFINITY:
- tracker.netdev_state[MLX5_LAG_P1].tx_enabled = false;
- tracker.netdev_state[MLX5_LAG_P1].link_up = false;
- tracker.netdev_state[MLX5_LAG_P2].tx_enabled = true;
- tracker.netdev_state[MLX5_LAG_P2].link_up = true;
+ tracker.netdev_state[idx0].tx_enabled = false;
+ tracker.netdev_state[idx0].link_up = false;
+ tracker.netdev_state[idx1].tx_enabled = true;
+ tracker.netdev_state[idx1].link_up = true;
break;
default:
- mlx5_core_warn(ldev->pf[MLX5_LAG_P1].dev,
+ mlx5_core_warn(ldev->pf[idx0].dev,
"Invalid affinity port %d", port);
return;
}
- if (tracker.netdev_state[MLX5_LAG_P1].tx_enabled)
- mlx5_notifier_call_chain(ldev->pf[MLX5_LAG_P1].dev->priv.events,
+ if (tracker.netdev_state[idx0].tx_enabled)
+ mlx5_notifier_call_chain(ldev->pf[idx0].dev->priv.events,
MLX5_DEV_EVENT_PORT_AFFINITY,
(void *)0);
- if (tracker.netdev_state[MLX5_LAG_P2].tx_enabled)
- mlx5_notifier_call_chain(ldev->pf[MLX5_LAG_P2].dev->priv.events,
+ if (tracker.netdev_state[idx1].tx_enabled)
+ mlx5_notifier_call_chain(ldev->pf[idx1].dev->priv.events,
MLX5_DEV_EVENT_PORT_AFFINITY,
(void *)0);
@@ -150,11 +155,15 @@ mlx5_lag_get_next_fib_dev(struct mlx5_lag *ldev,
static void mlx5_lag_fib_route_event(struct mlx5_lag *ldev, unsigned long event,
struct fib_entry_notifier_info *fen_info)
{
+ int idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
struct net_device *nh_dev0, *nh_dev1;
struct fib_info *fi = fen_info->fi;
struct lag_mp *mp = &ldev->lag_mp;
int i, dev_idx = 0;
+ if (idx < 0)
+ return;
+
/* Handle delete event */
if (event == FIB_EVENT_ENTRY_DEL) {
/* stop track */
@@ -180,14 +189,14 @@ static void mlx5_lag_fib_route_event(struct mlx5_lag *ldev, unsigned long event,
}
if (nh_dev0 == nh_dev1) {
- mlx5_core_warn(ldev->pf[MLX5_LAG_P1].dev,
+ mlx5_core_warn(ldev->pf[idx].dev,
"Multipath offload doesn't support routes with multiple nexthops of the same device");
return;
}
if (!nh_dev1) {
if (__mlx5_lag_is_active(ldev)) {
- ldev_for_each(i, 0, ldev) {
+ ldev_for_each(i, idx, ldev) {
dev_idx++;
if (ldev->pf[i].netdev == nh_dev0)
break;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index dd1b2caa0182..1e968abe5ab7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -68,13 +68,15 @@ static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
#define MLX5_LAG_MPESW_OFFLOADS_SUPPORTED_PORTS 4
static int enable_mpesw(struct mlx5_lag *ldev)
{
- struct mlx5_core_dev *dev0 = ldev->pf[MLX5_LAG_P1].dev;
+ int idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ struct mlx5_core_dev *dev0;
int err;
int i;
- if (ldev->mode != MLX5_LAG_MODE_NONE)
+ if (idx < 0 || ldev->mode != MLX5_LAG_MODE_NONE)
return -EINVAL;
+ dev0 = ldev->pf[idx].dev;
if (ldev->ports > MLX5_LAG_MPESW_OFFLOADS_SUPPORTED_PORTS)
return -EOPNOTSUPP;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
index 6b52b09ffc40..34b507eadd2d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
@@ -39,13 +39,18 @@ static int mlx5_lag_create_port_sel_table(struct mlx5_lag *ldev,
struct mlx5_lag_definer *lag_definer,
u8 *ports)
{
- struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
struct mlx5_flow_table_attr ft_attr = {};
struct mlx5_flow_destination dest = {};
MLX5_DECLARE_FLOW_ACT(flow_act);
struct mlx5_flow_namespace *ns;
+ struct mlx5_core_dev *dev;
int err, i, j, k, idx;
+ if (first_idx < 0)
+ return -EINVAL;
+
+ dev = ldev->pf[first_idx].dev;
ft_attr.max_fte = ldev->ports * ldev->buckets;
ft_attr.level = MLX5_LAG_FT_LEVEL_DEFINER;
@@ -293,11 +298,16 @@ static struct mlx5_lag_definer *
mlx5_lag_create_definer(struct mlx5_lag *ldev, enum netdev_lag_hash hash,
enum mlx5_traffic_types tt, bool tunnel, u8 *ports)
{
- struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
struct mlx5_lag_definer *lag_definer;
+ struct mlx5_core_dev *dev;
u32 *match_definer_mask;
int format_id, err;
+ if (first_idx < 0)
+ return ERR_PTR(-EINVAL);
+
+ dev = ldev->pf[first_idx].dev;
lag_definer = kzalloc(sizeof(*lag_definer), GFP_KERNEL);
if (!lag_definer)
return ERR_PTR(-ENOMEM);
@@ -339,12 +349,15 @@ mlx5_lag_create_definer(struct mlx5_lag *ldev, enum netdev_lag_hash hash,
static void mlx5_lag_destroy_definer(struct mlx5_lag *ldev,
struct mlx5_lag_definer *lag_definer)
{
- struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
- int idx;
- int i;
- int j;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+ struct mlx5_core_dev *dev;
+ int idx, i, j;
- ldev_for_each(i, 0, ldev) {
+ if (first_idx < 0)
+ return;
+
+ dev = ldev->pf[first_idx].dev;
+ ldev_for_each(i, first_idx, ldev) {
for (j = 0; j < ldev->buckets; j++) {
idx = i * ldev->buckets + j;
mlx5_del_flow_rules(lag_definer->rules[idx]);
@@ -499,10 +512,15 @@ static void mlx5_lag_set_outer_ttc_params(struct mlx5_lag *ldev,
static int mlx5_lag_create_ttc_table(struct mlx5_lag *ldev)
{
- struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
struct ttc_params ttc_params = {};
+ struct mlx5_core_dev *dev;
+ if (first_idx < 0)
+ return -EINVAL;
+
+ dev = ldev->pf[first_idx].dev;
mlx5_lag_set_outer_ttc_params(ldev, &ttc_params);
port_sel->outer.ttc = mlx5_create_ttc_table(dev, &ttc_params);
return PTR_ERR_OR_ZERO(port_sel->outer.ttc);
@@ -510,10 +528,15 @@ static int mlx5_lag_create_ttc_table(struct mlx5_lag *ldev)
static int mlx5_lag_create_inner_ttc_table(struct mlx5_lag *ldev)
{
- struct mlx5_core_dev *dev = ldev->pf[MLX5_LAG_P1].dev;
+ int first_idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
struct mlx5_lag_port_sel *port_sel = &ldev->port_sel;
struct ttc_params ttc_params = {};
+ struct mlx5_core_dev *dev;
+
+ if (first_idx < 0)
+ return -EINVAL;
+ dev = ldev->pf[first_idx].dev;
mlx5_lag_set_inner_ttc_params(ldev, &ttc_params);
port_sel->inner.ttc = mlx5_create_inner_ttc_table(dev, &ttc_params);
return PTR_ERR_OR_ZERO(port_sel->inner.ttc);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index 220a9ac75c8b..869bfecdd8ff 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -664,6 +664,10 @@ static int handle_hca_cap(struct mlx5_core_dev *dev, void *set_ctx)
MLX5_SET(cmd_hca_cap, set_hca_cap, log_max_current_uc_list,
ilog2(max_uc_list));
+ /* enable absolute native port num */
+ if (MLX5_CAP_GEN_MAX(dev, abs_native_port_num))
+ MLX5_SET(cmd_hca_cap, set_hca_cap, abs_native_port_num, 1);
+
return set_caps(dev, set_ctx, MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE);
}
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* [PATCH net-next 04/12] net/mlx5: fs, add counter object to flow destination
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (2 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 03/12] net/mlx5: LAG, Support LAG over Multi-Host NICs Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-12 17:20 ` Simon Horman
2024-12-11 13:42 ` [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API Tariq Toukan
` (7 subsequent siblings)
11 siblings, 1 reply; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Moshe Shemesh,
Yevgeny Kliteynik, Mark Bloch, Tariq Toukan
From: Moshe Shemesh <moshe@nvidia.com>
Currently mlx5_flow_destination includes counter_id which is assigned in
case we use flow counter on the flow steering rule. However, counter_id
is not enough data in case of using HW Steering. Thus, have mlx5_fc
object as part of mlx5_flow_destination instead of counter_id and assign
it where needed.
In case counter_id is received from user space, create a local counter
object to represent it.
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
drivers/infiniband/hw/mlx5/fs.c | 37 ++++++++++----
.../mellanox/mlx5/core/diag/fs_tracepoint.h | 2 +-
.../mellanox/mlx5/core/en_accel/ipsec_fs.c | 20 ++++----
.../net/ethernet/mellanox/mlx5/core/en_tc.c | 2 +-
.../mellanox/mlx5/core/esw/acl/egress_lgcy.c | 2 +-
.../mellanox/mlx5/core/esw/acl/ingress_lgcy.c | 2 +-
.../ethernet/mellanox/mlx5/core/esw/bridge.c | 20 ++++----
.../mellanox/mlx5/core/eswitch_offloads.c | 2 +-
.../net/ethernet/mellanox/mlx5/core/fs_cmd.c | 2 +-
.../net/ethernet/mellanox/mlx5/core/fs_core.c | 1 +
.../ethernet/mellanox/mlx5/core/fs_counters.c | 50 +++++++++++++++++++
.../mellanox/mlx5/core/lib/macsec_fs.c | 8 +--
.../mellanox/mlx5/core/steering/sws/fs_dr.c | 2 +-
include/linux/mlx5/fs.h | 4 +-
14 files changed, 112 insertions(+), 42 deletions(-)
diff --git a/drivers/infiniband/hw/mlx5/fs.c b/drivers/infiniband/hw/mlx5/fs.c
index 520034acf73a..162814ae8cb4 100644
--- a/drivers/infiniband/hw/mlx5/fs.c
+++ b/drivers/infiniband/hw/mlx5/fs.c
@@ -943,7 +943,7 @@ int mlx5_ib_fs_add_op_fc(struct mlx5_ib_dev *dev, u32 port_num,
}
dst.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dst.counter_id = mlx5_fc_id(opfc->fc);
+ dst.counter = opfc->fc;
flow_act.action =
MLX5_FLOW_CONTEXT_ACTION_COUNT | MLX5_FLOW_CONTEXT_ACTION_ALLOW;
@@ -1113,8 +1113,8 @@ static struct mlx5_ib_flow_handler *_create_flow_rule(struct mlx5_ib_dev *dev,
handler->ibcounters = flow_act.counters;
dest_arr[dest_num].type =
MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest_arr[dest_num].counter_id =
- mlx5_fc_id(mcounters->hw_cntrs_hndl);
+ dest_arr[dest_num].counter =
+ mcounters->hw_cntrs_hndl;
dest_num++;
}
@@ -1603,7 +1603,7 @@ static bool raw_fs_is_multicast(struct mlx5_ib_flow_matcher *fs_matcher,
static struct mlx5_ib_flow_handler *raw_fs_rule_add(
struct mlx5_ib_dev *dev, struct mlx5_ib_flow_matcher *fs_matcher,
struct mlx5_flow_context *flow_context, struct mlx5_flow_act *flow_act,
- u32 counter_id, void *cmd_in, int inlen, int dest_id, int dest_type)
+ struct mlx5_fc *counter, void *cmd_in, int inlen, int dest_id, int dest_type)
{
struct mlx5_flow_destination *dst;
struct mlx5_ib_flow_prio *ft_prio;
@@ -1652,8 +1652,12 @@ static struct mlx5_ib_flow_handler *raw_fs_rule_add(
}
if (flow_act->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
+ if (WARN_ON(!counter)) {
+ err = -EINVAL;
+ goto unlock;
+ }
dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dst[dst_num].counter_id = counter_id;
+ dst[dst_num].counter = counter;
dst_num++;
}
@@ -1878,7 +1882,8 @@ static int get_dests(struct uverbs_attr_bundle *attrs,
return 0;
}
-static bool is_flow_counter(void *obj, u32 offset, u32 *counter_id)
+static bool
+is_flow_counter(void *obj, u32 offset, u32 *counter_id, u32 *fc_bulk_size)
{
struct devx_obj *devx_obj = obj;
u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, opcode);
@@ -1888,6 +1893,7 @@ static bool is_flow_counter(void *obj, u32 offset, u32 *counter_id)
if (offset && offset >= devx_obj->flow_counter_bulk_size)
return false;
+ *fc_bulk_size = devx_obj->flow_counter_bulk_size;
*counter_id = MLX5_GET(dealloc_flow_counter_in,
devx_obj->dinbox,
flow_counter_id);
@@ -1904,13 +1910,13 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)(
{
struct mlx5_flow_context flow_context = {.flow_tag =
MLX5_FS_DEFAULT_FLOW_TAG};
- u32 *offset_attr, offset = 0, counter_id = 0;
int dest_id, dest_type = -1, inlen, len, ret, i;
struct mlx5_ib_flow_handler *flow_handler;
struct mlx5_ib_flow_matcher *fs_matcher;
struct ib_uobject **arr_flow_actions;
struct ib_uflow_resources *uflow_res;
struct mlx5_flow_act flow_act = {};
+ struct mlx5_fc *counter = NULL;
struct ib_qp *qp = NULL;
void *devx_obj, *cmd_in;
struct ib_uobject *uobj;
@@ -1937,6 +1943,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)(
len = uverbs_attr_get_uobjs_arr(attrs,
MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, &arr_flow_actions);
if (len) {
+ u32 *offset_attr, fc_bulk_size, offset = 0, counter_id = 0;
devx_obj = arr_flow_actions[0]->object;
if (uverbs_attr_is_valid(attrs,
@@ -1956,8 +1963,11 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)(
offset = *offset_attr;
}
- if (!is_flow_counter(devx_obj, offset, &counter_id))
+ if (!is_flow_counter(devx_obj, offset, &counter_id, &fc_bulk_size))
return -EINVAL;
+ counter = mlx5_fc_local_create(counter_id, offset, fc_bulk_size);
+ if (IS_ERR(counter))
+ return PTR_ERR(counter);
flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
}
@@ -1968,8 +1978,10 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)(
MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE);
uflow_res = flow_resources_alloc(MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS);
- if (!uflow_res)
- return -ENOMEM;
+ if (!uflow_res) {
+ ret = -ENOMEM;
+ goto destroy_counter;
+ }
len = uverbs_attr_get_uobjs_arr(attrs,
MLX5_IB_ATTR_CREATE_FLOW_ARR_FLOW_ACTIONS, &arr_flow_actions);
@@ -1996,7 +2008,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)(
flow_handler =
raw_fs_rule_add(dev, fs_matcher, &flow_context, &flow_act,
- counter_id, cmd_in, inlen, dest_id, dest_type);
+ counter, cmd_in, inlen, dest_id, dest_type);
if (IS_ERR(flow_handler)) {
ret = PTR_ERR(flow_handler);
goto err_out;
@@ -2007,6 +2019,9 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)(
return 0;
err_out:
ib_uverbs_flow_resources_free(uflow_res);
+destroy_counter:
+ if (counter)
+ mlx5_fc_local_destroy(counter);
return ret;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
index 9aed29fa4900..d6e736c1fb24 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fs_tracepoint.h
@@ -292,7 +292,7 @@ TRACE_EVENT(mlx5_fs_add_rule,
if (rule->dest_attr.type &
MLX5_FLOW_DESTINATION_TYPE_COUNTER)
__entry->counter_id =
- rule->dest_attr.counter_id;
+ mlx5_fc_id(rule->dest_attr.counter);
),
TP_printk("rule=%p fte=%p index=%u sw_action=<%s> [dst] %s\n",
__entry->rule, __entry->fte, __entry->index,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
index e51b03d4c717..687bd95d2c3e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_fs.c
@@ -194,7 +194,7 @@ static int rx_add_rule_drop_auth_trailer(struct mlx5e_ipsec_sa_entry *sa_entry,
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
flow_act.flags = FLOW_ACT_NO_APPEND;
dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest.counter_id = mlx5_fc_id(flow_counter);
+ dest.counter = flow_counter;
if (rx == ipsec->rx_esw)
spec->flow_context.flow_source = MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK;
@@ -223,7 +223,7 @@ static int rx_add_rule_drop_auth_trailer(struct mlx5e_ipsec_sa_entry *sa_entry,
}
sa_entry->ipsec_rule.trailer.fc = flow_counter;
- dest.counter_id = mlx5_fc_id(flow_counter);
+ dest.counter = flow_counter;
MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.ipsec_syndrome, 2);
rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
@@ -275,7 +275,7 @@ static int rx_add_rule_drop_replay(struct mlx5e_ipsec_sa_entry *sa_entry, struct
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
flow_act.flags = FLOW_ACT_NO_APPEND;
dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest.counter_id = mlx5_fc_id(flow_counter);
+ dest.counter = flow_counter;
if (rx == ipsec->rx_esw)
spec->flow_context.flow_source = MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK;
@@ -348,7 +348,7 @@ static int ipsec_rx_status_drop_all_create(struct mlx5e_ipsec *ipsec,
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest.counter_id = mlx5_fc_id(flow_counter);
+ dest.counter = flow_counter;
if (rx == ipsec->rx_esw)
spec->flow_context.flow_source = MLX5_FLOW_CONTEXT_FLOW_SOURCE_UPLINK;
rule = mlx5_add_flow_rules(ft, spec, &flow_act, &dest, 1);
@@ -686,7 +686,7 @@ static int rx_create(struct mlx5_core_dev *mdev, struct mlx5e_ipsec *ipsec,
rx->ft.status = ft;
dest[1].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest[1].counter_id = mlx5_fc_id(rx->fc->cnt);
+ dest[1].counter = rx->fc->cnt;
err = mlx5_ipsec_rx_status_create(ipsec, rx, dest);
if (err)
goto err_add;
@@ -873,7 +873,7 @@ static int ipsec_counter_rule_tx(struct mlx5_core_dev *mdev, struct mlx5e_ipsec_
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW |
MLX5_FLOW_CONTEXT_ACTION_COUNT;
dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest.counter_id = mlx5_fc_id(tx->fc->cnt);
+ dest.counter = tx->fc->cnt;
fte = mlx5_add_flow_rules(tx->ft.status, spec, &flow_act, &dest, 1);
if (IS_ERR(fte)) {
err = PTR_ERR(fte);
@@ -1649,7 +1649,7 @@ static int rx_add_rule(struct mlx5e_ipsec_sa_entry *sa_entry)
dest[0].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
dest[0].ft = rx->ft.status;
dest[1].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest[1].counter_id = mlx5_fc_id(counter);
+ dest[1].counter = counter;
rule = mlx5_add_flow_rules(rx->ft.sa, spec, &flow_act, dest, 2);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
@@ -1762,7 +1762,7 @@ static int tx_add_rule(struct mlx5e_ipsec_sa_entry *sa_entry)
dest[0].ft = tx->ft.status;
dest[0].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
dest[1].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest[1].counter_id = mlx5_fc_id(counter);
+ dest[1].counter = counter;
rule = mlx5_add_flow_rules(tx->ft.sa, spec, &flow_act, dest, 2);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
@@ -1835,7 +1835,7 @@ static int tx_add_policy(struct mlx5e_ipsec_pol_entry *pol_entry)
flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_DROP |
MLX5_FLOW_CONTEXT_ACTION_COUNT;
dest[dstn].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest[dstn].counter_id = mlx5_fc_id(tx->fc->drop);
+ dest[dstn].counter = tx->fc->drop;
dstn++;
break;
default:
@@ -1913,7 +1913,7 @@ static int rx_add_policy(struct mlx5e_ipsec_pol_entry *pol_entry)
case XFRM_POLICY_BLOCK:
flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
dest[dstn].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest[dstn].counter_id = mlx5_fc_id(rx->fc->drop);
+ dest[dstn].counter = rx->fc->drop;
dstn++;
break;
default:
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 6b3b1afe8312..9ba99609999f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -1282,7 +1282,7 @@ mlx5e_add_offloaded_nic_rule(struct mlx5e_priv *priv,
if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
dest[dest_ix].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest[dest_ix].counter_id = mlx5_fc_id(attr->counter);
+ dest[dest_ix].counter = attr->counter;
dest_ix++;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c
index 6b4c9ffad95b..7dd1dc3f77c7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c
@@ -135,7 +135,7 @@ int esw_acl_egress_lgcy_setup(struct mlx5_eswitch *esw,
if (drop_counter) {
flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
drop_ctr_dst.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- drop_ctr_dst.counter_id = mlx5_fc_id(drop_counter);
+ drop_ctr_dst.counter = drop_counter;
dst = &drop_ctr_dst;
dest_num++;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_lgcy.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_lgcy.c
index 093ed86a0acd..1c37098e09ea 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_lgcy.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/acl/ingress_lgcy.c
@@ -260,7 +260,7 @@ int esw_acl_ingress_lgcy_setup(struct mlx5_eswitch *esw,
if (counter) {
flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT;
drop_ctr_dst.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- drop_ctr_dst.counter_id = mlx5_fc_id(counter);
+ drop_ctr_dst.counter = counter;
dst = &drop_ctr_dst;
dest_num++;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c
index c5ea1d1d2b03..5f647358a05c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/bridge.c
@@ -570,7 +570,8 @@ mlx5_esw_bridge_egress_table_cleanup(struct mlx5_esw_bridge *bridge)
static struct mlx5_flow_handle *
mlx5_esw_bridge_ingress_flow_with_esw_create(u16 vport_num, const unsigned char *addr,
- struct mlx5_esw_bridge_vlan *vlan, u32 counter_id,
+ struct mlx5_esw_bridge_vlan *vlan,
+ struct mlx5_fc *counter,
struct mlx5_esw_bridge *bridge,
struct mlx5_eswitch *esw)
{
@@ -628,7 +629,7 @@ mlx5_esw_bridge_ingress_flow_with_esw_create(u16 vport_num, const unsigned char
dests[0].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
dests[0].ft = bridge->egress_ft;
dests[1].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dests[1].counter_id = counter_id;
+ dests[1].counter = counter;
handle = mlx5_add_flow_rules(br_offloads->ingress_ft, rule_spec, &flow_act, dests,
ARRAY_SIZE(dests));
@@ -639,17 +640,19 @@ mlx5_esw_bridge_ingress_flow_with_esw_create(u16 vport_num, const unsigned char
static struct mlx5_flow_handle *
mlx5_esw_bridge_ingress_flow_create(u16 vport_num, const unsigned char *addr,
- struct mlx5_esw_bridge_vlan *vlan, u32 counter_id,
+ struct mlx5_esw_bridge_vlan *vlan,
+ struct mlx5_fc *counter,
struct mlx5_esw_bridge *bridge)
{
- return mlx5_esw_bridge_ingress_flow_with_esw_create(vport_num, addr, vlan, counter_id,
+ return mlx5_esw_bridge_ingress_flow_with_esw_create(vport_num, addr, vlan, counter,
bridge, bridge->br_offloads->esw);
}
static struct mlx5_flow_handle *
mlx5_esw_bridge_ingress_flow_peer_create(u16 vport_num, u16 esw_owner_vhca_id,
const unsigned char *addr,
- struct mlx5_esw_bridge_vlan *vlan, u32 counter_id,
+ struct mlx5_esw_bridge_vlan *vlan,
+ struct mlx5_fc *counter,
struct mlx5_esw_bridge *bridge)
{
struct mlx5_devcom_comp_dev *devcom = bridge->br_offloads->esw->devcom, *pos;
@@ -671,7 +674,7 @@ mlx5_esw_bridge_ingress_flow_peer_create(u16 vport_num, u16 esw_owner_vhca_id,
goto out;
}
- handle = mlx5_esw_bridge_ingress_flow_with_esw_create(vport_num, addr, vlan, counter_id,
+ handle = mlx5_esw_bridge_ingress_flow_with_esw_create(vport_num, addr, vlan, counter,
bridge, peer_esw);
out:
@@ -1385,10 +1388,9 @@ mlx5_esw_bridge_fdb_entry_init(struct net_device *dev, u16 vport_num, u16 esw_ow
handle = peer ?
mlx5_esw_bridge_ingress_flow_peer_create(vport_num, esw_owner_vhca_id,
- addr, vlan, mlx5_fc_id(counter),
- bridge) :
+ addr, vlan, counter, bridge) :
mlx5_esw_bridge_ingress_flow_create(vport_num, addr, vlan,
- mlx5_fc_id(counter), bridge);
+ counter, bridge);
if (IS_ERR(handle)) {
err = PTR_ERR(handle);
esw_warn(esw->dev, "Failed to create ingress flow(vport=%u,err=%d,peer=%d)\n",
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index d5b42b3a19fd..8636f0485800 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -724,7 +724,7 @@ mlx5_eswitch_add_offloaded_rule(struct mlx5_eswitch *esw,
if (flow_act.action & MLX5_FLOW_CONTEXT_ACTION_COUNT) {
dest[i].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest[i].counter_id = mlx5_fc_id(attr->counter);
+ dest[i].counter = attr->counter;
i++;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
index 676005854dad..6bf0aade69d7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
@@ -718,7 +718,7 @@ static int mlx5_cmd_set_fte(struct mlx5_core_dev *dev,
continue;
MLX5_SET(flow_counter_list, in_dests, flow_counter_id,
- dst->dest_attr.counter_id);
+ mlx5_fc_id(dst->dest_attr.counter));
in_dests += dst_cnt_size;
list_size++;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index 2eabfcc247c6..f781f8f169b9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -658,6 +658,7 @@ static void del_sw_hw_rule(struct fs_node *node)
BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_ACTION) |
BIT(MLX5_SET_FTE_MODIFY_ENABLE_MASK_FLOW_COUNTERS);
fte->act_dests.action.action &= ~MLX5_FLOW_CONTEXT_ACTION_COUNT;
+ mlx5_fc_local_destroy(rule->dest_attr.counter);
goto out;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
index 62d0c689796b..8cd5e274d6d3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
@@ -43,6 +43,11 @@
#define MLX5_FC_POOL_MAX_THRESHOLD BIT(18)
#define MLX5_FC_POOL_USED_BUFF_RATIO 10
+enum mlx5_fc_type {
+ MLX5_FC_TYPE_ACQUIRED = 0,
+ MLX5_FC_TYPE_LOCAL,
+};
+
struct mlx5_fc_cache {
u64 packets;
u64 bytes;
@@ -52,6 +57,7 @@ struct mlx5_fc_cache {
struct mlx5_fc {
u32 id;
bool aging;
+ enum mlx5_fc_type type;
struct mlx5_fc_bulk *bulk;
struct mlx5_fc_cache cache;
/* last{packets,bytes} are used for calculating deltas since last reading. */
@@ -186,6 +192,9 @@ static void mlx5_fc_release(struct mlx5_core_dev *dev, struct mlx5_fc *counter)
{
struct mlx5_fc_stats *fc_stats = dev->priv.fc_stats;
+ if (WARN_ON(counter->type == MLX5_FC_TYPE_LOCAL))
+ return;
+
if (counter->bulk)
mlx5_fc_pool_release_counter(&fc_stats->fc_pool, counter);
else
@@ -536,6 +545,47 @@ static int mlx5_fc_bulk_release_fc(struct mlx5_fc_bulk *bulk, struct mlx5_fc *fc
return 0;
}
+/**
+ * mlx5_fc_local_create - Allocate mlx5_fc struct for a counter which
+ * was already acquired using its counter id and bulk data.
+ *
+ * @counter_id: counter acquired counter id
+ * @offset: counter offset from bulk base
+ * @bulk_size: counter's bulk size as was allocated
+ */
+struct mlx5_fc *
+mlx5_fc_local_create(u32 counter_id, u32 offset, u32 bulk_size)
+{
+ struct mlx5_fc_bulk *bulk;
+ struct mlx5_fc *counter;
+
+ counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+ if (!counter)
+ return ERR_PTR(-ENOMEM);
+ bulk = kzalloc(sizeof(*bulk), GFP_KERNEL);
+ if (!bulk) {
+ kfree(counter);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ counter->type = MLX5_FC_TYPE_LOCAL;
+ counter->id = counter_id;
+ bulk->base_id = counter_id - offset;
+ bulk->bulk_len = bulk_size;
+ return counter;
+}
+EXPORT_SYMBOL(mlx5_fc_local_create);
+
+void mlx5_fc_local_destroy(struct mlx5_fc *counter)
+{
+ if (!counter || counter->type != MLX5_FC_TYPE_LOCAL)
+ return;
+
+ kfree(counter->bulk);
+ kfree(counter);
+}
+EXPORT_SYMBOL(mlx5_fc_local_destroy);
+
/* Flow counters pool API */
static void mlx5_fc_pool_init(struct mlx5_fc_pool *fc_pool, struct mlx5_core_dev *dev)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
index 4a078113e292..762d55ba9e51 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -497,7 +497,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
memset(&dest, 0, sizeof(struct mlx5_flow_destination));
memset(&flow_act, 0, sizeof(flow_act));
dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest.counter_id = mlx5_fc_id(tx_tables->check_miss_rule_counter);
+ dest.counter = tx_tables->check_miss_rule_counter;
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
rule = mlx5_add_flow_rules(tx_tables->ft_check, NULL, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
@@ -519,7 +519,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
flow_act.flags = FLOW_ACT_NO_APPEND;
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW | MLX5_FLOW_CONTEXT_ACTION_COUNT;
dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest.counter_id = mlx5_fc_id(tx_tables->check_rule_counter);
+ dest.counter = tx_tables->check_rule_counter;
rule = mlx5_add_flow_rules(tx_tables->ft_check, spec, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
@@ -1200,7 +1200,7 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs
flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT |
MLX5_FLOW_CONTEXT_ACTION_COUNT;
roce_dest[dstn].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- roce_dest[dstn].counter_id = mlx5_fc_id(rx_tables->check_rule_counter);
+ roce_dest[dstn].counter = rx_tables->check_rule_counter;
rule = mlx5_add_flow_rules(rx_tables->ft_check, spec, flow_act, roce_dest, dstn + 1);
if (IS_ERR(rule)) {
@@ -1592,7 +1592,7 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
memset(&flow_act, 0, sizeof(flow_act));
dest.type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest.counter_id = mlx5_fc_id(rx_tables->check_miss_rule_counter);
+ dest.counter = rx_tables->check_miss_rule_counter;
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP | MLX5_FLOW_CONTEXT_ACTION_COUNT;
rule = mlx5_add_flow_rules(rx_tables->ft_check, NULL, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/fs_dr.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/fs_dr.c
index 4b349d4005e4..8007d3f523c9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/fs_dr.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/fs_dr.c
@@ -521,7 +521,7 @@ static int mlx5_cmd_dr_create_fte(struct mlx5_flow_root_namespace *ns,
goto free_actions;
}
- id = dst->dest_attr.counter_id;
+ id = mlx5_fc_id(dst->dest_attr.counter);
tmp_action =
mlx5dr_action_create_flow_counter(id);
if (!tmp_action) {
diff --git a/include/linux/mlx5/fs.h b/include/linux/mlx5/fs.h
index 438db888bde0..2a69d9d71276 100644
--- a/include/linux/mlx5/fs.h
+++ b/include/linux/mlx5/fs.h
@@ -163,7 +163,7 @@ struct mlx5_flow_destination {
u32 tir_num;
u32 ft_num;
struct mlx5_flow_table *ft;
- u32 counter_id;
+ struct mlx5_fc *counter;
struct {
u16 num;
u16 vhca_id;
@@ -299,6 +299,8 @@ int mlx5_modify_rule_destination(struct mlx5_flow_handle *handler,
struct mlx5_fc *mlx5_fc_create(struct mlx5_core_dev *dev, bool aging);
void mlx5_fc_destroy(struct mlx5_core_dev *dev, struct mlx5_fc *counter);
+struct mlx5_fc *mlx5_fc_local_create(u32 counter_id, u32 offset, u32 bulk_size);
+void mlx5_fc_local_destroy(struct mlx5_fc *counter);
u64 mlx5_fc_query_lastuse(struct mlx5_fc *counter);
void mlx5_fc_query_cached(struct mlx5_fc *counter,
u64 *bytes, u64 *packets, u64 *lastuse);
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* Re: [PATCH net-next 04/12] net/mlx5: fs, add counter object to flow destination
2024-12-11 13:42 ` [PATCH net-next 04/12] net/mlx5: fs, add counter object to flow destination Tariq Toukan
@ 2024-12-12 17:20 ` Simon Horman
2024-12-12 18:32 ` Tariq Toukan
0 siblings, 1 reply; 34+ messages in thread
From: Simon Horman @ 2024-12-12 17:20 UTC (permalink / raw)
To: Tariq Toukan
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Moshe Shemesh, Yevgeny Kliteynik,
Mark Bloch
On Wed, Dec 11, 2024 at 03:42:15PM +0200, Tariq Toukan wrote:
> From: Moshe Shemesh <moshe@nvidia.com>
>
> Currently mlx5_flow_destination includes counter_id which is assigned in
> case we use flow counter on the flow steering rule. However, counter_id
> is not enough data in case of using HW Steering. Thus, have mlx5_fc
> object as part of mlx5_flow_destination instead of counter_id and assign
> it where needed.
>
> In case counter_id is received from user space, create a local counter
> object to represent it.
>
> Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> Reviewed-by: Mark Bloch <mbloch@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Unfortunately, I think that this misses two counter_id instances
in mlx5_vnet.c and the following is needed:
diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
index 5f581e71e201..36099047560d 100644
--- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
+++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
@@ -1952,7 +1952,7 @@ static int mlx5_vdpa_add_mac_vlan_rules(struct mlx5_vdpa_net *ndev, u8 *mac,
goto out_free;
#if defined(CONFIG_MLX5_VDPA_STEERING_DEBUG)
- dests[1].counter_id = mlx5_fc_id(node->ucast_counter.counter);
+ dests[1].counter = node->ucast_counter.counter;
#endif
node->ucast_rule = mlx5_add_flow_rules(ndev->rxft, spec, &flow_act, dests, NUM_DESTS);
if (IS_ERR(node->ucast_rule)) {
@@ -1961,7 +1961,7 @@ static int mlx5_vdpa_add_mac_vlan_rules(struct mlx5_vdpa_net *ndev, u8 *mac,
}
#if defined(CONFIG_MLX5_VDPA_STEERING_DEBUG)
- dests[1].counter_id = mlx5_fc_id(node->mcast_counter.counter);
+ dests[1].counter = node->mcast_counter.counter;
#endif
memset(dmac_c, 0, ETH_ALEN);
You can observe this with an allmodconfig build.
Also, please consider including a "Returns:" section in
the Kernel doc of mlx5_fc_local_create().
--
pw-bot: changes-requested
^ permalink raw reply related [flat|nested] 34+ messages in thread* Re: [PATCH net-next 04/12] net/mlx5: fs, add counter object to flow destination
2024-12-12 17:20 ` Simon Horman
@ 2024-12-12 18:32 ` Tariq Toukan
0 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-12 18:32 UTC (permalink / raw)
To: Simon Horman, Tariq Toukan
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Moshe Shemesh, Yevgeny Kliteynik,
Mark Bloch
On 12/12/2024 19:20, Simon Horman wrote:
> On Wed, Dec 11, 2024 at 03:42:15PM +0200, Tariq Toukan wrote:
>> From: Moshe Shemesh <moshe@nvidia.com>
>>
>> Currently mlx5_flow_destination includes counter_id which is assigned in
>> case we use flow counter on the flow steering rule. However, counter_id
>> is not enough data in case of using HW Steering. Thus, have mlx5_fc
>> object as part of mlx5_flow_destination instead of counter_id and assign
>> it where needed.
>>
>> In case counter_id is received from user space, create a local counter
>> object to represent it.
>>
>> Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
>> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
>> Reviewed-by: Mark Bloch <mbloch@nvidia.com>
>> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
>
> Unfortunately, I think that this misses two counter_id instances
> in mlx5_vnet.c and the following is needed:
>
> diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> index 5f581e71e201..36099047560d 100644
> --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> @@ -1952,7 +1952,7 @@ static int mlx5_vdpa_add_mac_vlan_rules(struct mlx5_vdpa_net *ndev, u8 *mac,
> goto out_free;
>
> #if defined(CONFIG_MLX5_VDPA_STEERING_DEBUG)
> - dests[1].counter_id = mlx5_fc_id(node->ucast_counter.counter);
> + dests[1].counter = node->ucast_counter.counter;
> #endif
> node->ucast_rule = mlx5_add_flow_rules(ndev->rxft, spec, &flow_act, dests, NUM_DESTS);
> if (IS_ERR(node->ucast_rule)) {
> @@ -1961,7 +1961,7 @@ static int mlx5_vdpa_add_mac_vlan_rules(struct mlx5_vdpa_net *ndev, u8 *mac,
> }
>
> #if defined(CONFIG_MLX5_VDPA_STEERING_DEBUG)
> - dests[1].counter_id = mlx5_fc_id(node->mcast_counter.counter);
> + dests[1].counter = node->mcast_counter.counter;
> #endif
>
> memset(dmac_c, 0, ETH_ALEN);
>
> You can observe this with an allmodconfig build.
>
Thanks, will fix.
>
> Also, please consider including a "Returns:" section in
> the Kernel doc of mlx5_fc_local_create().
>
I'll add.
^ permalink raw reply [flat|nested] 34+ messages in thread
* [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (3 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 04/12] net/mlx5: fs, add counter object to flow destination Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-12 17:23 ` Simon Horman
2024-12-11 13:42 ` [PATCH net-next 06/12] net/mlx5: fs, retry insertion to hash table on EBUSY Tariq Toukan
` (6 subsequent siblings)
11 siblings, 1 reply; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Moshe Shemesh,
Yevgeny Kliteynik, Mark Bloch, Tariq Toukan
From: Moshe Shemesh <moshe@nvidia.com>
Refactor fc_pool API to create generic fs_pool API, as HW steering has
more flow steering elements which can take advantage of the same pool of
bulks API. Change fs_counters code to use the fs_pool API.
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/Makefile | 2 +-
.../ethernet/mellanox/mlx5/core/fs_counters.c | 294 +++++-------------
.../net/ethernet/mellanox/mlx5/core/fs_pool.c | 194 ++++++++++++
.../net/ethernet/mellanox/mlx5/core/fs_pool.h | 54 ++++
4 files changed, 331 insertions(+), 213 deletions(-)
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
index be3d0876c521..79fe09de0a9f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -17,7 +17,7 @@ mlx5_core-y := main.o cmd.o debugfs.o fw.o eq.o uar.o pagealloc.o \
fs_counters.o fs_ft_pool.o rl.o lag/debugfs.o lag/lag.o dev.o events.o wq.o lib/gid.o \
lib/devcom.o lib/pci_vsc.o lib/dm.o lib/fs_ttc.o diag/fs_tracepoint.o \
diag/fw_tracer.o diag/crdump.o devlink.o diag/rsc_dump.o diag/reporter_vnic.o \
- fw_reset.o qos.o lib/tout.o lib/aso.o wc.o
+ fw_reset.o qos.o lib/tout.o lib/aso.o wc.o fs_pool.o
#
# Netdev basic
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
index 8cd5e274d6d3..2a167a2fd25d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
@@ -34,6 +34,7 @@
#include <linux/mlx5/fs.h>
#include "mlx5_core.h"
#include "fs_core.h"
+#include "fs_pool.h"
#include "fs_cmd.h"
#define MLX5_FC_STATS_PERIOD msecs_to_jiffies(1000)
@@ -65,17 +66,6 @@ struct mlx5_fc {
u64 lastbytes;
};
-struct mlx5_fc_pool {
- struct mlx5_core_dev *dev;
- struct mutex pool_lock; /* protects pool lists */
- struct list_head fully_used;
- struct list_head partially_used;
- struct list_head unused;
- int available_fcs;
- int used_fcs;
- int threshold;
-};
-
struct mlx5_fc_stats {
struct xarray counters;
@@ -86,13 +76,13 @@ struct mlx5_fc_stats {
int bulk_query_len;
bool bulk_query_alloc_failed;
unsigned long next_bulk_query_alloc;
- struct mlx5_fc_pool fc_pool;
+ struct mlx5_fs_pool fc_pool;
};
-static void mlx5_fc_pool_init(struct mlx5_fc_pool *fc_pool, struct mlx5_core_dev *dev);
-static void mlx5_fc_pool_cleanup(struct mlx5_fc_pool *fc_pool);
-static struct mlx5_fc *mlx5_fc_pool_acquire_counter(struct mlx5_fc_pool *fc_pool);
-static void mlx5_fc_pool_release_counter(struct mlx5_fc_pool *fc_pool, struct mlx5_fc *fc);
+static void mlx5_fc_pool_init(struct mlx5_fs_pool *fc_pool, struct mlx5_core_dev *dev);
+static void mlx5_fc_pool_cleanup(struct mlx5_fs_pool *fc_pool);
+static struct mlx5_fc *mlx5_fc_pool_acquire_counter(struct mlx5_fs_pool *fc_pool);
+static void mlx5_fc_pool_release_counter(struct mlx5_fs_pool *fc_pool, struct mlx5_fc *fc);
static int get_init_bulk_query_len(struct mlx5_core_dev *dev)
{
@@ -447,11 +437,9 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
/* Flow counter bluks */
struct mlx5_fc_bulk {
- struct list_head pool_list;
+ struct mlx5_fs_bulk fs_bulk;
u32 base_id;
- int bulk_len;
- unsigned long *bitmask;
- struct mlx5_fc fcs[] __counted_by(bulk_len);
+ struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
};
static void mlx5_fc_init(struct mlx5_fc *counter, struct mlx5_fc_bulk *bulk,
@@ -461,16 +449,10 @@ static void mlx5_fc_init(struct mlx5_fc *counter, struct mlx5_fc_bulk *bulk,
counter->id = id;
}
-static int mlx5_fc_bulk_get_free_fcs_amount(struct mlx5_fc_bulk *bulk)
-{
- return bitmap_weight(bulk->bitmask, bulk->bulk_len);
-}
-
-static struct mlx5_fc_bulk *mlx5_fc_bulk_create(struct mlx5_core_dev *dev)
+static struct mlx5_fs_bulk *mlx5_fc_bulk_create(struct mlx5_core_dev *dev)
{
enum mlx5_fc_bulk_alloc_bitmask alloc_bitmask;
- struct mlx5_fc_bulk *bulk;
- int err = -ENOMEM;
+ struct mlx5_fc_bulk *fc_bulk;
int bulk_len;
u32 base_id;
int i;
@@ -478,71 +460,97 @@ static struct mlx5_fc_bulk *mlx5_fc_bulk_create(struct mlx5_core_dev *dev)
alloc_bitmask = MLX5_CAP_GEN(dev, flow_counter_bulk_alloc);
bulk_len = alloc_bitmask > 0 ? MLX5_FC_BULK_NUM_FCS(alloc_bitmask) : 1;
- bulk = kvzalloc(struct_size(bulk, fcs, bulk_len), GFP_KERNEL);
- if (!bulk)
- goto err_alloc_bulk;
+ fc_bulk = kvzalloc(struct_size(fc_bulk, fcs, bulk_len), GFP_KERNEL);
+ if (!fc_bulk)
+ return NULL;
- bulk->bitmask = kvcalloc(BITS_TO_LONGS(bulk_len), sizeof(unsigned long),
- GFP_KERNEL);
- if (!bulk->bitmask)
- goto err_alloc_bitmask;
+ if (mlx5_fs_bulk_init(dev, &fc_bulk->fs_bulk, bulk_len))
+ goto err_fs_bulk_init;
- err = mlx5_cmd_fc_bulk_alloc(dev, alloc_bitmask, &base_id);
- if (err)
- goto err_mlx5_cmd_bulk_alloc;
+ if (mlx5_cmd_fc_bulk_alloc(dev, alloc_bitmask, &base_id))
+ goto err_cmd_bulk_alloc;
+ fc_bulk->base_id = base_id;
+ for (i = 0; i < bulk_len; i++)
+ mlx5_fc_init(&fc_bulk->fcs[i], fc_bulk, base_id + i);
- bulk->base_id = base_id;
- bulk->bulk_len = bulk_len;
- for (i = 0; i < bulk_len; i++) {
- mlx5_fc_init(&bulk->fcs[i], bulk, base_id + i);
- set_bit(i, bulk->bitmask);
- }
+ return &fc_bulk->fs_bulk;
- return bulk;
-
-err_mlx5_cmd_bulk_alloc:
- kvfree(bulk->bitmask);
-err_alloc_bitmask:
- kvfree(bulk);
-err_alloc_bulk:
- return ERR_PTR(err);
+err_cmd_bulk_alloc:
+ mlx5_fs_bulk_cleanup(&fc_bulk->fs_bulk);
+err_fs_bulk_init:
+ kvfree(fc_bulk);
+ return NULL;
}
static int
-mlx5_fc_bulk_destroy(struct mlx5_core_dev *dev, struct mlx5_fc_bulk *bulk)
+mlx5_fc_bulk_destroy(struct mlx5_core_dev *dev, struct mlx5_fs_bulk *fs_bulk)
{
- if (mlx5_fc_bulk_get_free_fcs_amount(bulk) < bulk->bulk_len) {
+ struct mlx5_fc_bulk *fc_bulk = container_of(fs_bulk,
+ struct mlx5_fc_bulk,
+ fs_bulk);
+
+ if (mlx5_fs_bulk_get_free_amount(fs_bulk) < fs_bulk->bulk_len) {
mlx5_core_err(dev, "Freeing bulk before all counters were released\n");
return -EBUSY;
}
- mlx5_cmd_fc_free(dev, bulk->base_id);
- kvfree(bulk->bitmask);
- kvfree(bulk);
+ mlx5_cmd_fc_free(dev, fc_bulk->base_id);
+ mlx5_fs_bulk_cleanup(fs_bulk);
+ kvfree(fc_bulk);
return 0;
}
-static struct mlx5_fc *mlx5_fc_bulk_acquire_fc(struct mlx5_fc_bulk *bulk)
+static void mlx5_fc_pool_update_threshold(struct mlx5_fs_pool *fc_pool)
{
- int free_fc_index = find_first_bit(bulk->bitmask, bulk->bulk_len);
+ fc_pool->threshold = min_t(int, MLX5_FC_POOL_MAX_THRESHOLD,
+ fc_pool->used_units / MLX5_FC_POOL_USED_BUFF_RATIO);
+}
- if (free_fc_index >= bulk->bulk_len)
- return ERR_PTR(-ENOSPC);
+/* Flow counters pool API */
+
+static const struct mlx5_fs_pool_ops mlx5_fc_pool_ops = {
+ .bulk_destroy = mlx5_fc_bulk_destroy,
+ .bulk_create = mlx5_fc_bulk_create,
+ .update_threshold = mlx5_fc_pool_update_threshold,
+};
- clear_bit(free_fc_index, bulk->bitmask);
- return &bulk->fcs[free_fc_index];
+static void
+mlx5_fc_pool_init(struct mlx5_fs_pool *fc_pool, struct mlx5_core_dev *dev)
+{
+ mlx5_fs_pool_init(fc_pool, dev, &mlx5_fc_pool_ops);
}
-static int mlx5_fc_bulk_release_fc(struct mlx5_fc_bulk *bulk, struct mlx5_fc *fc)
+static void mlx5_fc_pool_cleanup(struct mlx5_fs_pool *fc_pool)
{
- int fc_index = fc->id - bulk->base_id;
+ mlx5_fs_pool_cleanup(fc_pool);
+}
- if (test_bit(fc_index, bulk->bitmask))
- return -EINVAL;
+static struct mlx5_fc *
+mlx5_fc_pool_acquire_counter(struct mlx5_fs_pool *fc_pool)
+{
+ struct mlx5_fs_pool_index pool_index = {};
+ struct mlx5_fc_bulk *fc_bulk;
+ int err;
- set_bit(fc_index, bulk->bitmask);
- return 0;
+ err = mlx5_fs_pool_acquire_index(fc_pool, &pool_index);
+ if (err)
+ return ERR_PTR(err);
+ fc_bulk = container_of(pool_index.fs_bulk, struct mlx5_fc_bulk, fs_bulk);
+ return &fc_bulk->fcs[pool_index.index];
+}
+
+static void
+mlx5_fc_pool_release_counter(struct mlx5_fs_pool *fc_pool, struct mlx5_fc *fc)
+{
+ struct mlx5_fs_bulk *fs_bulk = &fc->bulk->fs_bulk;
+ struct mlx5_fs_pool_index pool_index = {};
+ struct mlx5_core_dev *dev = fc_pool->dev;
+
+ pool_index.fs_bulk = fs_bulk;
+ pool_index.index = fc->id - fc->bulk->base_id;
+ if (mlx5_fs_pool_release_index(fc_pool, &pool_index))
+ mlx5_core_warn(dev, "Attempted to release a counter which is not acquired\n");
}
/**
@@ -556,22 +564,22 @@ static int mlx5_fc_bulk_release_fc(struct mlx5_fc_bulk *bulk, struct mlx5_fc *fc
struct mlx5_fc *
mlx5_fc_local_create(u32 counter_id, u32 offset, u32 bulk_size)
{
- struct mlx5_fc_bulk *bulk;
+ struct mlx5_fc_bulk *fc_bulk;
struct mlx5_fc *counter;
counter = kzalloc(sizeof(*counter), GFP_KERNEL);
if (!counter)
return ERR_PTR(-ENOMEM);
- bulk = kzalloc(sizeof(*bulk), GFP_KERNEL);
- if (!bulk) {
+ fc_bulk = kzalloc(sizeof(*fc_bulk), GFP_KERNEL);
+ if (!fc_bulk) {
kfree(counter);
return ERR_PTR(-ENOMEM);
}
counter->type = MLX5_FC_TYPE_LOCAL;
counter->id = counter_id;
- bulk->base_id = counter_id - offset;
- bulk->bulk_len = bulk_size;
+ fc_bulk->base_id = counter_id - offset;
+ fc_bulk->fs_bulk.bulk_len = bulk_size;
return counter;
}
EXPORT_SYMBOL(mlx5_fc_local_create);
@@ -585,141 +593,3 @@ void mlx5_fc_local_destroy(struct mlx5_fc *counter)
kfree(counter);
}
EXPORT_SYMBOL(mlx5_fc_local_destroy);
-
-/* Flow counters pool API */
-
-static void mlx5_fc_pool_init(struct mlx5_fc_pool *fc_pool, struct mlx5_core_dev *dev)
-{
- fc_pool->dev = dev;
- mutex_init(&fc_pool->pool_lock);
- INIT_LIST_HEAD(&fc_pool->fully_used);
- INIT_LIST_HEAD(&fc_pool->partially_used);
- INIT_LIST_HEAD(&fc_pool->unused);
- fc_pool->available_fcs = 0;
- fc_pool->used_fcs = 0;
- fc_pool->threshold = 0;
-}
-
-static void mlx5_fc_pool_cleanup(struct mlx5_fc_pool *fc_pool)
-{
- struct mlx5_core_dev *dev = fc_pool->dev;
- struct mlx5_fc_bulk *bulk;
- struct mlx5_fc_bulk *tmp;
-
- list_for_each_entry_safe(bulk, tmp, &fc_pool->fully_used, pool_list)
- mlx5_fc_bulk_destroy(dev, bulk);
- list_for_each_entry_safe(bulk, tmp, &fc_pool->partially_used, pool_list)
- mlx5_fc_bulk_destroy(dev, bulk);
- list_for_each_entry_safe(bulk, tmp, &fc_pool->unused, pool_list)
- mlx5_fc_bulk_destroy(dev, bulk);
-}
-
-static void mlx5_fc_pool_update_threshold(struct mlx5_fc_pool *fc_pool)
-{
- fc_pool->threshold = min_t(int, MLX5_FC_POOL_MAX_THRESHOLD,
- fc_pool->used_fcs / MLX5_FC_POOL_USED_BUFF_RATIO);
-}
-
-static struct mlx5_fc_bulk *
-mlx5_fc_pool_alloc_new_bulk(struct mlx5_fc_pool *fc_pool)
-{
- struct mlx5_core_dev *dev = fc_pool->dev;
- struct mlx5_fc_bulk *new_bulk;
-
- new_bulk = mlx5_fc_bulk_create(dev);
- if (!IS_ERR(new_bulk))
- fc_pool->available_fcs += new_bulk->bulk_len;
- mlx5_fc_pool_update_threshold(fc_pool);
- return new_bulk;
-}
-
-static void
-mlx5_fc_pool_free_bulk(struct mlx5_fc_pool *fc_pool, struct mlx5_fc_bulk *bulk)
-{
- struct mlx5_core_dev *dev = fc_pool->dev;
-
- fc_pool->available_fcs -= bulk->bulk_len;
- mlx5_fc_bulk_destroy(dev, bulk);
- mlx5_fc_pool_update_threshold(fc_pool);
-}
-
-static struct mlx5_fc *
-mlx5_fc_pool_acquire_from_list(struct list_head *src_list,
- struct list_head *next_list,
- bool move_non_full_bulk)
-{
- struct mlx5_fc_bulk *bulk;
- struct mlx5_fc *fc;
-
- if (list_empty(src_list))
- return ERR_PTR(-ENODATA);
-
- bulk = list_first_entry(src_list, struct mlx5_fc_bulk, pool_list);
- fc = mlx5_fc_bulk_acquire_fc(bulk);
- if (move_non_full_bulk || mlx5_fc_bulk_get_free_fcs_amount(bulk) == 0)
- list_move(&bulk->pool_list, next_list);
- return fc;
-}
-
-static struct mlx5_fc *
-mlx5_fc_pool_acquire_counter(struct mlx5_fc_pool *fc_pool)
-{
- struct mlx5_fc_bulk *new_bulk;
- struct mlx5_fc *fc;
-
- mutex_lock(&fc_pool->pool_lock);
-
- fc = mlx5_fc_pool_acquire_from_list(&fc_pool->partially_used,
- &fc_pool->fully_used, false);
- if (IS_ERR(fc))
- fc = mlx5_fc_pool_acquire_from_list(&fc_pool->unused,
- &fc_pool->partially_used,
- true);
- if (IS_ERR(fc)) {
- new_bulk = mlx5_fc_pool_alloc_new_bulk(fc_pool);
- if (IS_ERR(new_bulk)) {
- fc = ERR_CAST(new_bulk);
- goto out;
- }
- fc = mlx5_fc_bulk_acquire_fc(new_bulk);
- list_add(&new_bulk->pool_list, &fc_pool->partially_used);
- }
- fc_pool->available_fcs--;
- fc_pool->used_fcs++;
-
-out:
- mutex_unlock(&fc_pool->pool_lock);
- return fc;
-}
-
-static void
-mlx5_fc_pool_release_counter(struct mlx5_fc_pool *fc_pool, struct mlx5_fc *fc)
-{
- struct mlx5_core_dev *dev = fc_pool->dev;
- struct mlx5_fc_bulk *bulk = fc->bulk;
- int bulk_free_fcs_amount;
-
- mutex_lock(&fc_pool->pool_lock);
-
- if (mlx5_fc_bulk_release_fc(bulk, fc)) {
- mlx5_core_warn(dev, "Attempted to release a counter which is not acquired\n");
- goto unlock;
- }
-
- fc_pool->available_fcs++;
- fc_pool->used_fcs--;
-
- bulk_free_fcs_amount = mlx5_fc_bulk_get_free_fcs_amount(bulk);
- if (bulk_free_fcs_amount == 1)
- list_move_tail(&bulk->pool_list, &fc_pool->partially_used);
- if (bulk_free_fcs_amount == bulk->bulk_len) {
- list_del(&bulk->pool_list);
- if (fc_pool->available_fcs > fc_pool->threshold)
- mlx5_fc_pool_free_bulk(fc_pool, bulk);
- else
- list_add(&bulk->pool_list, &fc_pool->unused);
- }
-
-unlock:
- mutex_unlock(&fc_pool->pool_lock);
-}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c
new file mode 100644
index 000000000000..b891d7b9e3e0
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.c
@@ -0,0 +1,194 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2024 NVIDIA Corporation & Affiliates */
+
+#include <mlx5_core.h>
+#include "fs_pool.h"
+
+int mlx5_fs_bulk_init(struct mlx5_core_dev *dev, struct mlx5_fs_bulk *fs_bulk,
+ int bulk_len)
+{
+ int i;
+
+ fs_bulk->bitmask = kvcalloc(BITS_TO_LONGS(bulk_len), sizeof(unsigned long),
+ GFP_KERNEL);
+ if (!fs_bulk->bitmask)
+ return -ENOMEM;
+
+ fs_bulk->bulk_len = bulk_len;
+ for (i = 0; i < bulk_len; i++)
+ set_bit(i, fs_bulk->bitmask);
+
+ return 0;
+}
+
+void mlx5_fs_bulk_cleanup(struct mlx5_fs_bulk *fs_bulk)
+{
+ kvfree(fs_bulk->bitmask);
+}
+
+int mlx5_fs_bulk_get_free_amount(struct mlx5_fs_bulk *bulk)
+{
+ return bitmap_weight(bulk->bitmask, bulk->bulk_len);
+}
+
+static int mlx5_fs_bulk_acquire_index(struct mlx5_fs_bulk *fs_bulk,
+ struct mlx5_fs_pool_index *pool_index)
+{
+ int free_index = find_first_bit(fs_bulk->bitmask, fs_bulk->bulk_len);
+
+ WARN_ON_ONCE(!pool_index || !fs_bulk);
+ if (free_index >= fs_bulk->bulk_len)
+ return -ENOSPC;
+
+ clear_bit(free_index, fs_bulk->bitmask);
+ pool_index->fs_bulk = fs_bulk;
+ pool_index->index = free_index;
+ return 0;
+}
+
+static int mlx5_fs_bulk_release_index(struct mlx5_fs_bulk *fs_bulk, int index)
+{
+ if (test_bit(index, fs_bulk->bitmask))
+ return -EINVAL;
+
+ set_bit(index, fs_bulk->bitmask);
+ return 0;
+}
+
+void mlx5_fs_pool_init(struct mlx5_fs_pool *pool, struct mlx5_core_dev *dev,
+ const struct mlx5_fs_pool_ops *ops)
+{
+ WARN_ON_ONCE(!ops || !ops->bulk_destroy || !ops->bulk_create ||
+ !ops->update_threshold);
+ pool->dev = dev;
+ mutex_init(&pool->pool_lock);
+ INIT_LIST_HEAD(&pool->fully_used);
+ INIT_LIST_HEAD(&pool->partially_used);
+ INIT_LIST_HEAD(&pool->unused);
+ pool->available_units = 0;
+ pool->used_units = 0;
+ pool->threshold = 0;
+ pool->ops = ops;
+}
+
+void mlx5_fs_pool_cleanup(struct mlx5_fs_pool *pool)
+{
+ struct mlx5_core_dev *dev = pool->dev;
+ struct mlx5_fs_bulk *bulk;
+ struct mlx5_fs_bulk *tmp;
+
+ list_for_each_entry_safe(bulk, tmp, &pool->fully_used, pool_list)
+ pool->ops->bulk_destroy(dev, bulk);
+ list_for_each_entry_safe(bulk, tmp, &pool->partially_used, pool_list)
+ pool->ops->bulk_destroy(dev, bulk);
+ list_for_each_entry_safe(bulk, tmp, &pool->unused, pool_list)
+ pool->ops->bulk_destroy(dev, bulk);
+}
+
+static struct mlx5_fs_bulk *
+mlx5_fs_pool_alloc_new_bulk(struct mlx5_fs_pool *fs_pool)
+{
+ struct mlx5_core_dev *dev = fs_pool->dev;
+ struct mlx5_fs_bulk *new_bulk;
+
+ new_bulk = fs_pool->ops->bulk_create(dev);
+ if (new_bulk)
+ fs_pool->available_units += new_bulk->bulk_len;
+ fs_pool->ops->update_threshold(fs_pool);
+ return new_bulk;
+}
+
+static void
+mlx5_fs_pool_free_bulk(struct mlx5_fs_pool *fs_pool, struct mlx5_fs_bulk *bulk)
+{
+ struct mlx5_core_dev *dev = fs_pool->dev;
+
+ fs_pool->available_units -= bulk->bulk_len;
+ fs_pool->ops->bulk_destroy(dev, bulk);
+ fs_pool->ops->update_threshold(fs_pool);
+}
+
+static int
+mlx5_fs_pool_acquire_from_list(struct list_head *src_list,
+ struct list_head *next_list,
+ bool move_non_full_bulk,
+ struct mlx5_fs_pool_index *pool_index)
+{
+ struct mlx5_fs_bulk *fs_bulk;
+ int err;
+
+ if (list_empty(src_list))
+ return -ENODATA;
+
+ fs_bulk = list_first_entry(src_list, struct mlx5_fs_bulk, pool_list);
+ err = mlx5_fs_bulk_acquire_index(fs_bulk, pool_index);
+ if (move_non_full_bulk || mlx5_fs_bulk_get_free_amount(fs_bulk) == 0)
+ list_move(&fs_bulk->pool_list, next_list);
+ return err;
+}
+
+int mlx5_fs_pool_acquire_index(struct mlx5_fs_pool *fs_pool,
+ struct mlx5_fs_pool_index *pool_index)
+{
+ struct mlx5_fs_bulk *new_bulk;
+ int err;
+
+ mutex_lock(&fs_pool->pool_lock);
+
+ err = mlx5_fs_pool_acquire_from_list(&fs_pool->partially_used,
+ &fs_pool->fully_used, false,
+ pool_index);
+ if (err)
+ err = mlx5_fs_pool_acquire_from_list(&fs_pool->unused,
+ &fs_pool->partially_used,
+ true, pool_index);
+ if (err) {
+ new_bulk = mlx5_fs_pool_alloc_new_bulk(fs_pool);
+ if (!new_bulk) {
+ err = -ENOENT;
+ goto out;
+ }
+ err = mlx5_fs_bulk_acquire_index(new_bulk, pool_index);
+ WARN_ON_ONCE(err);
+ list_add(&new_bulk->pool_list, &fs_pool->partially_used);
+ }
+ fs_pool->available_units--;
+ fs_pool->used_units++;
+
+out:
+ mutex_unlock(&fs_pool->pool_lock);
+ return err;
+}
+
+int mlx5_fs_pool_release_index(struct mlx5_fs_pool *fs_pool,
+ struct mlx5_fs_pool_index *pool_index)
+{
+ struct mlx5_fs_bulk *bulk = pool_index->fs_bulk;
+ int bulk_free_amount;
+ int err;
+
+ mutex_lock(&fs_pool->pool_lock);
+
+ /* TBD would rather return void if there was no warn here in original code */
+ err = mlx5_fs_bulk_release_index(bulk, pool_index->index);
+ if (err)
+ goto unlock;
+
+ fs_pool->available_units++;
+ fs_pool->used_units--;
+
+ bulk_free_amount = mlx5_fs_bulk_get_free_amount(bulk);
+ if (bulk_free_amount == 1)
+ list_move_tail(&bulk->pool_list, &fs_pool->partially_used);
+ if (bulk_free_amount == bulk->bulk_len) {
+ list_del(&bulk->pool_list);
+ if (fs_pool->available_units > fs_pool->threshold)
+ mlx5_fs_pool_free_bulk(fs_pool, bulk);
+ else
+ list_add(&bulk->pool_list, &fs_pool->unused);
+ }
+
+unlock:
+ mutex_unlock(&fs_pool->pool_lock);
+ return err;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h
new file mode 100644
index 000000000000..3b149863260c
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_pool.h
@@ -0,0 +1,54 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2024 NVIDIA Corporation & Affiliates */
+
+#ifndef __MLX5_FS_POOL_H__
+#define __MLX5_FS_POOL_H__
+
+#include <linux/mlx5/driver.h>
+
+struct mlx5_fs_bulk {
+ struct list_head pool_list;
+ int bulk_len;
+ unsigned long *bitmask;
+};
+
+struct mlx5_fs_pool_index {
+ struct mlx5_fs_bulk *fs_bulk;
+ int index;
+};
+
+struct mlx5_fs_pool;
+
+struct mlx5_fs_pool_ops {
+ int (*bulk_destroy)(struct mlx5_core_dev *dev, struct mlx5_fs_bulk *bulk);
+ struct mlx5_fs_bulk * (*bulk_create)(struct mlx5_core_dev *dev);
+ void (*update_threshold)(struct mlx5_fs_pool *pool);
+};
+
+struct mlx5_fs_pool {
+ struct mlx5_core_dev *dev;
+ void *pool_ctx;
+ const struct mlx5_fs_pool_ops *ops;
+ struct mutex pool_lock; /* protects pool lists */
+ struct list_head fully_used;
+ struct list_head partially_used;
+ struct list_head unused;
+ int available_units;
+ int used_units;
+ int threshold;
+};
+
+int mlx5_fs_bulk_init(struct mlx5_core_dev *dev, struct mlx5_fs_bulk *fs_bulk,
+ int bulk_len);
+void mlx5_fs_bulk_cleanup(struct mlx5_fs_bulk *fs_bulk);
+int mlx5_fs_bulk_get_free_amount(struct mlx5_fs_bulk *bulk);
+
+void mlx5_fs_pool_init(struct mlx5_fs_pool *pool, struct mlx5_core_dev *dev,
+ const struct mlx5_fs_pool_ops *ops);
+void mlx5_fs_pool_cleanup(struct mlx5_fs_pool *pool);
+int mlx5_fs_pool_acquire_index(struct mlx5_fs_pool *fs_pool,
+ struct mlx5_fs_pool_index *pool_index);
+int mlx5_fs_pool_release_index(struct mlx5_fs_pool *fs_pool,
+ struct mlx5_fs_pool_index *pool_index);
+
+#endif /* __MLX5_FS_POOL_H__ */
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* Re: [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API
2024-12-11 13:42 ` [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API Tariq Toukan
@ 2024-12-12 17:23 ` Simon Horman
2024-12-15 13:39 ` Moshe Shemesh
0 siblings, 1 reply; 34+ messages in thread
From: Simon Horman @ 2024-12-12 17:23 UTC (permalink / raw)
To: Tariq Toukan
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Moshe Shemesh, Yevgeny Kliteynik,
Mark Bloch
On Wed, Dec 11, 2024 at 03:42:16PM +0200, Tariq Toukan wrote:
> From: Moshe Shemesh <moshe@nvidia.com>
>
> Refactor fc_pool API to create generic fs_pool API, as HW steering has
> more flow steering elements which can take advantage of the same pool of
> bulks API. Change fs_counters code to use the fs_pool API.
>
> Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> Reviewed-by: Mark Bloch <mbloch@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
...
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
...
> @@ -447,11 +437,9 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
> /* Flow counter bluks */
>
> struct mlx5_fc_bulk {
> - struct list_head pool_list;
> + struct mlx5_fs_bulk fs_bulk;
> u32 base_id;
> - int bulk_len;
> - unsigned long *bitmask;
> - struct mlx5_fc fcs[] __counted_by(bulk_len);
> + struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
> };
Unfortunately it seems that clang-19 doesn't know how to handle
__counted_by() when used like this:
drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c:442:36: error: 'counted_by' argument must be a simple declaration reference
442 | struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
| ^~~~~~~~~~~~~~~~
././include/linux/compiler_types.h:346:62: note: expanded from macro '__counted_by'
346 | # define __counted_by(member) __attribute__((__counted_by__(member)))
| ^~~~~~
...
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API
2024-12-12 17:23 ` Simon Horman
@ 2024-12-15 13:39 ` Moshe Shemesh
2024-12-18 5:22 ` Kees Cook
0 siblings, 1 reply; 34+ messages in thread
From: Moshe Shemesh @ 2024-12-15 13:39 UTC (permalink / raw)
To: Simon Horman, Tariq Toukan, Kees Cook
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Yevgeny Kliteynik, Mark Bloch
On 12/12/2024 7:23 PM, Simon Horman wrote:
> External email: Use caution opening links or attachments
>
>
> On Wed, Dec 11, 2024 at 03:42:16PM +0200, Tariq Toukan wrote:
>> From: Moshe Shemesh <moshe@nvidia.com>
>>
>> Refactor fc_pool API to create generic fs_pool API, as HW steering has
>> more flow steering elements which can take advantage of the same pool of
>> bulks API. Change fs_counters code to use the fs_pool API.
>>
>> Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
>> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
>> Reviewed-by: Mark Bloch <mbloch@nvidia.com>
>> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
>
> ...
>
>> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
>
> ...
>
>> @@ -447,11 +437,9 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
>> /* Flow counter bluks */
>>
>> struct mlx5_fc_bulk {
>> - struct list_head pool_list;
>> + struct mlx5_fs_bulk fs_bulk;
>> u32 base_id;
>> - int bulk_len;
>> - unsigned long *bitmask;
>> - struct mlx5_fc fcs[] __counted_by(bulk_len);
>> + struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
>> };
>
> Unfortunately it seems that clang-19 doesn't know how to handle
> __counted_by() when used like this:
>
> drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c:442:36: error: 'counted_by' argument must be a simple declaration reference
> 442 | struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
Thanks Simon, from code perspective, bulk_len should be now in the inner
struct fs_bulk.
Keen Cook, is that going to be supported in the future? for now I will
just remove __counted_by() from this struct.
> | ^~~~~~~~~~~~~~~~
> ././include/linux/compiler_types.h:346:62: note: expanded from macro '__counted_by'
> 346 | # define __counted_by(member) __attribute__((__counted_by__(member)))
> | ^~~~~~
>
> ...
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API
2024-12-15 13:39 ` Moshe Shemesh
@ 2024-12-18 5:22 ` Kees Cook
2024-12-18 8:21 ` Moshe Shemesh
0 siblings, 1 reply; 34+ messages in thread
From: Kees Cook @ 2024-12-18 5:22 UTC (permalink / raw)
To: Moshe Shemesh, morbo, qing.zhao
Cc: Simon Horman, Tariq Toukan, David S. Miller, Jakub Kicinski,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Yevgeny Kliteynik,
Mark Bloch, linux-hardening
On Sun, Dec 15, 2024 at 03:39:11PM +0200, Moshe Shemesh wrote:
>
>
> On 12/12/2024 7:23 PM, Simon Horman wrote:
> > External email: Use caution opening links or attachments
> >
> >
> > On Wed, Dec 11, 2024 at 03:42:16PM +0200, Tariq Toukan wrote:
> > > From: Moshe Shemesh <moshe@nvidia.com>
> > >
> > > Refactor fc_pool API to create generic fs_pool API, as HW steering has
> > > more flow steering elements which can take advantage of the same pool of
> > > bulks API. Change fs_counters code to use the fs_pool API.
> > >
> > > Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
> > > Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> > > Reviewed-by: Mark Bloch <mbloch@nvidia.com>
> > > Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> >
> > ...
> >
> > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
> >
> > ...
> >
> > > @@ -447,11 +437,9 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
> > > /* Flow counter bluks */
> > >
> > > struct mlx5_fc_bulk {
> > > - struct list_head pool_list;
> > > + struct mlx5_fs_bulk fs_bulk;
> > > u32 base_id;
> > > - int bulk_len;
> > > - unsigned long *bitmask;
> > > - struct mlx5_fc fcs[] __counted_by(bulk_len);
> > > + struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
> > > };
> >
> > Unfortunately it seems that clang-19 doesn't know how to handle
> > __counted_by() when used like this:
> >
> > drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c:442:36: error: 'counted_by' argument must be a simple declaration reference
> > 442 | struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
>
> Thanks Simon, from code perspective, bulk_len should be now in the inner
> struct fs_bulk.
>
> Keen Cook, is that going to be supported in the future? for now I will just
> remove __counted_by() from this struct.
I am expecting this will be supported in the future, yes, but there isn't
an ETA for it yet. Neither GCC 15 nor Clang are currently supporting
sub-struct members -- the counted_by member needs to be at the same
"level" in the struct as the annotated flexible array.
Is it possible to move "base_id" above "fs_bulk" and move "fcs" into
the of end struct mlx5_fs_bulk?
-Kees
--
Kees Cook
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API
2024-12-18 5:22 ` Kees Cook
@ 2024-12-18 8:21 ` Moshe Shemesh
0 siblings, 0 replies; 34+ messages in thread
From: Moshe Shemesh @ 2024-12-18 8:21 UTC (permalink / raw)
To: Kees Cook, morbo, qing.zhao
Cc: Simon Horman, Tariq Toukan, David S. Miller, Jakub Kicinski,
Paolo Abeni, Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Yevgeny Kliteynik,
Mark Bloch, linux-hardening
On 12/18/2024 7:22 AM, Kees Cook wrote:
>
> On Sun, Dec 15, 2024 at 03:39:11PM +0200, Moshe Shemesh wrote:
>>
>>
>> On 12/12/2024 7:23 PM, Simon Horman wrote:
>>> External email: Use caution opening links or attachments
>>>
>>>
>>> On Wed, Dec 11, 2024 at 03:42:16PM +0200, Tariq Toukan wrote:
>>>> From: Moshe Shemesh <moshe@nvidia.com>
>>>>
>>>> Refactor fc_pool API to create generic fs_pool API, as HW steering has
>>>> more flow steering elements which can take advantage of the same pool of
>>>> bulks API. Change fs_counters code to use the fs_pool API.
>>>>
>>>> Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
>>>> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
>>>> Reviewed-by: Mark Bloch <mbloch@nvidia.com>
>>>> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
>>>
>>> ...
>>>
>>>> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
>>>
>>> ...
>>>
>>>> @@ -447,11 +437,9 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
>>>> /* Flow counter bluks */
>>>>
>>>> struct mlx5_fc_bulk {
>>>> - struct list_head pool_list;
>>>> + struct mlx5_fs_bulk fs_bulk;
>>>> u32 base_id;
>>>> - int bulk_len;
>>>> - unsigned long *bitmask;
>>>> - struct mlx5_fc fcs[] __counted_by(bulk_len);
>>>> + struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
>>>> };
>>>
>>> Unfortunately it seems that clang-19 doesn't know how to handle
>>> __counted_by() when used like this:
>>>
>>> drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c:442:36: error: 'counted_by' argument must be a simple declaration reference
>>> 442 | struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
>>
>> Thanks Simon, from code perspective, bulk_len should be now in the inner
>> struct fs_bulk.
>>
>> Keen Cook, is that going to be supported in the future? for now I will just
>> remove __counted_by() from this struct.
>
> I am expecting this will be supported in the future, yes, but there isn't
> an ETA for it yet. Neither GCC 15 nor Clang are currently supporting
> sub-struct members -- the counted_by member needs to be at the same
> "level" in the struct as the annotated flexible array.
>
> Is it possible to move "base_id" above "fs_bulk" and move "fcs" into
> the of end struct mlx5_fs_bulk?
>
fcs is specific for mlx5_fc_bulk and so can't be part of fs_bulk.
In follow up patches for hws feature I am using mlx5_fs_bulk in other
type of bulks, fcs is specific for mlx5_fc_bulk.
Unfortunately, I don't see a nice way to have bulk_len and fcs on the
same struct.
> -Kees
>
> --
> Kees Cook
^ permalink raw reply [flat|nested] 34+ messages in thread
* [PATCH net-next 06/12] net/mlx5: fs, retry insertion to hash table on EBUSY
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (4 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 07/12] net/mlx5: HWS, no need to expose mlx5hws_send_queues_open/close Tariq Toukan
` (5 subsequent siblings)
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Mark Bloch,
Tariq Toukan
From: Mark Bloch <mbloch@nvidia.com>
When inserting into an rhashtable faster than it can grow, an -EBUSY error
may be encountered. Modify the insertion logic to retry on -EBUSY until
either a successful insertion or a genuine error is returned.
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index f781f8f169b9..ae1a5705b26d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -821,11 +821,17 @@ static int insert_fte(struct mlx5_flow_group *fg, struct fs_fte *fte)
return index;
fte->index = index + fg->start_index;
+retry_insert:
ret = rhashtable_insert_fast(&fg->ftes_hash,
&fte->hash,
rhash_fte);
- if (ret)
+ if (ret) {
+ if (ret == -EBUSY) {
+ cond_resched();
+ goto retry_insert;
+ }
goto err_ida_remove;
+ }
tree_add_node(&fte->node, &fg->node);
list_add_tail(&fte->node.list, &fg->node.children);
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* [PATCH net-next 07/12] net/mlx5: HWS, no need to expose mlx5hws_send_queues_open/close
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (5 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 06/12] net/mlx5: fs, retry insertion to hash table on EBUSY Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 08/12] net/mlx5: HWS, do not initialize native API queues Tariq Toukan
` (4 subsequent siblings)
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma,
Yevgeny Kliteynik, Mark Bloch, Itamar Gozlan, Tariq Toukan
From: Yevgeny Kliteynik <kliteyn@nvidia.com>
No need to have mlx5hws_send_queues_open/close in header.
Make them static and remove from header.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Itamar Gozlan <igozlan@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
.../ethernet/mellanox/mlx5/core/steering/hws/send.c | 12 ++++++------
.../ethernet/mellanox/mlx5/core/steering/hws/send.h | 6 ------
2 files changed, 6 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
index 883b4ed30892..a93da4f71646 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
@@ -896,15 +896,15 @@ static int mlx5hws_send_ring_open(struct mlx5hws_context *ctx,
return err;
}
-void mlx5hws_send_queue_close(struct mlx5hws_send_engine *queue)
+static void hws_send_queue_close(struct mlx5hws_send_engine *queue)
{
hws_send_ring_close(queue);
kfree(queue->completed.entries);
}
-int mlx5hws_send_queue_open(struct mlx5hws_context *ctx,
- struct mlx5hws_send_engine *queue,
- u16 queue_size)
+static int hws_send_queue_open(struct mlx5hws_context *ctx,
+ struct mlx5hws_send_engine *queue,
+ u16 queue_size)
{
int err;
@@ -936,7 +936,7 @@ int mlx5hws_send_queue_open(struct mlx5hws_context *ctx,
static void __hws_send_queues_close(struct mlx5hws_context *ctx, u16 queues)
{
while (queues--)
- mlx5hws_send_queue_close(&ctx->send_queue[queues]);
+ hws_send_queue_close(&ctx->send_queue[queues]);
}
static void hws_send_queues_bwc_locks_destroy(struct mlx5hws_context *ctx)
@@ -1022,7 +1022,7 @@ int mlx5hws_send_queues_open(struct mlx5hws_context *ctx,
}
for (i = 0; i < ctx->queues; i++) {
- err = mlx5hws_send_queue_open(ctx, &ctx->send_queue[i], queue_size);
+ err = hws_send_queue_open(ctx, &ctx->send_queue[i], queue_size);
if (err)
goto close_send_queues;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.h
index b50825d6dc53..f833092235c1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.h
@@ -189,12 +189,6 @@ void mlx5hws_send_abort_new_dep_wqe(struct mlx5hws_send_engine *queue);
void mlx5hws_send_all_dep_wqe(struct mlx5hws_send_engine *queue);
-void mlx5hws_send_queue_close(struct mlx5hws_send_engine *queue);
-
-int mlx5hws_send_queue_open(struct mlx5hws_context *ctx,
- struct mlx5hws_send_engine *queue,
- u16 queue_size);
-
void mlx5hws_send_queues_close(struct mlx5hws_context *ctx);
int mlx5hws_send_queues_open(struct mlx5hws_context *ctx,
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* [PATCH net-next 08/12] net/mlx5: HWS, do not initialize native API queues
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (6 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 07/12] net/mlx5: HWS, no need to expose mlx5hws_send_queues_open/close Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 09/12] net/mlx5: DR, expand SWS STE callbacks and consolidate common structs Tariq Toukan
` (3 subsequent siblings)
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma,
Yevgeny Kliteynik, Itamar Gozlan, Mark Bloch, Tariq Toukan
From: Yevgeny Kliteynik <kliteyn@nvidia.com>
HWS has two types of APIs:
- Native: fastest and slimmest, async API.
The user of this API is required to manage rule handles memory,
and to poll for completion for each rule.
- BWC: backward compatible API, similar semantics to SWS API.
This layer is implemented above native API and it does all
the work for the user, so that it is easy to switch between
SWS and HWS.
Right now the existing users of HWS require only BWC API.
Therefore, in order to not waste resources, this patch disables
send queues allocation for native API.
If in the future support for faster HWS rule insertion will be required
(such as for Connection Tracking), native queues can be enabled.
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Itamar Gozlan <igozlan@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
.../mellanox/mlx5/core/steering/hws/bwc.h | 6 ++-
.../mellanox/mlx5/core/steering/hws/context.c | 6 ++-
.../mellanox/mlx5/core/steering/hws/context.h | 6 +++
.../mellanox/mlx5/core/steering/hws/mlx5hws.h | 1 -
.../mellanox/mlx5/core/steering/hws/send.c | 38 ++++++++++++++-----
5 files changed, 43 insertions(+), 14 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/bwc.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/bwc.h
index 0b745968e21e..3d4965213b01 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/bwc.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/bwc.h
@@ -60,9 +60,11 @@ void mlx5hws_bwc_rule_fill_attr(struct mlx5hws_bwc_matcher *bwc_matcher,
static inline u16 mlx5hws_bwc_queues(struct mlx5hws_context *ctx)
{
/* Besides the control queue, half of the queues are
- * reguler HWS queues, and the other half are BWC queues.
+ * regular HWS queues, and the other half are BWC queues.
*/
- return (ctx->queues - 1) / 2;
+ if (mlx5hws_context_bwc_supported(ctx))
+ return (ctx->queues - 1) / 2;
+ return 0;
}
static inline u16 mlx5hws_bwc_get_queue_id(struct mlx5hws_context *ctx, u16 idx)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.c
index fd48b05e91e0..4a8928f33bb9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.c
@@ -161,8 +161,10 @@ static int hws_context_init_hws(struct mlx5hws_context *ctx,
if (ret)
goto uninit_pd;
- if (attr->bwc)
- ctx->flags |= MLX5HWS_CONTEXT_FLAG_BWC_SUPPORT;
+ /* Context has support for backward compatible API,
+ * and does not have support for native HWS API.
+ */
+ ctx->flags |= MLX5HWS_CONTEXT_FLAG_BWC_SUPPORT;
ret = mlx5hws_send_queues_open(ctx, attr->queues, attr->queue_size);
if (ret)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.h
index 47f5cc8de73f..1c9cc4fba083 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/context.h
@@ -8,6 +8,7 @@ enum mlx5hws_context_flags {
MLX5HWS_CONTEXT_FLAG_HWS_SUPPORT = 1 << 0,
MLX5HWS_CONTEXT_FLAG_PRIVATE_PD = 1 << 1,
MLX5HWS_CONTEXT_FLAG_BWC_SUPPORT = 1 << 2,
+ MLX5HWS_CONTEXT_FLAG_NATIVE_SUPPORT = 1 << 3,
};
enum mlx5hws_context_shared_stc_type {
@@ -58,6 +59,11 @@ static inline bool mlx5hws_context_bwc_supported(struct mlx5hws_context *ctx)
return ctx->flags & MLX5HWS_CONTEXT_FLAG_BWC_SUPPORT;
}
+static inline bool mlx5hws_context_native_supported(struct mlx5hws_context *ctx)
+{
+ return ctx->flags & MLX5HWS_CONTEXT_FLAG_NATIVE_SUPPORT;
+}
+
bool mlx5hws_context_cap_dynamic_reparse(struct mlx5hws_context *ctx);
u8 mlx5hws_context_get_reparse_mode(struct mlx5hws_context *ctx);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/mlx5hws.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/mlx5hws.h
index f39d636ff39a..5121951f2778 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/mlx5hws.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/mlx5hws.h
@@ -70,7 +70,6 @@ enum mlx5hws_send_queue_actions {
struct mlx5hws_context_attr {
u16 queues;
u16 queue_size;
- bool bwc; /* add support for backward compatible API*/
};
struct mlx5hws_table_attr {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
index a93da4f71646..e3d621f013f6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/hws/send.c
@@ -898,6 +898,9 @@ static int mlx5hws_send_ring_open(struct mlx5hws_context *ctx,
static void hws_send_queue_close(struct mlx5hws_send_engine *queue)
{
+ if (!queue->num_entries)
+ return; /* this queue wasn't initialized */
+
hws_send_ring_close(queue);
kfree(queue->completed.entries);
}
@@ -1000,12 +1003,33 @@ static int hws_bwc_send_queues_init(struct mlx5hws_context *ctx)
return -ENOMEM;
}
+static int hws_send_queues_open(struct mlx5hws_context *ctx, u16 queue_size)
+{
+ int err = 0;
+ u32 i = 0;
+
+ /* If native API isn't supported, skip the unused native queues:
+ * initialize BWC queues and control queue only.
+ */
+ if (!mlx5hws_context_native_supported(ctx))
+ i = mlx5hws_bwc_get_queue_id(ctx, 0);
+
+ for (; i < ctx->queues; i++) {
+ err = hws_send_queue_open(ctx, &ctx->send_queue[i], queue_size);
+ if (err) {
+ __hws_send_queues_close(ctx, i);
+ return err;
+ }
+ }
+
+ return 0;
+}
+
int mlx5hws_send_queues_open(struct mlx5hws_context *ctx,
u16 queues,
u16 queue_size)
{
int err = 0;
- u32 i;
/* Open one extra queue for control path */
ctx->queues = queues + 1;
@@ -1021,17 +1045,13 @@ int mlx5hws_send_queues_open(struct mlx5hws_context *ctx,
goto free_bwc_locks;
}
- for (i = 0; i < ctx->queues; i++) {
- err = hws_send_queue_open(ctx, &ctx->send_queue[i], queue_size);
- if (err)
- goto close_send_queues;
- }
+ err = hws_send_queues_open(ctx, queue_size);
+ if (err)
+ goto free_queues;
return 0;
-close_send_queues:
- __hws_send_queues_close(ctx, i);
-
+free_queues:
kfree(ctx->send_queue);
free_bwc_locks:
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* [PATCH net-next 09/12] net/mlx5: DR, expand SWS STE callbacks and consolidate common structs
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (7 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 08/12] net/mlx5: HWS, do not initialize native API queues Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering Tariq Toukan
` (2 subsequent siblings)
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Itamar Gozlan,
Yevgeny Kliteynik, Tariq Toukan
From: Itamar Gozlan <igozlan@nvidia.com>
Expand SWS STE callbacks to support ConnectX-8 hardware.
Move common enums and structures to a shared header file.
Signed-off-by: Itamar Gozlan <igozlan@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
.../mellanox/mlx5/core/steering/sws/dr_ste.c | 4 +-
.../mellanox/mlx5/core/steering/sws/dr_ste.h | 18 +-
.../mlx5/core/steering/sws/dr_ste_v0.c | 6 +-
.../mlx5/core/steering/sws/dr_ste_v1.c | 207 ++++--------------
.../mlx5/core/steering/sws/dr_ste_v1.h | 147 ++++++++++++-
.../mlx5/core/steering/sws/dr_ste_v2.c | 169 +-------------
.../mlx5/core/steering/sws/dr_ste_v2.h | 168 ++++++++++++++
7 files changed, 377 insertions(+), 342 deletions(-)
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.h
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c
index e94fbb015efa..01ba8eae2983 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c
@@ -555,7 +555,7 @@ void mlx5dr_ste_set_actions_tx(struct mlx5dr_ste_ctx *ste_ctx,
struct mlx5dr_ste_actions_attr *attr,
u32 *added_stes)
{
- ste_ctx->set_actions_tx(dmn, action_type_set, ste_ctx->actions_caps,
+ ste_ctx->set_actions_tx(ste_ctx, dmn, action_type_set, ste_ctx->actions_caps,
hw_ste_arr, attr, added_stes);
}
@@ -566,7 +566,7 @@ void mlx5dr_ste_set_actions_rx(struct mlx5dr_ste_ctx *ste_ctx,
struct mlx5dr_ste_actions_attr *attr,
u32 *added_stes)
{
- ste_ctx->set_actions_rx(dmn, action_type_set, ste_ctx->actions_caps,
+ ste_ctx->set_actions_rx(ste_ctx, dmn, action_type_set, ste_ctx->actions_caps,
hw_ste_arr, attr, added_stes);
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h
index 54a6619c3ecb..b6ec8d30d990 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h
@@ -160,13 +160,15 @@ struct mlx5dr_ste_ctx {
/* Actions */
u32 actions_caps;
- void (*set_actions_rx)(struct mlx5dr_domain *dmn,
+ void (*set_actions_rx)(struct mlx5dr_ste_ctx *ste_ctx,
+ struct mlx5dr_domain *dmn,
u8 *action_type_set,
u32 actions_caps,
u8 *hw_ste_arr,
struct mlx5dr_ste_actions_attr *attr,
u32 *added_stes);
- void (*set_actions_tx)(struct mlx5dr_domain *dmn,
+ void (*set_actions_tx)(struct mlx5dr_ste_ctx *ste_ctx,
+ struct mlx5dr_domain *dmn,
u8 *action_type_set,
u32 actions_caps,
u8 *hw_ste_arr,
@@ -197,7 +199,17 @@ struct mlx5dr_ste_ctx {
u16 *used_hw_action_num);
int (*alloc_modify_hdr_chunk)(struct mlx5dr_action *action);
void (*dealloc_modify_hdr_chunk)(struct mlx5dr_action *action);
-
+ /* Actions bit set */
+ void (*set_encap)(u8 *hw_ste_p, u8 *d_action,
+ u32 reformat_id, int size);
+ void (*set_push_vlan)(u8 *ste, u8 *d_action,
+ u32 vlan_hdr);
+ void (*set_pop_vlan)(u8 *hw_ste_p, u8 *s_action,
+ u8 vlans_num);
+ void (*set_rx_decap)(u8 *hw_ste_p, u8 *s_action);
+ void (*set_encap_l3)(u8 *hw_ste_p, u8 *frst_s_action,
+ u8 *scnd_d_action, u32 reformat_id,
+ int size);
/* Send */
void (*prepare_for_postsend)(u8 *hw_ste_p, u32 ste_size);
};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v0.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v0.c
index e9f6c7ed7a7b..42536bee55e2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v0.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v0.c
@@ -406,7 +406,8 @@ static void dr_ste_v0_arr_init_next(u8 **last_ste,
}
static void
-dr_ste_v0_set_actions_tx(struct mlx5dr_domain *dmn,
+dr_ste_v0_set_actions_tx(struct mlx5dr_ste_ctx *ste_ctx,
+ struct mlx5dr_domain *dmn,
u8 *action_type_set,
u32 actions_caps,
u8 *last_ste,
@@ -476,7 +477,8 @@ dr_ste_v0_set_actions_tx(struct mlx5dr_domain *dmn,
}
static void
-dr_ste_v0_set_actions_rx(struct mlx5dr_domain *dmn,
+dr_ste_v0_set_actions_rx(struct mlx5dr_ste_ctx *ste_ctx,
+ struct mlx5dr_domain *dmn,
u8 *action_type_set,
u32 actions_caps,
u8 *last_ste,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.c
index 1d49704b9542..7f83d77c43ef 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.c
@@ -5,136 +5,6 @@
#include "mlx5_ifc_dr_ste_v1.h"
#include "dr_ste_v1.h"
-#define DR_STE_CALC_DFNR_TYPE(lookup_type, inner) \
- ((inner) ? DR_STE_V1_LU_TYPE_##lookup_type##_I : \
- DR_STE_V1_LU_TYPE_##lookup_type##_O)
-
-enum dr_ste_v1_entry_format {
- DR_STE_V1_TYPE_BWC_BYTE = 0x0,
- DR_STE_V1_TYPE_BWC_DW = 0x1,
- DR_STE_V1_TYPE_MATCH = 0x2,
- DR_STE_V1_TYPE_MATCH_RANGES = 0x7,
-};
-
-/* Lookup type is built from 2B: [ Definer mode 1B ][ Definer index 1B ] */
-enum {
- DR_STE_V1_LU_TYPE_NOP = 0x0000,
- DR_STE_V1_LU_TYPE_ETHL2_TNL = 0x0002,
- DR_STE_V1_LU_TYPE_IBL3_EXT = 0x0102,
- DR_STE_V1_LU_TYPE_ETHL2_O = 0x0003,
- DR_STE_V1_LU_TYPE_IBL4 = 0x0103,
- DR_STE_V1_LU_TYPE_ETHL2_I = 0x0004,
- DR_STE_V1_LU_TYPE_SRC_QP_GVMI = 0x0104,
- DR_STE_V1_LU_TYPE_ETHL2_SRC_O = 0x0005,
- DR_STE_V1_LU_TYPE_ETHL2_HEADERS_O = 0x0105,
- DR_STE_V1_LU_TYPE_ETHL2_SRC_I = 0x0006,
- DR_STE_V1_LU_TYPE_ETHL2_HEADERS_I = 0x0106,
- DR_STE_V1_LU_TYPE_ETHL3_IPV4_5_TUPLE_O = 0x0007,
- DR_STE_V1_LU_TYPE_IPV6_DES_O = 0x0107,
- DR_STE_V1_LU_TYPE_ETHL3_IPV4_5_TUPLE_I = 0x0008,
- DR_STE_V1_LU_TYPE_IPV6_DES_I = 0x0108,
- DR_STE_V1_LU_TYPE_ETHL4_O = 0x0009,
- DR_STE_V1_LU_TYPE_IPV6_SRC_O = 0x0109,
- DR_STE_V1_LU_TYPE_ETHL4_I = 0x000a,
- DR_STE_V1_LU_TYPE_IPV6_SRC_I = 0x010a,
- DR_STE_V1_LU_TYPE_ETHL2_SRC_DST_O = 0x000b,
- DR_STE_V1_LU_TYPE_MPLS_O = 0x010b,
- DR_STE_V1_LU_TYPE_ETHL2_SRC_DST_I = 0x000c,
- DR_STE_V1_LU_TYPE_MPLS_I = 0x010c,
- DR_STE_V1_LU_TYPE_ETHL3_IPV4_MISC_O = 0x000d,
- DR_STE_V1_LU_TYPE_GRE = 0x010d,
- DR_STE_V1_LU_TYPE_FLEX_PARSER_TNL_HEADER = 0x000e,
- DR_STE_V1_LU_TYPE_GENERAL_PURPOSE = 0x010e,
- DR_STE_V1_LU_TYPE_ETHL3_IPV4_MISC_I = 0x000f,
- DR_STE_V1_LU_TYPE_STEERING_REGISTERS_0 = 0x010f,
- DR_STE_V1_LU_TYPE_STEERING_REGISTERS_1 = 0x0110,
- DR_STE_V1_LU_TYPE_FLEX_PARSER_OK = 0x0011,
- DR_STE_V1_LU_TYPE_FLEX_PARSER_0 = 0x0111,
- DR_STE_V1_LU_TYPE_FLEX_PARSER_1 = 0x0112,
- DR_STE_V1_LU_TYPE_ETHL4_MISC_O = 0x0113,
- DR_STE_V1_LU_TYPE_ETHL4_MISC_I = 0x0114,
- DR_STE_V1_LU_TYPE_INVALID = 0x00ff,
- DR_STE_V1_LU_TYPE_DONT_CARE = MLX5DR_STE_LU_TYPE_DONT_CARE,
-};
-
-enum dr_ste_v1_header_anchors {
- DR_STE_HEADER_ANCHOR_START_OUTER = 0x00,
- DR_STE_HEADER_ANCHOR_1ST_VLAN = 0x02,
- DR_STE_HEADER_ANCHOR_IPV6_IPV4 = 0x07,
- DR_STE_HEADER_ANCHOR_INNER_MAC = 0x13,
- DR_STE_HEADER_ANCHOR_INNER_IPV6_IPV4 = 0x19,
-};
-
-enum dr_ste_v1_action_size {
- DR_STE_ACTION_SINGLE_SZ = 4,
- DR_STE_ACTION_DOUBLE_SZ = 8,
- DR_STE_ACTION_TRIPLE_SZ = 12,
-};
-
-enum dr_ste_v1_action_insert_ptr_attr {
- DR_STE_V1_ACTION_INSERT_PTR_ATTR_NONE = 0, /* Regular push header (e.g. push vlan) */
- DR_STE_V1_ACTION_INSERT_PTR_ATTR_ENCAP = 1, /* Encapsulation / Tunneling */
- DR_STE_V1_ACTION_INSERT_PTR_ATTR_ESP = 2, /* IPsec */
-};
-
-enum dr_ste_v1_action_id {
- DR_STE_V1_ACTION_ID_NOP = 0x00,
- DR_STE_V1_ACTION_ID_COPY = 0x05,
- DR_STE_V1_ACTION_ID_SET = 0x06,
- DR_STE_V1_ACTION_ID_ADD = 0x07,
- DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE = 0x08,
- DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER = 0x09,
- DR_STE_V1_ACTION_ID_INSERT_INLINE = 0x0a,
- DR_STE_V1_ACTION_ID_INSERT_POINTER = 0x0b,
- DR_STE_V1_ACTION_ID_FLOW_TAG = 0x0c,
- DR_STE_V1_ACTION_ID_QUEUE_ID_SEL = 0x0d,
- DR_STE_V1_ACTION_ID_ACCELERATED_LIST = 0x0e,
- DR_STE_V1_ACTION_ID_MODIFY_LIST = 0x0f,
- DR_STE_V1_ACTION_ID_ASO = 0x12,
- DR_STE_V1_ACTION_ID_TRAILER = 0x13,
- DR_STE_V1_ACTION_ID_COUNTER_ID = 0x14,
- DR_STE_V1_ACTION_ID_MAX = 0x21,
- /* use for special cases */
- DR_STE_V1_ACTION_ID_SPECIAL_ENCAP_L3 = 0x22,
-};
-
-enum {
- DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_0 = 0x00,
- DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_1 = 0x01,
- DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_2 = 0x02,
- DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_0 = 0x08,
- DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_1 = 0x09,
- DR_STE_V1_ACTION_MDFY_FLD_L3_OUT_0 = 0x0e,
- DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_0 = 0x18,
- DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_1 = 0x19,
- DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_0 = 0x40,
- DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_1 = 0x41,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_0 = 0x44,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_1 = 0x45,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_2 = 0x46,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_3 = 0x47,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_0 = 0x4c,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_1 = 0x4d,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_2 = 0x4e,
- DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_3 = 0x4f,
- DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_0 = 0x5e,
- DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_1 = 0x5f,
- DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_0 = 0x6f,
- DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_1 = 0x70,
- DR_STE_V1_ACTION_MDFY_FLD_METADATA_2_CQE = 0x7b,
- DR_STE_V1_ACTION_MDFY_FLD_GNRL_PURPOSE = 0x7c,
- DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_0 = 0x8c,
- DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_1 = 0x8d,
- DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_0 = 0x8e,
- DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_1 = 0x8f,
- DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_0 = 0x90,
- DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_1 = 0x91,
-};
-
-enum dr_ste_v1_aso_ctx_type {
- DR_STE_V1_ASO_CTX_TYPE_POLICERS = 0x2,
-};
-
static const struct mlx5dr_ste_action_modify_field dr_ste_v1_action_modify_field_arr[] = {
[MLX5_ACTION_IN_FIELD_OUT_SMAC_47_16] = {
.hw_field = DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_0, .start = 0, .end = 31,
@@ -379,13 +249,12 @@ static void dr_ste_v1_set_counter_id(u8 *hw_ste_p, u32 ctr_id)
MLX5_SET(ste_match_bwc_v1, hw_ste_p, counter_id, ctr_id);
}
-static void dr_ste_v1_set_reparse(u8 *hw_ste_p)
+void dr_ste_v1_set_reparse(u8 *hw_ste_p)
{
MLX5_SET(ste_match_bwc_v1, hw_ste_p, reparse, 1);
}
-static void dr_ste_v1_set_encap(u8 *hw_ste_p, u8 *d_action,
- u32 reformat_id, int size)
+void dr_ste_v1_set_encap(u8 *hw_ste_p, u8 *d_action, u32 reformat_id, int size)
{
MLX5_SET(ste_double_action_insert_with_ptr_v1, d_action, action_id,
DR_STE_V1_ACTION_ID_INSERT_POINTER);
@@ -432,8 +301,7 @@ static void dr_ste_v1_set_remove_hdr(u8 *hw_ste_p, u8 *s_action,
dr_ste_v1_set_reparse(hw_ste_p);
}
-static void dr_ste_v1_set_push_vlan(u8 *hw_ste_p, u8 *d_action,
- u32 vlan_hdr)
+void dr_ste_v1_set_push_vlan(u8 *hw_ste_p, u8 *d_action, u32 vlan_hdr)
{
MLX5_SET(ste_double_action_insert_with_inline_v1, d_action,
action_id, DR_STE_V1_ACTION_ID_INSERT_INLINE);
@@ -446,7 +314,7 @@ static void dr_ste_v1_set_push_vlan(u8 *hw_ste_p, u8 *d_action,
dr_ste_v1_set_reparse(hw_ste_p);
}
-static void dr_ste_v1_set_pop_vlan(u8 *hw_ste_p, u8 *s_action, u8 vlans_num)
+void dr_ste_v1_set_pop_vlan(u8 *hw_ste_p, u8 *s_action, u8 vlans_num)
{
MLX5_SET(ste_single_action_remove_header_size_v1, s_action,
action_id, DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE);
@@ -459,11 +327,8 @@ static void dr_ste_v1_set_pop_vlan(u8 *hw_ste_p, u8 *s_action, u8 vlans_num)
dr_ste_v1_set_reparse(hw_ste_p);
}
-static void dr_ste_v1_set_encap_l3(u8 *hw_ste_p,
- u8 *frst_s_action,
- u8 *scnd_d_action,
- u32 reformat_id,
- int size)
+void dr_ste_v1_set_encap_l3(u8 *hw_ste_p, u8 *frst_s_action, u8 *scnd_d_action,
+ u32 reformat_id, int size)
{
/* Remove L2 headers */
MLX5_SET(ste_single_action_remove_header_v1, frst_s_action, action_id,
@@ -483,7 +348,7 @@ static void dr_ste_v1_set_encap_l3(u8 *hw_ste_p,
dr_ste_v1_set_reparse(hw_ste_p);
}
-static void dr_ste_v1_set_rx_decap(u8 *hw_ste_p, u8 *s_action)
+void dr_ste_v1_set_rx_decap(u8 *hw_ste_p, u8 *s_action)
{
MLX5_SET(ste_single_action_remove_header_v1, s_action, action_id,
DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER);
@@ -620,7 +485,8 @@ static void dr_ste_v1_arr_init_next_match_range(u8 **last_ste,
dr_ste_v1_set_entry_type(*last_ste, DR_STE_V1_TYPE_MATCH_RANGES);
}
-void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn,
+void dr_ste_v1_set_actions_tx(struct mlx5dr_ste_ctx *ste_ctx,
+ struct mlx5dr_domain *dmn,
u8 *action_type_set,
u32 actions_caps,
u8 *last_ste,
@@ -640,7 +506,7 @@ void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn,
last_ste, action);
action_sz = DR_STE_ACTION_TRIPLE_SZ;
}
- dr_ste_v1_set_pop_vlan(last_ste, action, attr->vlans.count);
+ ste_ctx->set_pop_vlan(last_ste, action, attr->vlans.count);
action_sz -= DR_STE_ACTION_SINGLE_SZ;
action += DR_STE_ACTION_SINGLE_SZ;
@@ -677,8 +543,8 @@ void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn,
action_sz = DR_STE_ACTION_TRIPLE_SZ;
allow_encap = true;
}
- dr_ste_v1_set_push_vlan(last_ste, action,
- attr->vlans.headers[i]);
+ ste_ctx->set_push_vlan(last_ste, action,
+ attr->vlans.headers[i]);
action_sz -= DR_STE_ACTION_DOUBLE_SZ;
action += DR_STE_ACTION_DOUBLE_SZ;
}
@@ -691,9 +557,9 @@ void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn,
action_sz = DR_STE_ACTION_TRIPLE_SZ;
allow_encap = true;
}
- dr_ste_v1_set_encap(last_ste, action,
- attr->reformat.id,
- attr->reformat.size);
+ ste_ctx->set_encap(last_ste, action,
+ attr->reformat.id,
+ attr->reformat.size);
action_sz -= DR_STE_ACTION_DOUBLE_SZ;
action += DR_STE_ACTION_DOUBLE_SZ;
} else if (action_type_set[DR_ACTION_TYP_L2_TO_TNL_L3]) {
@@ -706,10 +572,10 @@ void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn,
}
d_action = action + DR_STE_ACTION_SINGLE_SZ;
- dr_ste_v1_set_encap_l3(last_ste,
- action, d_action,
- attr->reformat.id,
- attr->reformat.size);
+ ste_ctx->set_encap_l3(last_ste,
+ action, d_action,
+ attr->reformat.id,
+ attr->reformat.size);
action_sz -= DR_STE_ACTION_TRIPLE_SZ;
action += DR_STE_ACTION_TRIPLE_SZ;
} else if (action_type_set[DR_ACTION_TYP_INSERT_HDR]) {
@@ -776,7 +642,8 @@ void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn,
dr_ste_v1_set_hit_addr(last_ste, attr->final_icm_addr, 1);
}
-void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn,
+void dr_ste_v1_set_actions_rx(struct mlx5dr_ste_ctx *ste_ctx,
+ struct mlx5dr_domain *dmn,
u8 *action_type_set,
u32 actions_caps,
u8 *last_ste,
@@ -799,7 +666,7 @@ void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn,
allow_modify_hdr = false;
allow_ctr = false;
} else if (action_type_set[DR_ACTION_TYP_TNL_L2_TO_L2]) {
- dr_ste_v1_set_rx_decap(last_ste, action);
+ ste_ctx->set_rx_decap(last_ste, action);
action_sz -= DR_STE_ACTION_SINGLE_SZ;
action += DR_STE_ACTION_SINGLE_SZ;
allow_modify_hdr = false;
@@ -827,7 +694,7 @@ void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn,
action_sz = DR_STE_ACTION_TRIPLE_SZ;
}
- dr_ste_v1_set_pop_vlan(last_ste, action, attr->vlans.count);
+ ste_ctx->set_pop_vlan(last_ste, action, attr->vlans.count);
action_sz -= DR_STE_ACTION_SINGLE_SZ;
action += DR_STE_ACTION_SINGLE_SZ;
allow_ctr = false;
@@ -868,8 +735,8 @@ void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn,
last_ste, action);
action_sz = DR_STE_ACTION_TRIPLE_SZ;
}
- dr_ste_v1_set_push_vlan(last_ste, action,
- attr->vlans.headers[i]);
+ ste_ctx->set_push_vlan(last_ste, action,
+ attr->vlans.headers[i]);
action_sz -= DR_STE_ACTION_DOUBLE_SZ;
action += DR_STE_ACTION_DOUBLE_SZ;
}
@@ -895,9 +762,9 @@ void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn,
action = MLX5_ADDR_OF(ste_mask_and_match_v1, last_ste, action);
action_sz = DR_STE_ACTION_TRIPLE_SZ;
}
- dr_ste_v1_set_encap(last_ste, action,
- attr->reformat.id,
- attr->reformat.size);
+ ste_ctx->set_encap(last_ste, action,
+ attr->reformat.id,
+ attr->reformat.size);
action_sz -= DR_STE_ACTION_DOUBLE_SZ;
action += DR_STE_ACTION_DOUBLE_SZ;
allow_modify_hdr = false;
@@ -912,10 +779,10 @@ void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn,
d_action = action + DR_STE_ACTION_SINGLE_SZ;
- dr_ste_v1_set_encap_l3(last_ste,
- action, d_action,
- attr->reformat.id,
- attr->reformat.size);
+ ste_ctx->set_encap_l3(last_ste,
+ action, d_action,
+ attr->reformat.id,
+ attr->reformat.size);
action_sz -= DR_STE_ACTION_TRIPLE_SZ;
allow_modify_hdr = false;
} else if (action_type_set[DR_ACTION_TYP_INSERT_HDR]) {
@@ -1027,9 +894,6 @@ void dr_ste_v1_set_action_copy(u8 *d_action,
MLX5_SET(ste_double_action_copy_v1, d_action, source_right_shifter, src_shifter);
}
-#define DR_STE_DECAP_L3_ACTION_NUM 8
-#define DR_STE_L2_HDR_MAX_SZ 20
-
int dr_ste_v1_set_action_decap_l3_list(void *data,
u32 data_sz,
u8 *hw_action,
@@ -2330,7 +2194,12 @@ static struct mlx5dr_ste_ctx ste_ctx_v1 = {
.set_action_decap_l3_list = &dr_ste_v1_set_action_decap_l3_list,
.alloc_modify_hdr_chunk = &dr_ste_v1_alloc_modify_hdr_ptrn_arg,
.dealloc_modify_hdr_chunk = &dr_ste_v1_free_modify_hdr_ptrn_arg,
-
+ /* Actions bit set */
+ .set_encap = &dr_ste_v1_set_encap,
+ .set_push_vlan = &dr_ste_v1_set_push_vlan,
+ .set_pop_vlan = &dr_ste_v1_set_pop_vlan,
+ .set_rx_decap = &dr_ste_v1_set_rx_decap,
+ .set_encap_l3 = &dr_ste_v1_set_encap_l3,
/* Send */
.prepare_for_postsend = &dr_ste_v1_prepare_for_postsend,
};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.h
index e2fc69867088..a8d9e308d339 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v1.h
@@ -7,6 +7,138 @@
#include "dr_types.h"
#include "dr_ste.h"
+#define DR_STE_DECAP_L3_ACTION_NUM 8
+#define DR_STE_L2_HDR_MAX_SZ 20
+#define DR_STE_CALC_DFNR_TYPE(lookup_type, inner) \
+ ((inner) ? DR_STE_V1_LU_TYPE_##lookup_type##_I : \
+ DR_STE_V1_LU_TYPE_##lookup_type##_O)
+
+enum dr_ste_v1_entry_format {
+ DR_STE_V1_TYPE_BWC_BYTE = 0x0,
+ DR_STE_V1_TYPE_BWC_DW = 0x1,
+ DR_STE_V1_TYPE_MATCH = 0x2,
+ DR_STE_V1_TYPE_MATCH_RANGES = 0x7,
+};
+
+/* Lookup type is built from 2B: [ Definer mode 1B ][ Definer index 1B ] */
+enum {
+ DR_STE_V1_LU_TYPE_NOP = 0x0000,
+ DR_STE_V1_LU_TYPE_ETHL2_TNL = 0x0002,
+ DR_STE_V1_LU_TYPE_IBL3_EXT = 0x0102,
+ DR_STE_V1_LU_TYPE_ETHL2_O = 0x0003,
+ DR_STE_V1_LU_TYPE_IBL4 = 0x0103,
+ DR_STE_V1_LU_TYPE_ETHL2_I = 0x0004,
+ DR_STE_V1_LU_TYPE_SRC_QP_GVMI = 0x0104,
+ DR_STE_V1_LU_TYPE_ETHL2_SRC_O = 0x0005,
+ DR_STE_V1_LU_TYPE_ETHL2_HEADERS_O = 0x0105,
+ DR_STE_V1_LU_TYPE_ETHL2_SRC_I = 0x0006,
+ DR_STE_V1_LU_TYPE_ETHL2_HEADERS_I = 0x0106,
+ DR_STE_V1_LU_TYPE_ETHL3_IPV4_5_TUPLE_O = 0x0007,
+ DR_STE_V1_LU_TYPE_IPV6_DES_O = 0x0107,
+ DR_STE_V1_LU_TYPE_ETHL3_IPV4_5_TUPLE_I = 0x0008,
+ DR_STE_V1_LU_TYPE_IPV6_DES_I = 0x0108,
+ DR_STE_V1_LU_TYPE_ETHL4_O = 0x0009,
+ DR_STE_V1_LU_TYPE_IPV6_SRC_O = 0x0109,
+ DR_STE_V1_LU_TYPE_ETHL4_I = 0x000a,
+ DR_STE_V1_LU_TYPE_IPV6_SRC_I = 0x010a,
+ DR_STE_V1_LU_TYPE_ETHL2_SRC_DST_O = 0x000b,
+ DR_STE_V1_LU_TYPE_MPLS_O = 0x010b,
+ DR_STE_V1_LU_TYPE_ETHL2_SRC_DST_I = 0x000c,
+ DR_STE_V1_LU_TYPE_MPLS_I = 0x010c,
+ DR_STE_V1_LU_TYPE_ETHL3_IPV4_MISC_O = 0x000d,
+ DR_STE_V1_LU_TYPE_GRE = 0x010d,
+ DR_STE_V1_LU_TYPE_FLEX_PARSER_TNL_HEADER = 0x000e,
+ DR_STE_V1_LU_TYPE_GENERAL_PURPOSE = 0x010e,
+ DR_STE_V1_LU_TYPE_ETHL3_IPV4_MISC_I = 0x000f,
+ DR_STE_V1_LU_TYPE_STEERING_REGISTERS_0 = 0x010f,
+ DR_STE_V1_LU_TYPE_STEERING_REGISTERS_1 = 0x0110,
+ DR_STE_V1_LU_TYPE_FLEX_PARSER_OK = 0x0011,
+ DR_STE_V1_LU_TYPE_FLEX_PARSER_0 = 0x0111,
+ DR_STE_V1_LU_TYPE_FLEX_PARSER_1 = 0x0112,
+ DR_STE_V1_LU_TYPE_ETHL4_MISC_O = 0x0113,
+ DR_STE_V1_LU_TYPE_ETHL4_MISC_I = 0x0114,
+ DR_STE_V1_LU_TYPE_INVALID = 0x00ff,
+ DR_STE_V1_LU_TYPE_DONT_CARE = MLX5DR_STE_LU_TYPE_DONT_CARE,
+};
+
+enum dr_ste_v1_header_anchors {
+ DR_STE_HEADER_ANCHOR_START_OUTER = 0x00,
+ DR_STE_HEADER_ANCHOR_1ST_VLAN = 0x02,
+ DR_STE_HEADER_ANCHOR_IPV6_IPV4 = 0x07,
+ DR_STE_HEADER_ANCHOR_INNER_MAC = 0x13,
+ DR_STE_HEADER_ANCHOR_INNER_IPV6_IPV4 = 0x19,
+};
+
+enum dr_ste_v1_action_size {
+ DR_STE_ACTION_SINGLE_SZ = 4,
+ DR_STE_ACTION_DOUBLE_SZ = 8,
+ DR_STE_ACTION_TRIPLE_SZ = 12,
+};
+
+enum dr_ste_v1_action_insert_ptr_attr {
+ DR_STE_V1_ACTION_INSERT_PTR_ATTR_NONE = 0, /* Regular push header (e.g. push vlan) */
+ DR_STE_V1_ACTION_INSERT_PTR_ATTR_ENCAP = 1, /* Encapsulation / Tunneling */
+ DR_STE_V1_ACTION_INSERT_PTR_ATTR_ESP = 2, /* IPsec */
+};
+
+enum dr_ste_v1_action_id {
+ DR_STE_V1_ACTION_ID_NOP = 0x00,
+ DR_STE_V1_ACTION_ID_COPY = 0x05,
+ DR_STE_V1_ACTION_ID_SET = 0x06,
+ DR_STE_V1_ACTION_ID_ADD = 0x07,
+ DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE = 0x08,
+ DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER = 0x09,
+ DR_STE_V1_ACTION_ID_INSERT_INLINE = 0x0a,
+ DR_STE_V1_ACTION_ID_INSERT_POINTER = 0x0b,
+ DR_STE_V1_ACTION_ID_FLOW_TAG = 0x0c,
+ DR_STE_V1_ACTION_ID_QUEUE_ID_SEL = 0x0d,
+ DR_STE_V1_ACTION_ID_ACCELERATED_LIST = 0x0e,
+ DR_STE_V1_ACTION_ID_MODIFY_LIST = 0x0f,
+ DR_STE_V1_ACTION_ID_ASO = 0x12,
+ DR_STE_V1_ACTION_ID_TRAILER = 0x13,
+ DR_STE_V1_ACTION_ID_COUNTER_ID = 0x14,
+ DR_STE_V1_ACTION_ID_MAX = 0x21,
+ /* use for special cases */
+ DR_STE_V1_ACTION_ID_SPECIAL_ENCAP_L3 = 0x22,
+};
+
+enum {
+ DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_0 = 0x00,
+ DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_1 = 0x01,
+ DR_STE_V1_ACTION_MDFY_FLD_L2_OUT_2 = 0x02,
+ DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_0 = 0x08,
+ DR_STE_V1_ACTION_MDFY_FLD_SRC_L2_OUT_1 = 0x09,
+ DR_STE_V1_ACTION_MDFY_FLD_L3_OUT_0 = 0x0e,
+ DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_0 = 0x18,
+ DR_STE_V1_ACTION_MDFY_FLD_L4_OUT_1 = 0x19,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_0 = 0x40,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV4_OUT_1 = 0x41,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_0 = 0x44,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_1 = 0x45,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_2 = 0x46,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_DST_OUT_3 = 0x47,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_0 = 0x4c,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_1 = 0x4d,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_2 = 0x4e,
+ DR_STE_V1_ACTION_MDFY_FLD_IPV6_SRC_OUT_3 = 0x4f,
+ DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_0 = 0x5e,
+ DR_STE_V1_ACTION_MDFY_FLD_TCP_MISC_1 = 0x5f,
+ DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_0 = 0x6f,
+ DR_STE_V1_ACTION_MDFY_FLD_CFG_HDR_0_1 = 0x70,
+ DR_STE_V1_ACTION_MDFY_FLD_METADATA_2_CQE = 0x7b,
+ DR_STE_V1_ACTION_MDFY_FLD_GNRL_PURPOSE = 0x7c,
+ DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_0 = 0x8c,
+ DR_STE_V1_ACTION_MDFY_FLD_REGISTER_2_1 = 0x8d,
+ DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_0 = 0x8e,
+ DR_STE_V1_ACTION_MDFY_FLD_REGISTER_1_1 = 0x8f,
+ DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_0 = 0x90,
+ DR_STE_V1_ACTION_MDFY_FLD_REGISTER_0_1 = 0x91,
+};
+
+enum dr_ste_v1_aso_ctx_type {
+ DR_STE_V1_ASO_CTX_TYPE_POLICERS = 0x2,
+};
+
bool dr_ste_v1_is_miss_addr_set(u8 *hw_ste_p);
void dr_ste_v1_set_miss_addr(u8 *hw_ste_p, u64 miss_addr);
u64 dr_ste_v1_get_miss_addr(u8 *hw_ste_p);
@@ -17,11 +149,18 @@ u16 dr_ste_v1_get_next_lu_type(u8 *hw_ste_p);
void dr_ste_v1_set_hit_addr(u8 *hw_ste_p, u64 icm_addr, u32 ht_size);
void dr_ste_v1_init(u8 *hw_ste_p, u16 lu_type, bool is_rx, u16 gvmi);
void dr_ste_v1_prepare_for_postsend(u8 *hw_ste_p, u32 ste_size);
-void dr_ste_v1_set_actions_tx(struct mlx5dr_domain *dmn, u8 *action_type_set,
- u32 actions_caps, u8 *last_ste,
+void dr_ste_v1_set_reparse(u8 *hw_ste_p);
+void dr_ste_v1_set_encap(u8 *hw_ste_p, u8 *d_action, u32 reformat_id, int size);
+void dr_ste_v1_set_push_vlan(u8 *hw_ste_p, u8 *d_action, u32 vlan_hdr);
+void dr_ste_v1_set_pop_vlan(u8 *hw_ste_p, u8 *s_action, u8 vlans_num);
+void dr_ste_v1_set_encap_l3(u8 *hw_ste_p, u8 *frst_s_action, u8 *scnd_d_action,
+ u32 reformat_id, int size);
+void dr_ste_v1_set_rx_decap(u8 *hw_ste_p, u8 *s_action);
+void dr_ste_v1_set_actions_tx(struct mlx5dr_ste_ctx *ste_ctx, struct mlx5dr_domain *dmn,
+ u8 *action_type_set, u32 actions_caps, u8 *last_ste,
struct mlx5dr_ste_actions_attr *attr, u32 *added_stes);
-void dr_ste_v1_set_actions_rx(struct mlx5dr_domain *dmn, u8 *action_type_set,
- u32 actions_caps, u8 *last_ste,
+void dr_ste_v1_set_actions_rx(struct mlx5dr_ste_ctx *ste_ctx, struct mlx5dr_domain *dmn,
+ u8 *action_type_set, u32 actions_caps, u8 *last_ste,
struct mlx5dr_ste_actions_attr *attr, u32 *added_stes);
void dr_ste_v1_set_action_set(u8 *d_action, u8 hw_field, u8 shifter,
u8 length, u32 data);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.c
index 808b013cf48c..0882dba0f64b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.c
@@ -2,167 +2,7 @@
/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
#include "dr_ste_v1.h"
-
-enum {
- DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_0 = 0x00,
- DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1 = 0x01,
- DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_2 = 0x02,
- DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_0 = 0x08,
- DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_1 = 0x09,
- DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0 = 0x0e,
- DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0 = 0x18,
- DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_1 = 0x19,
- DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_0 = 0x40,
- DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_1 = 0x41,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_0 = 0x44,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_1 = 0x45,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_2 = 0x46,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_3 = 0x47,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_0 = 0x4c,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_1 = 0x4d,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_2 = 0x4e,
- DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_3 = 0x4f,
- DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_0 = 0x5e,
- DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_1 = 0x5f,
- DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_0 = 0x6f,
- DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_1 = 0x70,
- DR_STE_V2_ACTION_MDFY_FLD_METADATA_2_CQE = 0x7b,
- DR_STE_V2_ACTION_MDFY_FLD_GNRL_PURPOSE = 0x7c,
- DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_0 = 0x90,
- DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_1 = 0x91,
- DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_0 = 0x92,
- DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_1 = 0x93,
- DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_0 = 0x94,
- DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_1 = 0x95,
-};
-
-static const struct mlx5dr_ste_action_modify_field dr_ste_v2_action_modify_field_arr[] = {
- [MLX5_ACTION_IN_FIELD_OUT_SMAC_47_16] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_0, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_SMAC_15_0] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_1, .start = 16, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_ETHERTYPE] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1, .start = 0, .end = 15,
- },
- [MLX5_ACTION_IN_FIELD_OUT_DMAC_47_16] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_0, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_DMAC_15_0] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1, .start = 16, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_IP_DSCP] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 18, .end = 23,
- },
- [MLX5_ACTION_IN_FIELD_OUT_TCP_FLAGS] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_1, .start = 16, .end = 24,
- .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
- },
- [MLX5_ACTION_IN_FIELD_OUT_TCP_SPORT] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
- .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
- },
- [MLX5_ACTION_IN_FIELD_OUT_TCP_DPORT] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
- .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
- },
- [MLX5_ACTION_IN_FIELD_OUT_IP_TTL] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
- },
- [MLX5_ACTION_IN_FIELD_OUT_IPV6_HOPLIMIT] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_UDP_SPORT] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
- .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
- },
- [MLX5_ACTION_IN_FIELD_OUT_UDP_DPORT] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
- .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
- },
- [MLX5_ACTION_IN_FIELD_OUT_SIPV6_127_96] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_0, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_SIPV6_95_64] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_1, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_SIPV6_63_32] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_2, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_SIPV6_31_0] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_3, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_DIPV6_127_96] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_0, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_DIPV6_95_64] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_1, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_DIPV6_63_32] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_2, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_DIPV6_31_0] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_3, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
- },
- [MLX5_ACTION_IN_FIELD_OUT_SIPV4] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_0, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
- },
- [MLX5_ACTION_IN_FIELD_OUT_DIPV4] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_1, .start = 0, .end = 31,
- .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_A] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_GNRL_PURPOSE, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_B] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_METADATA_2_CQE, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_C_0] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_0, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_C_1] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_1, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_C_2] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_0, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_C_3] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_1, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_C_4] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_0, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_METADATA_REG_C_5] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_1, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_TCP_SEQ_NUM] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_0, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_TCP_ACK_NUM] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_1, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_FIRST_VID] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_2, .start = 0, .end = 15,
- },
- [MLX5_ACTION_IN_FIELD_OUT_EMD_31_0] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_1, .start = 0, .end = 31,
- },
- [MLX5_ACTION_IN_FIELD_OUT_EMD_47_32] = {
- .hw_field = DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_0, .start = 0, .end = 15,
- },
-};
+#include "dr_ste_v2.h"
static struct mlx5dr_ste_ctx ste_ctx_v2 = {
/* Builders */
@@ -223,7 +63,12 @@ static struct mlx5dr_ste_ctx ste_ctx_v2 = {
.set_action_decap_l3_list = &dr_ste_v1_set_action_decap_l3_list,
.alloc_modify_hdr_chunk = &dr_ste_v1_alloc_modify_hdr_ptrn_arg,
.dealloc_modify_hdr_chunk = &dr_ste_v1_free_modify_hdr_ptrn_arg,
-
+ /* Actions bit set */
+ .set_encap = &dr_ste_v1_set_encap,
+ .set_push_vlan = &dr_ste_v1_set_push_vlan,
+ .set_pop_vlan = &dr_ste_v1_set_pop_vlan,
+ .set_rx_decap = &dr_ste_v1_set_rx_decap,
+ .set_encap_l3 = &dr_ste_v1_set_encap_l3,
/* Send */
.prepare_for_postsend = &dr_ste_v1_prepare_for_postsend,
};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.h
new file mode 100644
index 000000000000..d853fde49cfc
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v2.h
@@ -0,0 +1,168 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#ifndef _DR_STE_V2_
+#define _DR_STE_V2_
+
+enum {
+ DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_0 = 0x00,
+ DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1 = 0x01,
+ DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_2 = 0x02,
+ DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_0 = 0x08,
+ DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_1 = 0x09,
+ DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0 = 0x0e,
+ DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0 = 0x18,
+ DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_1 = 0x19,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_0 = 0x40,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_1 = 0x41,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_0 = 0x44,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_1 = 0x45,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_2 = 0x46,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_3 = 0x47,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_0 = 0x4c,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_1 = 0x4d,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_2 = 0x4e,
+ DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_3 = 0x4f,
+ DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_0 = 0x5e,
+ DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_1 = 0x5f,
+ DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_0 = 0x6f,
+ DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_1 = 0x70,
+ DR_STE_V2_ACTION_MDFY_FLD_METADATA_2_CQE = 0x7b,
+ DR_STE_V2_ACTION_MDFY_FLD_GNRL_PURPOSE = 0x7c,
+ DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_0 = 0x90,
+ DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_1 = 0x91,
+ DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_0 = 0x92,
+ DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_1 = 0x93,
+ DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_0 = 0x94,
+ DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_1 = 0x95,
+};
+
+static const struct mlx5dr_ste_action_modify_field dr_ste_v2_action_modify_field_arr[] = {
+ [MLX5_ACTION_IN_FIELD_OUT_SMAC_47_16] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_0, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_SMAC_15_0] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_SRC_L2_OUT_1, .start = 16, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_ETHERTYPE] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1, .start = 0, .end = 15,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_DMAC_47_16] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_0, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_DMAC_15_0] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_1, .start = 16, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_IP_DSCP] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 18, .end = 23,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_TCP_FLAGS] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_1, .start = 16, .end = 24,
+ .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_TCP_SPORT] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
+ .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_TCP_DPORT] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
+ .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_TCP,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_IP_TTL] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_IPV6_HOPLIMIT] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L3_OUT_0, .start = 8, .end = 15,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_UDP_SPORT] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 16, .end = 31,
+ .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_UDP_DPORT] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L4_OUT_0, .start = 0, .end = 15,
+ .l4_type = DR_STE_ACTION_MDFY_TYPE_L4_UDP,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_SIPV6_127_96] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_0, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_SIPV6_95_64] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_1, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_SIPV6_63_32] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_2, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_SIPV6_31_0] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_SRC_OUT_3, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_DIPV6_127_96] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_0, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_DIPV6_95_64] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_1, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_DIPV6_63_32] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_2, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_DIPV6_31_0] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV6_DST_OUT_3, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV6,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_SIPV4] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_0, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_DIPV4] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_IPV4_OUT_1, .start = 0, .end = 31,
+ .l3_type = DR_STE_ACTION_MDFY_TYPE_L3_IPV4,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_A] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_GNRL_PURPOSE, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_B] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_METADATA_2_CQE, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_C_0] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_0, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_C_1] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_0_1, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_C_2] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_0, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_C_3] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_1_1, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_C_4] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_0, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_METADATA_REG_C_5] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_REGISTER_2_1, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_TCP_SEQ_NUM] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_0, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_TCP_ACK_NUM] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_TCP_MISC_1, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_FIRST_VID] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_L2_OUT_2, .start = 0, .end = 15,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_EMD_31_0] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_1, .start = 0, .end = 31,
+ },
+ [MLX5_ACTION_IN_FIELD_OUT_EMD_47_32] = {
+ .hw_field = DR_STE_V2_ACTION_MDFY_FLD_CFG_HDR_0_0, .start = 0, .end = 15,
+ },
+};
+
+#endif /* _DR_STE_V2_ */
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (8 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 09/12] net/mlx5: DR, expand SWS STE callbacks and consolidate common structs Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-12 17:31 ` Simon Horman
2024-12-11 13:42 ` [PATCH net-next 11/12] net/mlx5: Remove PTM support log message Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 12/12] net/mlx5: fs, Add support for RDMA RX steering over IB link layer Tariq Toukan
11 siblings, 1 reply; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Itamar Gozlan,
Yevgeny Kliteynik, Tariq Toukan
From: Itamar Gozlan <igozlan@nvidia.com>
Add support for a new steering format version that is implemented by
ConnectX-8.
Except for several differences, the STEv3 is identical to STEv2, so
for most callbacks STEv3 context struct will call STEv2 functions.
Signed-off-by: Itamar Gozlan <igozlan@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/Makefile | 1 +
.../mlx5/core/steering/sws/dr_domain.c | 2 +-
.../mellanox/mlx5/core/steering/sws/dr_ste.c | 2 +
.../mellanox/mlx5/core/steering/sws/dr_ste.h | 1 +
.../mlx5/core/steering/sws/dr_ste_v3.c | 221 ++++++++++++++++++
.../mlx5/core/steering/sws/mlx5_ifc_dr.h | 40 ++++
.../mellanox/mlx5/core/steering/sws/mlx5dr.h | 2 +-
7 files changed, 267 insertions(+), 2 deletions(-)
create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v3.c
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
index 79fe09de0a9f..10a763e668ed 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -123,6 +123,7 @@ mlx5_core-$(CONFIG_MLX5_SW_STEERING) += steering/sws/dr_domain.o \
steering/sws/dr_ste_v0.o \
steering/sws/dr_ste_v1.o \
steering/sws/dr_ste_v2.o \
+ steering/sws/dr_ste_v3.o \
steering/sws/dr_cmd.o \
steering/sws/dr_fw.o \
steering/sws/dr_action.o \
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
index 3d74109f8230..bd361ba6658c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
@@ -8,7 +8,7 @@
#define DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, dmn_type) \
((dmn)->info.caps.dmn_type##_sw_owner || \
((dmn)->info.caps.dmn_type##_sw_owner_v2 && \
- (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_7))
+ (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_8))
bool mlx5dr_domain_is_support_ptrn_arg(struct mlx5dr_domain *dmn)
{
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c
index 01ba8eae2983..c8b8ff80c7c7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.c
@@ -1458,6 +1458,8 @@ struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx(u8 version)
return mlx5dr_ste_get_ctx_v1();
else if (version == MLX5_STEERING_FORMAT_CONNECTX_7)
return mlx5dr_ste_get_ctx_v2();
+ else if (version == MLX5_STEERING_FORMAT_CONNECTX_8)
+ return mlx5dr_ste_get_ctx_v3();
return NULL;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h
index b6ec8d30d990..5f409dc30aca 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste.h
@@ -217,5 +217,6 @@ struct mlx5dr_ste_ctx {
struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v0(void);
struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v1(void);
struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v2(void);
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v3(void);
#endif /* _DR_STE_ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v3.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v3.c
new file mode 100644
index 000000000000..cc60ce1d274e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v3.c
@@ -0,0 +1,221 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
+
+#include "dr_ste_v1.h"
+#include "dr_ste_v2.h"
+
+static void dr_ste_v3_set_encap(u8 *hw_ste_p, u8 *d_action,
+ u32 reformat_id, int size)
+{
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, d_action, action_id,
+ DR_STE_V1_ACTION_ID_INSERT_POINTER);
+ /* The hardware expects here size in words (2 byte) */
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, d_action, size, size / 2);
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, d_action, pointer, reformat_id);
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, d_action, attributes,
+ DR_STE_V1_ACTION_INSERT_PTR_ATTR_ENCAP);
+ dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v3_set_push_vlan(u8 *ste, u8 *d_action,
+ u32 vlan_hdr)
+{
+ MLX5_SET(ste_double_action_insert_with_inline_v3, d_action, action_id,
+ DR_STE_V1_ACTION_ID_INSERT_INLINE);
+ /* The hardware expects here offset to vlan header in words (2 byte) */
+ MLX5_SET(ste_double_action_insert_with_inline_v3, d_action, start_offset,
+ HDR_LEN_L2_MACS >> 1);
+ MLX5_SET(ste_double_action_insert_with_inline_v3, d_action, inline_data, vlan_hdr);
+ dr_ste_v1_set_reparse(ste);
+}
+
+static void dr_ste_v3_set_pop_vlan(u8 *hw_ste_p, u8 *s_action,
+ u8 vlans_num)
+{
+ MLX5_SET(ste_single_action_remove_header_size_v3, s_action,
+ action_id, DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE);
+ MLX5_SET(ste_single_action_remove_header_size_v3, s_action,
+ start_anchor, DR_STE_HEADER_ANCHOR_1ST_VLAN);
+ /* The hardware expects here size in words (2 byte) */
+ MLX5_SET(ste_single_action_remove_header_size_v3, s_action,
+ remove_size, (HDR_LEN_L2_VLAN >> 1) * vlans_num);
+
+ dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v3_set_encap_l3(u8 *hw_ste_p,
+ u8 *frst_s_action,
+ u8 *scnd_d_action,
+ u32 reformat_id,
+ int size)
+{
+ /* Remove L2 headers */
+ MLX5_SET(ste_single_action_remove_header_v3, frst_s_action, action_id,
+ DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER);
+ MLX5_SET(ste_single_action_remove_header_v3, frst_s_action, end_anchor,
+ DR_STE_HEADER_ANCHOR_IPV6_IPV4);
+
+ /* Encapsulate with given reformat ID */
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, scnd_d_action, action_id,
+ DR_STE_V1_ACTION_ID_INSERT_POINTER);
+ /* The hardware expects here size in words (2 byte) */
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, scnd_d_action, size, size / 2);
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, scnd_d_action, pointer, reformat_id);
+ MLX5_SET(ste_double_action_insert_with_ptr_v3, scnd_d_action, attributes,
+ DR_STE_V1_ACTION_INSERT_PTR_ATTR_ENCAP);
+
+ dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static void dr_ste_v3_set_rx_decap(u8 *hw_ste_p, u8 *s_action)
+{
+ MLX5_SET(ste_single_action_remove_header_v3, s_action, action_id,
+ DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER);
+ MLX5_SET(ste_single_action_remove_header_v3, s_action, decap, 1);
+ MLX5_SET(ste_single_action_remove_header_v3, s_action, vni_to_cqe, 1);
+ MLX5_SET(ste_single_action_remove_header_v3, s_action, end_anchor,
+ DR_STE_HEADER_ANCHOR_INNER_MAC);
+
+ dr_ste_v1_set_reparse(hw_ste_p);
+}
+
+static int
+dr_ste_v3_set_action_decap_l3_list(void *data, u32 data_sz,
+ u8 *hw_action, u32 hw_action_sz,
+ uint16_t *used_hw_action_num)
+{
+ u8 padded_data[DR_STE_L2_HDR_MAX_SZ] = {};
+ void *data_ptr = padded_data;
+ u16 used_actions = 0;
+ u32 inline_data_sz;
+ u32 i;
+
+ if (hw_action_sz / DR_STE_ACTION_DOUBLE_SZ < DR_STE_DECAP_L3_ACTION_NUM)
+ return -EINVAL;
+
+ inline_data_sz =
+ MLX5_FLD_SZ_BYTES(ste_double_action_insert_with_inline_v3, inline_data);
+
+ /* Add an alignment padding */
+ memcpy(padded_data + data_sz % inline_data_sz, data, data_sz);
+
+ /* Remove L2L3 outer headers */
+ MLX5_SET(ste_single_action_remove_header_v3, hw_action, action_id,
+ DR_STE_V1_ACTION_ID_REMOVE_HEADER_TO_HEADER);
+ MLX5_SET(ste_single_action_remove_header_v3, hw_action, decap, 1);
+ MLX5_SET(ste_single_action_remove_header_v3, hw_action, vni_to_cqe, 1);
+ MLX5_SET(ste_single_action_remove_header_v3, hw_action, end_anchor,
+ DR_STE_HEADER_ANCHOR_INNER_IPV6_IPV4);
+ hw_action += DR_STE_ACTION_DOUBLE_SZ;
+ used_actions++; /* Remove and NOP are a single double action */
+
+ /* Point to the last dword of the header */
+ data_ptr += (data_sz / inline_data_sz) * inline_data_sz;
+
+ /* Add the new header using inline action 4Byte at a time, the header
+ * is added in reversed order to the beginning of the packet to avoid
+ * incorrect parsing by the HW. Since header is 14B or 18B an extra
+ * two bytes are padded and later removed.
+ */
+ for (i = 0; i < data_sz / inline_data_sz + 1; i++) {
+ void *addr_inline;
+
+ MLX5_SET(ste_double_action_insert_with_inline_v3, hw_action, action_id,
+ DR_STE_V1_ACTION_ID_INSERT_INLINE);
+ /* The hardware expects here offset to words (2 bytes) */
+ MLX5_SET(ste_double_action_insert_with_inline_v3, hw_action, start_offset, 0);
+
+ /* Copy bytes one by one to avoid endianness problem */
+ addr_inline = MLX5_ADDR_OF(ste_double_action_insert_with_inline_v3,
+ hw_action, inline_data);
+ memcpy(addr_inline, data_ptr - i * inline_data_sz, inline_data_sz);
+ hw_action += DR_STE_ACTION_DOUBLE_SZ;
+ used_actions++;
+ }
+
+ /* Remove first 2 extra bytes */
+ MLX5_SET(ste_single_action_remove_header_size_v3, hw_action, action_id,
+ DR_STE_V1_ACTION_ID_REMOVE_BY_SIZE);
+ MLX5_SET(ste_single_action_remove_header_size_v3, hw_action, start_offset, 0);
+ /* The hardware expects here size in words (2 bytes) */
+ MLX5_SET(ste_single_action_remove_header_size_v3, hw_action, remove_size, 1);
+ used_actions++;
+
+ *used_hw_action_num = used_actions;
+
+ return 0;
+}
+
+static struct mlx5dr_ste_ctx ste_ctx_v3 = {
+ /* Builders */
+ .build_eth_l2_src_dst_init = &dr_ste_v1_build_eth_l2_src_dst_init,
+ .build_eth_l3_ipv6_src_init = &dr_ste_v1_build_eth_l3_ipv6_src_init,
+ .build_eth_l3_ipv6_dst_init = &dr_ste_v1_build_eth_l3_ipv6_dst_init,
+ .build_eth_l3_ipv4_5_tuple_init = &dr_ste_v1_build_eth_l3_ipv4_5_tuple_init,
+ .build_eth_l2_src_init = &dr_ste_v1_build_eth_l2_src_init,
+ .build_eth_l2_dst_init = &dr_ste_v1_build_eth_l2_dst_init,
+ .build_eth_l2_tnl_init = &dr_ste_v1_build_eth_l2_tnl_init,
+ .build_eth_l3_ipv4_misc_init = &dr_ste_v1_build_eth_l3_ipv4_misc_init,
+ .build_eth_ipv6_l3_l4_init = &dr_ste_v1_build_eth_ipv6_l3_l4_init,
+ .build_mpls_init = &dr_ste_v1_build_mpls_init,
+ .build_tnl_gre_init = &dr_ste_v1_build_tnl_gre_init,
+ .build_tnl_mpls_init = &dr_ste_v1_build_tnl_mpls_init,
+ .build_tnl_mpls_over_udp_init = &dr_ste_v1_build_tnl_mpls_over_udp_init,
+ .build_tnl_mpls_over_gre_init = &dr_ste_v1_build_tnl_mpls_over_gre_init,
+ .build_icmp_init = &dr_ste_v1_build_icmp_init,
+ .build_general_purpose_init = &dr_ste_v1_build_general_purpose_init,
+ .build_eth_l4_misc_init = &dr_ste_v1_build_eth_l4_misc_init,
+ .build_tnl_vxlan_gpe_init = &dr_ste_v1_build_flex_parser_tnl_vxlan_gpe_init,
+ .build_tnl_geneve_init = &dr_ste_v1_build_flex_parser_tnl_geneve_init,
+ .build_tnl_geneve_tlv_opt_init = &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_init,
+ .build_tnl_geneve_tlv_opt_exist_init =
+ &dr_ste_v1_build_flex_parser_tnl_geneve_tlv_opt_exist_init,
+ .build_register_0_init = &dr_ste_v1_build_register_0_init,
+ .build_register_1_init = &dr_ste_v1_build_register_1_init,
+ .build_src_gvmi_qpn_init = &dr_ste_v1_build_src_gvmi_qpn_init,
+ .build_flex_parser_0_init = &dr_ste_v1_build_flex_parser_0_init,
+ .build_flex_parser_1_init = &dr_ste_v1_build_flex_parser_1_init,
+ .build_tnl_gtpu_init = &dr_ste_v1_build_flex_parser_tnl_gtpu_init,
+ .build_tnl_header_0_1_init = &dr_ste_v1_build_tnl_header_0_1_init,
+ .build_tnl_gtpu_flex_parser_0_init = &dr_ste_v1_build_tnl_gtpu_flex_parser_0_init,
+ .build_tnl_gtpu_flex_parser_1_init = &dr_ste_v1_build_tnl_gtpu_flex_parser_1_init,
+
+ /* Getters and Setters */
+ .ste_init = &dr_ste_v1_init,
+ .set_next_lu_type = &dr_ste_v1_set_next_lu_type,
+ .get_next_lu_type = &dr_ste_v1_get_next_lu_type,
+ .is_miss_addr_set = &dr_ste_v1_is_miss_addr_set,
+ .set_miss_addr = &dr_ste_v1_set_miss_addr,
+ .get_miss_addr = &dr_ste_v1_get_miss_addr,
+ .set_hit_addr = &dr_ste_v1_set_hit_addr,
+ .set_byte_mask = &dr_ste_v1_set_byte_mask,
+ .get_byte_mask = &dr_ste_v1_get_byte_mask,
+
+ /* Actions */
+ .actions_caps = DR_STE_CTX_ACTION_CAP_TX_POP |
+ DR_STE_CTX_ACTION_CAP_RX_PUSH |
+ DR_STE_CTX_ACTION_CAP_RX_ENCAP,
+ .set_actions_rx = &dr_ste_v1_set_actions_rx,
+ .set_actions_tx = &dr_ste_v1_set_actions_tx,
+ .modify_field_arr_sz = ARRAY_SIZE(dr_ste_v2_action_modify_field_arr),
+ .modify_field_arr = dr_ste_v2_action_modify_field_arr,
+ .set_action_set = &dr_ste_v1_set_action_set,
+ .set_action_add = &dr_ste_v1_set_action_add,
+ .set_action_copy = &dr_ste_v1_set_action_copy,
+ .set_action_decap_l3_list = &dr_ste_v3_set_action_decap_l3_list,
+ .alloc_modify_hdr_chunk = &dr_ste_v1_alloc_modify_hdr_ptrn_arg,
+ .dealloc_modify_hdr_chunk = &dr_ste_v1_free_modify_hdr_ptrn_arg,
+ /* Actions bit set */
+ .set_encap = &dr_ste_v3_set_encap,
+ .set_push_vlan = &dr_ste_v3_set_push_vlan,
+ .set_pop_vlan = &dr_ste_v3_set_pop_vlan,
+ .set_rx_decap = &dr_ste_v3_set_rx_decap,
+ .set_encap_l3 = &dr_ste_v3_set_encap_l3,
+ /* Send */
+ .prepare_for_postsend = &dr_ste_v1_prepare_for_postsend,
+};
+
+struct mlx5dr_ste_ctx *mlx5dr_ste_get_ctx_v3(void)
+{
+ return &ste_ctx_v3;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5_ifc_dr.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5_ifc_dr.h
index fb078fa0f0cc..898c3618ff26 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5_ifc_dr.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5_ifc_dr.h
@@ -600,4 +600,44 @@ struct mlx5_ifc_ste_double_action_aso_v1_bits {
};
};
+struct mlx5_ifc_ste_single_action_remove_header_v3_bits {
+ u8 action_id[0x8];
+ u8 start_anchor[0x7];
+ u8 end_anchor[0x7];
+ u8 reserved_at_16[0x1];
+ u8 outer_l4_remove[0x1];
+ u8 reserved_at_18[0x4];
+ u8 decap[0x1];
+ u8 vni_to_cqe[0x1];
+ u8 qos_profile[0x2];
+};
+
+struct mlx5_ifc_ste_single_action_remove_header_size_v3_bits {
+ u8 action_id[0x8];
+ u8 start_anchor[0x7];
+ u8 start_offset[0x8];
+ u8 outer_l4_remove[0x1];
+ u8 reserved_at_18[0x2];
+ u8 remove_size[0x6];
+};
+
+struct mlx5_ifc_ste_double_action_insert_with_inline_v3_bits {
+ u8 action_id[0x8];
+ u8 start_anchor[0x7];
+ u8 start_offset[0x8];
+ u8 reserved_at_17[0x9];
+
+ u8 inline_data[0x20];
+};
+
+struct mlx5_ifc_ste_double_action_insert_with_ptr_v3_bits {
+ u8 action_id[0x8];
+ u8 start_anchor[0x7];
+ u8 start_offset[0x8];
+ u8 size[0x6];
+ u8 attributes[0x3];
+
+ u8 pointer[0x20];
+};
+
#endif /* MLX5_IFC_DR_H */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5dr.h b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5dr.h
index 3ac7dc67509f..0bb3724c10c2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5dr.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/mlx5dr.h
@@ -160,7 +160,7 @@ mlx5dr_is_supported(struct mlx5_core_dev *dev)
(MLX5_CAP_ESW_FLOWTABLE_FDB(dev, sw_owner) ||
(MLX5_CAP_ESW_FLOWTABLE_FDB(dev, sw_owner_v2) &&
(MLX5_CAP_GEN(dev, steering_format_version) <=
- MLX5_STEERING_FORMAT_CONNECTX_7)));
+ MLX5_STEERING_FORMAT_CONNECTX_8)));
}
/* buddy functions & structure */
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* Re: [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-11 13:42 ` [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering Tariq Toukan
@ 2024-12-12 17:31 ` Simon Horman
2024-12-12 18:31 ` Tariq Toukan
0 siblings, 1 reply; 34+ messages in thread
From: Simon Horman @ 2024-12-12 17:31 UTC (permalink / raw)
To: Tariq Toukan
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Itamar Gozlan, Yevgeny Kliteynik
On Wed, Dec 11, 2024 at 03:42:21PM +0200, Tariq Toukan wrote:
> From: Itamar Gozlan <igozlan@nvidia.com>
>
> Add support for a new steering format version that is implemented by
> ConnectX-8.
> Except for several differences, the STEv3 is identical to STEv2, so
> for most callbacks STEv3 context struct will call STEv2 functions.
>
> Signed-off-by: Itamar Gozlan <igozlan@nvidia.com>
> Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> ---
> .../net/ethernet/mellanox/mlx5/core/Makefile | 1 +
> .../mlx5/core/steering/sws/dr_domain.c | 2 +-
> .../mellanox/mlx5/core/steering/sws/dr_ste.c | 2 +
> .../mellanox/mlx5/core/steering/sws/dr_ste.h | 1 +
> .../mlx5/core/steering/sws/dr_ste_v3.c | 221 ++++++++++++++++++
> .../mlx5/core/steering/sws/mlx5_ifc_dr.h | 40 ++++
> .../mellanox/mlx5/core/steering/sws/mlx5dr.h | 2 +-
> 7 files changed, 267 insertions(+), 2 deletions(-)
> create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v3.c
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
> index 79fe09de0a9f..10a763e668ed 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
> @@ -123,6 +123,7 @@ mlx5_core-$(CONFIG_MLX5_SW_STEERING) += steering/sws/dr_domain.o \
> steering/sws/dr_ste_v0.o \
> steering/sws/dr_ste_v1.o \
> steering/sws/dr_ste_v2.o \
> + steering/sws/dr_ste_v3.o \
> steering/sws/dr_cmd.o \
> steering/sws/dr_fw.o \
> steering/sws/dr_action.o \
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> index 3d74109f8230..bd361ba6658c 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> @@ -8,7 +8,7 @@
> #define DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, dmn_type) \
> ((dmn)->info.caps.dmn_type##_sw_owner || \
> ((dmn)->info.caps.dmn_type##_sw_owner_v2 && \
> - (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_7))
> + (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_8))
A definition for MLX5_STEERING_FORMAT_CONNECTX_8 seems to be missing
from this patch.
>
> bool mlx5dr_domain_is_support_ptrn_arg(struct mlx5dr_domain *dmn)
> {
...
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-12 17:31 ` Simon Horman
@ 2024-12-12 18:31 ` Tariq Toukan
2024-12-13 1:11 ` Jakub Kicinski
2024-12-13 10:31 ` Simon Horman
0 siblings, 2 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-12 18:31 UTC (permalink / raw)
To: Simon Horman, Tariq Toukan
Cc: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky, netdev, Saeed Mahameed,
Gal Pressman, linux-rdma, Itamar Gozlan, Yevgeny Kliteynik
On 12/12/2024 19:31, Simon Horman wrote:
> On Wed, Dec 11, 2024 at 03:42:21PM +0200, Tariq Toukan wrote:
>> From: Itamar Gozlan <igozlan@nvidia.com>
>>
>> Add support for a new steering format version that is implemented by
>> ConnectX-8.
>> Except for several differences, the STEv3 is identical to STEv2, so
>> for most callbacks STEv3 context struct will call STEv2 functions.
>>
>> Signed-off-by: Itamar Gozlan <igozlan@nvidia.com>
>> Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
>> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
>> ---
>> .../net/ethernet/mellanox/mlx5/core/Makefile | 1 +
>> .../mlx5/core/steering/sws/dr_domain.c | 2 +-
>> .../mellanox/mlx5/core/steering/sws/dr_ste.c | 2 +
>> .../mellanox/mlx5/core/steering/sws/dr_ste.h | 1 +
>> .../mlx5/core/steering/sws/dr_ste_v3.c | 221 ++++++++++++++++++
>> .../mlx5/core/steering/sws/mlx5_ifc_dr.h | 40 ++++
>> .../mellanox/mlx5/core/steering/sws/mlx5dr.h | 2 +-
>> 7 files changed, 267 insertions(+), 2 deletions(-)
>> create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v3.c
>>
>> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
>> index 79fe09de0a9f..10a763e668ed 100644
>> --- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
>> +++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
>> @@ -123,6 +123,7 @@ mlx5_core-$(CONFIG_MLX5_SW_STEERING) += steering/sws/dr_domain.o \
>> steering/sws/dr_ste_v0.o \
>> steering/sws/dr_ste_v1.o \
>> steering/sws/dr_ste_v2.o \
>> + steering/sws/dr_ste_v3.o \
>> steering/sws/dr_cmd.o \
>> steering/sws/dr_fw.o \
>> steering/sws/dr_action.o \
>> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
>> index 3d74109f8230..bd361ba6658c 100644
>> --- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
>> +++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
>> @@ -8,7 +8,7 @@
>> #define DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, dmn_type) \
>> ((dmn)->info.caps.dmn_type##_sw_owner || \
>> ((dmn)->info.caps.dmn_type##_sw_owner_v2 && \
>> - (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_7))
>> + (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_8))
>
> A definition for MLX5_STEERING_FORMAT_CONNECTX_8 seems to be missing
> from this patch.
>
Should be pulled from mlx5-next, as described in the cover letter.
Copying here for your convenience:
It requires pulling 4 IFC patches that were applied to
mlx5-next:
https://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux.git/log/?h=mlx5-next
>>
>> bool mlx5dr_domain_is_support_ptrn_arg(struct mlx5dr_domain *dmn)
>> {
>
> ...
>
^ permalink raw reply [flat|nested] 34+ messages in thread* Re: [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-12 18:31 ` Tariq Toukan
@ 2024-12-13 1:11 ` Jakub Kicinski
2024-12-15 6:25 ` Tariq Toukan
2024-12-13 10:31 ` Simon Horman
1 sibling, 1 reply; 34+ messages in thread
From: Jakub Kicinski @ 2024-12-13 1:11 UTC (permalink / raw)
To: Tariq Toukan
Cc: Simon Horman, Tariq Toukan, David S. Miller, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Itamar Gozlan,
Yevgeny Kliteynik
On Thu, 12 Dec 2024 20:31:30 +0200 Tariq Toukan wrote:
> It requires pulling 4 IFC patches that were applied to
> mlx5-next:
> https://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux.git/log/?h=mlx5-next
What do you expect we'll do with this series?
If you expect it to be set to Awaiting Upstream - could you make sure
that the cover letter has "mlx5-next" in the subject? That will makes
it easier to automate in patchwork.
If you expect the series to be applied / merged - LMK, I can try
to explain why that's impossible..
^ permalink raw reply [flat|nested] 34+ messages in thread
* Re: [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-13 1:11 ` Jakub Kicinski
@ 2024-12-15 6:25 ` Tariq Toukan
2024-12-15 21:12 ` Jakub Kicinski
0 siblings, 1 reply; 34+ messages in thread
From: Tariq Toukan @ 2024-12-15 6:25 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Simon Horman, Tariq Toukan, David S. Miller, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Itamar Gozlan,
Yevgeny Kliteynik
On 13/12/2024 3:11, Jakub Kicinski wrote:
> On Thu, 12 Dec 2024 20:31:30 +0200 Tariq Toukan wrote:
>> It requires pulling 4 IFC patches that were applied to
>> mlx5-next:
>> https://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux.git/log/?h=mlx5-next
>
> What do you expect we'll do with this series?
>
> If you expect it to be set to Awaiting Upstream - could you make sure
> that the cover letter has "mlx5-next" in the subject? That will makes
> it easier to automate in patchwork.
>
The relevant patches have mlx5-next in their topic.
Should the cover letter as well?
What about other non-IFC patches, keep them with net-next?
> If you expect the series to be applied / merged - LMK, I can try
> to explain why that's impossible..
The motivation is to avoid potential conflicts with rdma trees.
AFAIK this is the agreed practice and is being followed for some time...
If not, what's the suggested procedure then?
How do you suggest getting these IFC changes to both net and rdma trees?
^ permalink raw reply [flat|nested] 34+ messages in thread
* Re: [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-15 6:25 ` Tariq Toukan
@ 2024-12-15 21:12 ` Jakub Kicinski
2024-12-16 12:50 ` Tariq Toukan
0 siblings, 1 reply; 34+ messages in thread
From: Jakub Kicinski @ 2024-12-15 21:12 UTC (permalink / raw)
To: Tariq Toukan
Cc: Simon Horman, Tariq Toukan, David S. Miller, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Itamar Gozlan,
Yevgeny Kliteynik
On Sun, 15 Dec 2024 08:25:44 +0200 Tariq Toukan wrote:
> > What do you expect we'll do with this series?
> >
> > If you expect it to be set to Awaiting Upstream - could you make sure
> > that the cover letter has "mlx5-next" in the subject? That will makes
> > it easier to automate in patchwork.
>
> The relevant patches have mlx5-next in their topic.
> Should the cover letter as well?
> What about other non-IFC patches, keep them with net-next?
>
> > If you expect the series to be applied / merged - LMK, I can try
> > to explain why that's impossible..
>
> The motivation is to avoid potential conflicts with rdma trees.
> AFAIK this is the agreed practice and is being followed for some time...
>
> If not, what's the suggested procedure then?
> How do you suggest getting these IFC changes to both net and rdma trees?
You can post just the mlx5-next patches (preferably) or the combined
set (with mlx5-next in the cover letter tag). Wait a day or two (normal
review period, like netdev maintainers would when applying to
net-next). Apply the mlx5-next patches to mlx5-next. Send us a pull
request with just the mlx5-next stuff.
Post the net-next patches which depend on mlx5-next interface changes.
We can count this as the posting, so feel free to apply patch 1 to
mlx5-next and send the PR.
^ permalink raw reply [flat|nested] 34+ messages in thread
* Re: [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-15 21:12 ` Jakub Kicinski
@ 2024-12-16 12:50 ` Tariq Toukan
0 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-16 12:50 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Simon Horman, Tariq Toukan, David S. Miller, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Itamar Gozlan,
Yevgeny Kliteynik
On 15/12/2024 23:12, Jakub Kicinski wrote:
> On Sun, 15 Dec 2024 08:25:44 +0200 Tariq Toukan wrote:
>>> What do you expect we'll do with this series?
>>>
>>> If you expect it to be set to Awaiting Upstream - could you make sure
>>> that the cover letter has "mlx5-next" in the subject? That will makes
>>> it easier to automate in patchwork.
>>
>> The relevant patches have mlx5-next in their topic.
>> Should the cover letter as well?
>> What about other non-IFC patches, keep them with net-next?
>>
>>> If you expect the series to be applied / merged - LMK, I can try
>>> to explain why that's impossible..
>>
>> The motivation is to avoid potential conflicts with rdma trees.
>> AFAIK this is the agreed practice and is being followed for some time...
>>
>> If not, what's the suggested procedure then?
>> How do you suggest getting these IFC changes to both net and rdma trees?
>
> You can post just the mlx5-next patches (preferably) or the combined
> set (with mlx5-next in the cover letter tag). Wait a day or two (normal
> review period, like netdev maintainers would when applying to
> net-next). Apply the mlx5-next patches to mlx5-next. Send us a pull
> request with just the mlx5-next stuff.
>
Done.
> Post the net-next patches which depend on mlx5-next interface changes.
>
> We can count this as the posting, so feel free to apply patch 1 to
> mlx5-next and send the PR.
Done.
Let me know of any issue.
Thanks for your help.
^ permalink raw reply [flat|nested] 34+ messages in thread
* Re: [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering
2024-12-12 18:31 ` Tariq Toukan
2024-12-13 1:11 ` Jakub Kicinski
@ 2024-12-13 10:31 ` Simon Horman
1 sibling, 0 replies; 34+ messages in thread
From: Simon Horman @ 2024-12-13 10:31 UTC (permalink / raw)
To: Tariq Toukan
Cc: Tariq Toukan, David S. Miller, Jakub Kicinski, Paolo Abeni,
Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
Saeed Mahameed, Gal Pressman, linux-rdma, Itamar Gozlan,
Yevgeny Kliteynik
On Thu, Dec 12, 2024 at 08:31:30PM +0200, Tariq Toukan wrote:
>
>
> On 12/12/2024 19:31, Simon Horman wrote:
> > On Wed, Dec 11, 2024 at 03:42:21PM +0200, Tariq Toukan wrote:
> > > From: Itamar Gozlan <igozlan@nvidia.com>
> > >
> > > Add support for a new steering format version that is implemented by
> > > ConnectX-8.
> > > Except for several differences, the STEv3 is identical to STEv2, so
> > > for most callbacks STEv3 context struct will call STEv2 functions.
> > >
> > > Signed-off-by: Itamar Gozlan <igozlan@nvidia.com>
> > > Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> > > Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> > > ---
> > > .../net/ethernet/mellanox/mlx5/core/Makefile | 1 +
> > > .../mlx5/core/steering/sws/dr_domain.c | 2 +-
> > > .../mellanox/mlx5/core/steering/sws/dr_ste.c | 2 +
> > > .../mellanox/mlx5/core/steering/sws/dr_ste.h | 1 +
> > > .../mlx5/core/steering/sws/dr_ste_v3.c | 221 ++++++++++++++++++
> > > .../mlx5/core/steering/sws/mlx5_ifc_dr.h | 40 ++++
> > > .../mellanox/mlx5/core/steering/sws/mlx5dr.h | 2 +-
> > > 7 files changed, 267 insertions(+), 2 deletions(-)
> > > create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_ste_v3.c
> > >
> > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
> > > index 79fe09de0a9f..10a763e668ed 100644
> > > --- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
> > > +++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
> > > @@ -123,6 +123,7 @@ mlx5_core-$(CONFIG_MLX5_SW_STEERING) += steering/sws/dr_domain.o \
> > > steering/sws/dr_ste_v0.o \
> > > steering/sws/dr_ste_v1.o \
> > > steering/sws/dr_ste_v2.o \
> > > + steering/sws/dr_ste_v3.o \
> > > steering/sws/dr_cmd.o \
> > > steering/sws/dr_fw.o \
> > > steering/sws/dr_action.o \
> > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> > > index 3d74109f8230..bd361ba6658c 100644
> > > --- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> > > +++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_domain.c
> > > @@ -8,7 +8,7 @@
> > > #define DR_DOMAIN_SW_STEERING_SUPPORTED(dmn, dmn_type) \
> > > ((dmn)->info.caps.dmn_type##_sw_owner || \
> > > ((dmn)->info.caps.dmn_type##_sw_owner_v2 && \
> > > - (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_7))
> > > + (dmn)->info.caps.sw_format_ver <= MLX5_STEERING_FORMAT_CONNECTX_8))
> >
> > A definition for MLX5_STEERING_FORMAT_CONNECTX_8 seems to be missing
> > from this patch.
> >
>
> Should be pulled from mlx5-next, as described in the cover letter.
>
> Copying here for your convenience:
>
> It requires pulling 4 IFC patches that were applied to
> mlx5-next:
> https://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux.git/log/?h=mlx5-next
Thanks, sorry for missing that.
^ permalink raw reply [flat|nested] 34+ messages in thread
* [PATCH net-next 11/12] net/mlx5: Remove PTM support log message
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (9 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 10/12] net/mlx5: DR, add support for ConnectX-8 steering Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
2024-12-11 13:42 ` [PATCH net-next 12/12] net/mlx5: fs, Add support for RDMA RX steering over IB link layer Tariq Toukan
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma, Carolina Jubran,
Dragos Tatulea, Tariq Toukan
From: Carolina Jubran <cjubran@nvidia.com>
The absence of Precision Time Measurement support should not emit a
message, as it can be misleading in contexts where PTM is not required.
Remove the log message indicating the lack of PCIe PTM support.
Signed-off-by: Carolina Jubran <cjubran@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/main.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index 869bfecdd8ff..a108d8c726f8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -945,9 +945,7 @@ static int mlx5_pci_init(struct mlx5_core_dev *dev, struct pci_dev *pdev,
mlx5_pci_vsc_init(dev);
- err = pci_enable_ptm(pdev, NULL);
- if (err)
- mlx5_core_info(dev, "PTM is not supported by PCIe\n");
+ pci_enable_ptm(pdev, NULL);
return 0;
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread* [PATCH net-next 12/12] net/mlx5: fs, Add support for RDMA RX steering over IB link layer
2024-12-11 13:42 [PATCH net-next 00/12] mlx5 misc changes 2024-12-11 Tariq Toukan
` (10 preceding siblings ...)
2024-12-11 13:42 ` [PATCH net-next 11/12] net/mlx5: Remove PTM support log message Tariq Toukan
@ 2024-12-11 13:42 ` Tariq Toukan
11 siblings, 0 replies; 34+ messages in thread
From: Tariq Toukan @ 2024-12-11 13:42 UTC (permalink / raw)
To: David S. Miller, Jakub Kicinski, Paolo Abeni, Eric Dumazet,
Andrew Lunn, Leon Romanovsky
Cc: netdev, Saeed Mahameed, Gal Pressman, linux-rdma,
Patrisious Haddad, Mark Bloch, Tariq Toukan
From: Patrisious Haddad <phaddad@nvidia.com>
Relax the capability check for creating the RDMA RX steering domain
by considering only the capabilities reported by the firmware
as necessary for its creation, which in turn allows RDMA RX creation
over devices with IB link layer as well.
The table_miss_action_domain capability is required only for a specific
priority, which is handled in mlx5_rdma_enable_roce_steering().
The additional capability check for this case is already in place.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c | 3 ++-
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 3 +--
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
index 6bf0aade69d7..ae20c061e0fb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
@@ -217,7 +217,8 @@ static int mlx5_cmd_update_root_ft(struct mlx5_flow_root_namespace *ns,
int err;
if ((MLX5_CAP_GEN(dev, port_type) == MLX5_CAP_PORT_TYPE_IB) &&
- underlay_qpn == 0)
+ underlay_qpn == 0 &&
+ (ft->type != FS_FT_RDMA_RX && ft->type != FS_FT_RDMA_TX))
return 0;
if (ft->type == FS_FT_FDB &&
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index ae1a5705b26d..41b5e98a0495 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -3665,8 +3665,7 @@ int mlx5_fs_core_init(struct mlx5_core_dev *dev)
goto err;
}
- if (MLX5_CAP_FLOWTABLE_RDMA_RX(dev, ft_support) &&
- MLX5_CAP_FLOWTABLE_RDMA_RX(dev, table_miss_action_domain)) {
+ if (MLX5_CAP_FLOWTABLE_RDMA_RX(dev, ft_support)) {
err = init_rdma_rx_root_ns(steering);
if (err)
goto err;
--
2.44.0
^ permalink raw reply related [flat|nested] 34+ messages in thread