* [PATCH mlx5-next v1 01/14] macsec: add functions to get macsec real netdevice and check offload
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 02/14] net/mlx5e: Move MACsec flow steering operations to be used as core library Leon Romanovsky
` (13 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Given a macsec net_device add two functions to return the real net_device
for that device, and check if that macsec device is offloaded or not.
This is needed for auxiliary drivers that implement MACsec offload, but
have flows which are triggered over the macsec net_device, this allows
the drivers in such cases to verify if the device is offloaded or not,
and to access the real device of that macsec device, which would
belong to the driver, and would be needed for the offload procedure.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Raed Salem <raeds@nvidia.com>
Reviewed-by: Mark Zhang <markzhang@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
drivers/net/macsec.c | 15 +++++++++++++++
include/net/macsec.h | 2 ++
2 files changed, 17 insertions(+)
diff --git a/drivers/net/macsec.c b/drivers/net/macsec.c
index 984dfa5d6c11..ffc421d2de16 100644
--- a/drivers/net/macsec.c
+++ b/drivers/net/macsec.c
@@ -4240,6 +4240,21 @@ static struct net *macsec_get_link_net(const struct net_device *dev)
return dev_net(macsec_priv(dev)->real_dev);
}
+struct net_device *macsec_get_real_dev(const struct net_device *dev)
+{
+ return macsec_priv(dev)->real_dev;
+}
+EXPORT_SYMBOL_GPL(macsec_get_real_dev);
+
+bool macsec_netdev_is_offloaded(struct net_device *dev)
+{
+ if (!dev)
+ return false;
+
+ return macsec_is_offloaded(macsec_priv(dev));
+}
+EXPORT_SYMBOL_GPL(macsec_netdev_is_offloaded);
+
static size_t macsec_get_size(const struct net_device *dev)
{
return nla_total_size_64bit(8) + /* IFLA_MACSEC_SCI */
diff --git a/include/net/macsec.h b/include/net/macsec.h
index 441ed8fd4b5f..75a6f4863c83 100644
--- a/include/net/macsec.h
+++ b/include/net/macsec.h
@@ -312,6 +312,8 @@ static inline bool macsec_send_sci(const struct macsec_secy *secy)
return tx_sc->send_sci ||
(secy->n_rx_sc > 1 && !tx_sc->end_station && !tx_sc->scb);
}
+struct net_device *macsec_get_real_dev(const struct net_device *dev);
+bool macsec_netdev_is_offloaded(struct net_device *dev);
static inline void *macsec_netdev_priv(const struct net_device *dev)
{
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 02/14] net/mlx5e: Move MACsec flow steering operations to be used as core library
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 01/14] macsec: add functions to get macsec real netdevice and check offload Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 03/14] net/mlx5: Remove dependency of macsec flow steering on ethernet Leon Romanovsky
` (12 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Move MACsec flow steering operations(macsec_fs) from core/en_accel to
core/lib, this mandates moving MACsec statistics structure from the
general MACsec code header(en_accel/macsec.h) to macsec_fs header to
remove macsec_fs.h dependency over en_accel/macsec.h.
This to lay the ground for RoCE MACsec by moving all the data
that will need to be accessed by both ethernet MACsec and
RoCE MACsec to be shared at core.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/Kconfig | 2 +-
.../net/ethernet/mellanox/mlx5/core/Makefile | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en.h | 2 +-
.../mellanox/mlx5/core/en_accel/en_accel.h | 4 ++--
.../mellanox/mlx5/core/en_accel/macsec.c | 1 -
.../mellanox/mlx5/core/en_accel/macsec.h | 22 +++----------------
.../ethernet/mellanox/mlx5/core/en_stats.c | 2 +-
.../mlx5/core/{en_accel => lib}/macsec_fs.c | 2 +-
.../mlx5/core/{en_accel => lib}/macsec_fs.h | 19 ++++++++++++++--
9 files changed, 27 insertions(+), 29 deletions(-)
rename drivers/net/ethernet/mellanox/mlx5/core/{en_accel => lib}/macsec_fs.c (99%)
rename drivers/net/ethernet/mellanox/mlx5/core/{en_accel => lib}/macsec_fs.h (65%)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Kconfig b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
index bb1d7b039a7e..f3b284db1b5b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
@@ -139,7 +139,7 @@ config MLX5_CORE_IPOIB
help
MLX5 IPoIB offloads & acceleration support.
-config MLX5_EN_MACSEC
+config MLX5_MACSEC
bool "Connect-X support for MACSec offload"
depends on MLX5_CORE_EN
depends on MACSEC
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
index 63a2f2bb80a6..7d95950eb903 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile
@@ -98,7 +98,7 @@ mlx5_core-$(CONFIG_MLX5_CORE_IPOIB) += ipoib/ipoib.o ipoib/ethtool.o ipoib/ipoib
#
mlx5_core-$(CONFIG_MLX5_FPGA) += fpga/cmd.o fpga/core.o fpga/conn.o fpga/sdk.o
-mlx5_core-$(CONFIG_MLX5_EN_MACSEC) += en_accel/macsec.o en_accel/macsec_fs.o \
+mlx5_core-$(CONFIG_MLX5_MACSEC) += en_accel/macsec.o lib/macsec_fs.o \
en_accel/macsec_stats.o
mlx5_core-$(CONFIG_MLX5_EN_IPSEC) += en_accel/ipsec.o en_accel/ipsec_rxtx.o \
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 0f8f70b91485..955fb2428ba0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -917,7 +917,7 @@ struct mlx5e_priv {
const struct mlx5e_profile *profile;
void *ppriv;
-#ifdef CONFIG_MLX5_EN_MACSEC
+#ifdef CONFIG_MLX5_MACSEC
struct mlx5e_macsec *macsec;
#endif
#ifdef CONFIG_MLX5_EN_IPSEC
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
index bac4717548c6..caa34b9c161e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
@@ -138,7 +138,7 @@ static inline bool mlx5e_accel_tx_begin(struct net_device *dev,
}
#endif
-#ifdef CONFIG_MLX5_EN_MACSEC
+#ifdef CONFIG_MLX5_MACSEC
if (unlikely(mlx5e_macsec_skb_is_offload(skb))) {
struct mlx5e_priv *priv = netdev_priv(dev);
@@ -173,7 +173,7 @@ static inline void mlx5e_accel_tx_eseg(struct mlx5e_priv *priv,
mlx5e_ipsec_tx_build_eseg(priv, skb, eseg);
#endif
-#ifdef CONFIG_MLX5_EN_MACSEC
+#ifdef CONFIG_MLX5_MACSEC
if (unlikely(mlx5e_macsec_skb_is_offload(skb)))
mlx5e_macsec_tx_build_eseg(priv->macsec, skb, eseg);
#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
index 592b165530ff..b26044efdec6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
@@ -10,7 +10,6 @@
#include "lib/aso.h"
#include "lib/crypto.h"
#include "en_accel/macsec.h"
-#include "en_accel/macsec_fs.h"
#define MLX5_MACSEC_EPN_SCOPE_MID 0x80000000L
#define MLX5E_MACSEC_ASO_CTX_SZ MLX5_ST_SZ_BYTES(macsec_aso)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
index 347380a2cd9c..1f9c4a2723b2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
@@ -4,32 +4,16 @@
#ifndef __MLX5_EN_ACCEL_MACSEC_H__
#define __MLX5_EN_ACCEL_MACSEC_H__
-#ifdef CONFIG_MLX5_EN_MACSEC
+#ifdef CONFIG_MLX5_MACSEC
#include <linux/mlx5/driver.h>
#include <net/macsec.h>
#include <net/dst_metadata.h>
-
-/* Bit31 - 30: MACsec marker, Bit15-0: MACsec id */
-#define MLX5_MACEC_RX_FS_ID_MAX USHRT_MAX /* Must be power of two */
-#define MLX5_MACSEC_RX_FS_ID_MASK MLX5_MACEC_RX_FS_ID_MAX
-#define MLX5_MACSEC_METADATA_MARKER(metadata) ((((metadata) >> 30) & 0x3) == 0x1)
-#define MLX5_MACSEC_RX_METADAT_HANDLE(metadata) ((metadata) & MLX5_MACSEC_RX_FS_ID_MASK)
+#include "lib/macsec_fs.h"
struct mlx5e_priv;
struct mlx5e_macsec;
-struct mlx5e_macsec_stats {
- u64 macsec_rx_pkts;
- u64 macsec_rx_bytes;
- u64 macsec_rx_pkts_drop;
- u64 macsec_rx_bytes_drop;
- u64 macsec_tx_pkts;
- u64 macsec_tx_bytes;
- u64 macsec_tx_pkts_drop;
- u64 macsec_tx_bytes_drop;
-};
-
void mlx5e_macsec_build_netdev(struct mlx5e_priv *priv);
int mlx5e_macsec_init(struct mlx5e_priv *priv);
void mlx5e_macsec_cleanup(struct mlx5e_priv *priv);
@@ -68,6 +52,6 @@ static inline void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev,
struct mlx5_cqe64 *cqe)
{}
static inline bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev) { return false; }
-#endif /* CONFIG_MLX5_EN_MACSEC */
+#endif /* CONFIG_MLX5_MACSEC */
#endif /* __MLX5_ACCEL_EN_MACSEC_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
index 07b84d668fcc..cb47ae494856 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
@@ -2490,7 +2490,7 @@ mlx5e_stats_grp_t mlx5e_nic_stats_grps[] = {
&MLX5E_STATS_GRP(per_port_buff_congest),
&MLX5E_STATS_GRP(ptp),
&MLX5E_STATS_GRP(qos),
-#ifdef CONFIG_MLX5_EN_MACSEC
+#ifdef CONFIG_MLX5_MACSEC
&MLX5E_STATS_GRP(macsec_hw),
#endif
};
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
similarity index 99%
rename from drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c
rename to drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
index 414e28584881..46c0af66d72c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -7,7 +7,7 @@
#include <linux/if_vlan.h>
#include "fs_core.h"
#include "en/fs.h"
-#include "en_accel/macsec_fs.h"
+#include "lib/macsec_fs.h"
#include "mlx5_core.h"
/* MACsec TX flow steering */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
similarity index 65%
rename from drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.h
rename to drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
index b429648d4ee7..b282c0850e16 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_fs.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
@@ -4,9 +4,13 @@
#ifndef __MLX5_MACSEC_STEERING_H__
#define __MLX5_MACSEC_STEERING_H__
-#ifdef CONFIG_MLX5_EN_MACSEC
+#ifdef CONFIG_MLX5_MACSEC
-#include "en_accel/macsec.h"
+/* Bit31 - 30: MACsec marker, Bit15-0: MACsec id */
+#define MLX5_MACEC_RX_FS_ID_MAX USHRT_MAX /* Must be power of two */
+#define MLX5_MACSEC_RX_FS_ID_MASK MLX5_MACEC_RX_FS_ID_MAX
+#define MLX5_MACSEC_METADATA_MARKER(metadata) ((((metadata) >> 30) & 0x3) == 0x1)
+#define MLX5_MACSEC_RX_METADAT_HANDLE(metadata) ((metadata) & MLX5_MACSEC_RX_FS_ID_MASK)
#define MLX5_MACSEC_NUM_OF_SUPPORTED_INTERFACES 16
@@ -20,6 +24,17 @@ struct mlx5_macsec_rule_attrs {
int action;
};
+struct mlx5e_macsec_stats {
+ u64 macsec_rx_pkts;
+ u64 macsec_rx_bytes;
+ u64 macsec_rx_pkts_drop;
+ u64 macsec_rx_bytes_drop;
+ u64 macsec_tx_pkts;
+ u64 macsec_tx_bytes;
+ u64 macsec_tx_pkts_drop;
+ u64 macsec_tx_bytes_drop;
+};
+
enum mlx5_macsec_action {
MLX5_ACCEL_MACSEC_ACTION_ENCRYPT,
MLX5_ACCEL_MACSEC_ACTION_DECRYPT,
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 03/14] net/mlx5: Remove dependency of macsec flow steering on ethernet
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 01/14] macsec: add functions to get macsec real netdevice and check offload Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 02/14] net/mlx5e: Move MACsec flow steering operations to be used as core library Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 04/14] net/mlx5e: Rename MACsec flow steering functions/parameters to suit core naming style Leon Romanovsky
` (11 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Since macsec flow steering was moved to core, it should be independent
of all ethernet code and structures hence we remove all ethernet header
includes and redefine ethernet structs internally for macsec_fs usage
where needed.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../mellanox/mlx5/core/lib/macsec_fs.c | 43 +++++++++++++++----
1 file changed, 34 insertions(+), 9 deletions(-)
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 46c0af66d72c..ace6b67f1c97 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -6,7 +6,6 @@
#include <linux/mlx5/qp.h>
#include <linux/if_vlan.h>
#include "fs_core.h"
-#include "en/fs.h"
#include "lib/macsec_fs.h"
#include "mlx5_core.h"
@@ -57,8 +56,14 @@ struct mlx5e_macsec_tx_rule {
u32 fs_id;
};
+struct mlx5_macsec_flow_table {
+ int num_groups;
+ struct mlx5_flow_table *t;
+ struct mlx5_flow_group **g;
+};
+
struct mlx5e_macsec_tables {
- struct mlx5e_flow_table ft_crypto;
+ struct mlx5_macsec_flow_table ft_crypto;
struct mlx5_flow_handle *crypto_miss_rule;
struct mlx5_flow_table *ft_check;
@@ -103,6 +108,26 @@ struct mlx5e_macsec_fs {
struct mlx5e_macsec_rx *rx_fs;
};
+static void macsec_fs_destroy_groups(struct mlx5_macsec_flow_table *ft)
+{
+ int i;
+
+ for (i = ft->num_groups - 1; i >= 0; i--) {
+ if (!IS_ERR_OR_NULL(ft->g[i]))
+ mlx5_destroy_flow_group(ft->g[i]);
+ ft->g[i] = NULL;
+ }
+ ft->num_groups = 0;
+}
+
+static void macsec_fs_destroy_flow_table(struct mlx5_macsec_flow_table *ft)
+{
+ macsec_fs_destroy_groups(ft);
+ kfree(ft->g);
+ mlx5_destroy_flow_table(ft->t);
+ ft->t = NULL;
+}
+
static void macsec_fs_tx_destroy(struct mlx5e_macsec_fs *macsec_fs)
{
struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
@@ -142,10 +167,10 @@ static void macsec_fs_tx_destroy(struct mlx5e_macsec_fs *macsec_fs)
tx_tables->crypto_miss_rule = NULL;
}
- mlx5e_destroy_flow_table(&tx_tables->ft_crypto);
+ macsec_fs_destroy_flow_table(&tx_tables->ft_crypto);
}
-static int macsec_fs_tx_create_crypto_table_groups(struct mlx5e_flow_table *ft)
+static int macsec_fs_tx_create_crypto_table_groups(struct mlx5_macsec_flow_table *ft)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
int mclen = MLX5_ST_SZ_BYTES(fte_match_param);
@@ -245,7 +270,7 @@ static int macsec_fs_tx_create(struct mlx5e_macsec_fs *macsec_fs)
struct mlx5_flow_destination dest = {};
struct mlx5e_macsec_tables *tx_tables;
struct mlx5_flow_act flow_act = {};
- struct mlx5e_flow_table *ft_crypto;
+ struct mlx5_macsec_flow_table *ft_crypto;
struct mlx5_flow_table *flow_table;
struct mlx5_flow_group *flow_group;
struct mlx5_flow_namespace *ns;
@@ -734,10 +759,10 @@ static void macsec_fs_rx_destroy(struct mlx5e_macsec_fs *macsec_fs)
rx_tables->crypto_miss_rule = NULL;
}
- mlx5e_destroy_flow_table(&rx_tables->ft_crypto);
+ macsec_fs_destroy_flow_table(&rx_tables->ft_crypto);
}
-static int macsec_fs_rx_create_crypto_table_groups(struct mlx5e_flow_table *ft)
+static int macsec_fs_rx_create_crypto_table_groups(struct mlx5_macsec_flow_table *ft)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
int mclen = MLX5_ST_SZ_BYTES(fte_match_param);
@@ -895,10 +920,10 @@ static int macsec_fs_rx_create(struct mlx5e_macsec_fs *macsec_fs)
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
struct net_device *netdev = macsec_fs->netdev;
+ struct mlx5_macsec_flow_table *ft_crypto;
struct mlx5_flow_table_attr ft_attr = {};
struct mlx5_flow_destination dest = {};
struct mlx5e_macsec_tables *rx_tables;
- struct mlx5e_flow_table *ft_crypto;
struct mlx5_flow_table *flow_table;
struct mlx5_flow_group *flow_group;
struct mlx5_flow_act flow_act = {};
@@ -1123,11 +1148,11 @@ macsec_fs_rx_add_rule(struct mlx5e_macsec_fs *macsec_fs,
struct net_device *netdev = macsec_fs->netdev;
union mlx5e_macsec_rule *macsec_rule = NULL;
struct mlx5_modify_hdr *modify_hdr = NULL;
+ struct mlx5_macsec_flow_table *ft_crypto;
struct mlx5_flow_destination dest = {};
struct mlx5e_macsec_tables *rx_tables;
struct mlx5e_macsec_rx_rule *rx_rule;
struct mlx5_flow_act flow_act = {};
- struct mlx5e_flow_table *ft_crypto;
struct mlx5_flow_handle *rule;
struct mlx5_flow_spec *spec;
int err = 0;
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 04/14] net/mlx5e: Rename MACsec flow steering functions/parameters to suit core naming style
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (2 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 03/14] net/mlx5: Remove dependency of macsec flow steering on ethernet Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 05/14] net/mlx5e: Move MACsec flow steering and statistics database from ethernet to core Leon Romanovsky
` (10 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Rename MACsec flow steering(macsec_fs) functions and parameters from
ethernet(core/en_accel) naming convention to core naming convention.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../mellanox/mlx5/core/en_accel/macsec.c | 22 +--
.../mellanox/mlx5/core/en_accel/macsec.h | 2 +-
.../mlx5/core/en_accel/macsec_stats.c | 16 +-
.../mellanox/mlx5/core/lib/macsec_fs.c | 168 +++++++++---------
.../mellanox/mlx5/core/lib/macsec_fs.h | 30 ++--
5 files changed, 119 insertions(+), 119 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
index b26044efdec6..831d83094abd 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
@@ -67,7 +67,7 @@ struct mlx5e_macsec_sa {
struct rhash_head hash;
u32 fs_id;
- union mlx5e_macsec_rule *macsec_rule;
+ union mlx5_macsec_rule *macsec_rule;
struct rcu_head rcu_head;
struct mlx5e_macsec_epn_state epn_state;
};
@@ -124,7 +124,7 @@ struct mlx5e_macsec_device {
struct mlx5e_macsec {
struct list_head macsec_device_list_head;
int num_of_devices;
- struct mlx5e_macsec_fs *macsec_fs;
+ struct mlx5_macsec_fs *macsec_fs;
struct mutex lock; /* Protects mlx5e_macsec internal contexts */
/* Tx sci -> fs id mapping handling */
@@ -136,7 +136,7 @@ struct mlx5e_macsec {
struct mlx5_core_dev *mdev;
/* Stats manage */
- struct mlx5e_macsec_stats stats;
+ struct mlx5_macsec_stats stats;
/* ASO */
struct mlx5e_macsec_aso aso;
@@ -343,7 +343,7 @@ static void mlx5e_macsec_cleanup_sa(struct mlx5e_macsec *macsec,
if (!sa->macsec_rule)
return;
- mlx5e_macsec_fs_del_rule(macsec->macsec_fs, sa->macsec_rule, action);
+ mlx5_macsec_fs_del_rule(macsec->macsec_fs, sa->macsec_rule, action);
mlx5e_macsec_destroy_object(macsec->mdev, sa->macsec_obj_id);
sa->macsec_rule = NULL;
}
@@ -358,7 +358,7 @@ static int mlx5e_macsec_init_sa(struct macsec_context *ctx,
struct mlx5_macsec_rule_attrs rule_attrs;
struct mlx5_core_dev *mdev = priv->mdev;
struct mlx5_macsec_obj_attrs obj_attrs;
- union mlx5e_macsec_rule *macsec_rule;
+ union mlx5_macsec_rule *macsec_rule;
int err;
obj_attrs.next_pn = sa->next_pn;
@@ -386,7 +386,7 @@ static int mlx5e_macsec_init_sa(struct macsec_context *ctx,
rule_attrs.action = (is_tx) ? MLX5_ACCEL_MACSEC_ACTION_ENCRYPT :
MLX5_ACCEL_MACSEC_ACTION_DECRYPT;
- macsec_rule = mlx5e_macsec_fs_add_rule(macsec->macsec_fs, ctx, &rule_attrs, &sa->fs_id);
+ macsec_rule = mlx5_macsec_fs_add_rule(macsec->macsec_fs, ctx, &rule_attrs, &sa->fs_id);
if (!macsec_rule) {
err = -ENOMEM;
goto destroy_macsec_object;
@@ -1679,10 +1679,10 @@ bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev)
void mlx5e_macsec_get_stats_fill(struct mlx5e_macsec *macsec, void *macsec_stats)
{
- mlx5e_macsec_fs_get_stats_fill(macsec->macsec_fs, macsec_stats);
+ mlx5_macsec_fs_get_stats_fill(macsec->macsec_fs, macsec_stats);
}
-struct mlx5e_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec)
+struct mlx5_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec)
{
if (!macsec)
return NULL;
@@ -1781,7 +1781,7 @@ int mlx5e_macsec_init(struct mlx5e_priv *priv)
{
struct mlx5_core_dev *mdev = priv->mdev;
struct mlx5e_macsec *macsec = NULL;
- struct mlx5e_macsec_fs *macsec_fs;
+ struct mlx5_macsec_fs *macsec_fs;
int err;
if (!mlx5e_is_macsec_device(priv->mdev)) {
@@ -1821,7 +1821,7 @@ int mlx5e_macsec_init(struct mlx5e_priv *priv)
macsec->mdev = mdev;
- macsec_fs = mlx5e_macsec_fs_init(mdev, priv->netdev);
+ macsec_fs = mlx5_macsec_fs_init(mdev, priv->netdev);
if (!macsec_fs) {
err = -ENOMEM;
goto err_out;
@@ -1857,7 +1857,7 @@ void mlx5e_macsec_cleanup(struct mlx5e_priv *priv)
return;
mlx5_notifier_unregister(mdev, &macsec->nb);
- mlx5e_macsec_fs_cleanup(macsec->macsec_fs);
+ mlx5_macsec_fs_cleanup(macsec->macsec_fs);
destroy_workqueue(macsec->wq);
mlx5e_macsec_aso_cleanup(&macsec->aso, mdev);
rhashtable_destroy(&macsec->sci_hash);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
index 1f9c4a2723b2..47dc1d4448d9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
@@ -38,7 +38,7 @@ void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev, struct sk_buf
struct mlx5_cqe64 *cqe);
bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev);
void mlx5e_macsec_get_stats_fill(struct mlx5e_macsec *macsec, void *macsec_stats);
-struct mlx5e_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec);
+struct mlx5_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec);
#else
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
index e50a2e3f3d18..8326a593b3fb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
@@ -8,14 +8,14 @@
#include "en_accel/macsec.h"
static const struct counter_desc mlx5e_macsec_hw_stats_desc[] = {
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_pkts) },
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_bytes) },
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_pkts_drop) },
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_rx_bytes_drop) },
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_pkts) },
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_bytes) },
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_pkts_drop) },
- { MLX5E_DECLARE_STAT(struct mlx5e_macsec_stats, macsec_tx_bytes_drop) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_rx_pkts) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_rx_bytes) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_rx_pkts_drop) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_rx_bytes_drop) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_tx_pkts) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_tx_bytes) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_tx_pkts_drop) },
+ { MLX5E_DECLARE_STAT(struct mlx5_macsec_stats, macsec_tx_bytes_drop) },
};
#define NUM_MACSEC_HW_COUNTERS ARRAY_SIZE(mlx5e_macsec_hw_stats_desc)
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 ace6b67f1c97..8feb7db6a220 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -50,7 +50,7 @@ struct mlx5_sectag_header {
u8 sci[MACSEC_SCI_LEN]; /* optional */
} __packed;
-struct mlx5e_macsec_tx_rule {
+struct mlx5_macsec_tx_rule {
struct mlx5_flow_handle *rule;
struct mlx5_pkt_reformat *pkt_reformat;
u32 fs_id;
@@ -62,7 +62,7 @@ struct mlx5_macsec_flow_table {
struct mlx5_flow_group **g;
};
-struct mlx5e_macsec_tables {
+struct mlx5_macsec_tables {
struct mlx5_macsec_flow_table ft_crypto;
struct mlx5_flow_handle *crypto_miss_rule;
@@ -75,37 +75,37 @@ struct mlx5e_macsec_tables {
u32 refcnt;
};
-struct mlx5e_macsec_tx {
+struct mlx5_macsec_tx {
struct mlx5_flow_handle *crypto_mke_rule;
struct mlx5_flow_handle *check_rule;
struct ida tx_halloc;
- struct mlx5e_macsec_tables tables;
+ struct mlx5_macsec_tables tables;
};
-struct mlx5e_macsec_rx_rule {
+struct mlx5_macsec_rx_rule {
struct mlx5_flow_handle *rule[RX_NUM_OF_RULES_PER_SA];
struct mlx5_modify_hdr *meta_modhdr;
};
-struct mlx5e_macsec_rx {
+struct mlx5_macsec_rx {
struct mlx5_flow_handle *check_rule[2];
struct mlx5_pkt_reformat *check_rule_pkt_reformat[2];
- struct mlx5e_macsec_tables tables;
+ struct mlx5_macsec_tables tables;
};
-union mlx5e_macsec_rule {
- struct mlx5e_macsec_tx_rule tx_rule;
- struct mlx5e_macsec_rx_rule rx_rule;
+union mlx5_macsec_rule {
+ struct mlx5_macsec_tx_rule tx_rule;
+ struct mlx5_macsec_rx_rule rx_rule;
};
-struct mlx5e_macsec_fs {
+struct mlx5_macsec_fs {
struct mlx5_core_dev *mdev;
struct net_device *netdev;
- struct mlx5e_macsec_tx *tx_fs;
- struct mlx5e_macsec_rx *rx_fs;
+ struct mlx5_macsec_tx *tx_fs;
+ struct mlx5_macsec_rx *rx_fs;
};
static void macsec_fs_destroy_groups(struct mlx5_macsec_flow_table *ft)
@@ -128,10 +128,10 @@ static void macsec_fs_destroy_flow_table(struct mlx5_macsec_flow_table *ft)
ft->t = NULL;
}
-static void macsec_fs_tx_destroy(struct mlx5e_macsec_fs *macsec_fs)
+static void macsec_fs_tx_destroy(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
- struct mlx5e_macsec_tables *tx_tables;
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_macsec_tables *tx_tables;
tx_tables = &tx_fs->tables;
@@ -261,14 +261,14 @@ static struct mlx5_flow_table
return fdb;
}
-static int macsec_fs_tx_create(struct mlx5e_macsec_fs *macsec_fs)
+static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
- struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
struct net_device *netdev = macsec_fs->netdev;
struct mlx5_flow_table_attr ft_attr = {};
struct mlx5_flow_destination dest = {};
- struct mlx5e_macsec_tables *tx_tables;
+ struct mlx5_macsec_tables *tx_tables;
struct mlx5_flow_act flow_act = {};
struct mlx5_macsec_flow_table *ft_crypto;
struct mlx5_flow_table *flow_table;
@@ -414,10 +414,10 @@ static int macsec_fs_tx_create(struct mlx5e_macsec_fs *macsec_fs)
return err;
}
-static int macsec_fs_tx_ft_get(struct mlx5e_macsec_fs *macsec_fs)
+static int macsec_fs_tx_ft_get(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
- struct mlx5e_macsec_tables *tx_tables;
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_macsec_tables *tx_tables;
int err = 0;
tx_tables = &tx_fs->tables;
@@ -433,9 +433,9 @@ static int macsec_fs_tx_ft_get(struct mlx5e_macsec_fs *macsec_fs)
return err;
}
-static void macsec_fs_tx_ft_put(struct mlx5e_macsec_fs *macsec_fs)
+static void macsec_fs_tx_ft_put(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_tables *tx_tables = &macsec_fs->tx_fs->tables;
+ struct mlx5_macsec_tables *tx_tables = &macsec_fs->tx_fs->tables;
if (--tx_tables->refcnt)
return;
@@ -443,13 +443,13 @@ static void macsec_fs_tx_ft_put(struct mlx5e_macsec_fs *macsec_fs)
macsec_fs_tx_destroy(macsec_fs);
}
-static int macsec_fs_tx_setup_fte(struct mlx5e_macsec_fs *macsec_fs,
+static int macsec_fs_tx_setup_fte(struct mlx5_macsec_fs *macsec_fs,
struct mlx5_flow_spec *spec,
struct mlx5_flow_act *flow_act,
u32 macsec_obj_id,
u32 *fs_id)
{
- struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
int err = 0;
u32 id;
@@ -512,8 +512,8 @@ static void macsec_fs_tx_create_sectag_header(const struct macsec_context *ctx,
memcpy(reformatbf, §ag, *reformat_size);
}
-static void macsec_fs_tx_del_rule(struct mlx5e_macsec_fs *macsec_fs,
- struct mlx5e_macsec_tx_rule *tx_rule)
+static void macsec_fs_tx_del_rule(struct mlx5_macsec_fs *macsec_fs,
+ struct mlx5_macsec_tx_rule *tx_rule)
{
if (tx_rule->rule) {
mlx5_del_flow_rules(tx_rule->rule);
@@ -537,20 +537,20 @@ static void macsec_fs_tx_del_rule(struct mlx5e_macsec_fs *macsec_fs,
#define MLX5_REFORMAT_PARAM_ADD_MACSEC_OFFSET_4_BYTES 1
-static union mlx5e_macsec_rule *
-macsec_fs_tx_add_rule(struct mlx5e_macsec_fs *macsec_fs,
+static union mlx5_macsec_rule *
+macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
const struct macsec_context *macsec_ctx,
struct mlx5_macsec_rule_attrs *attrs,
u32 *sa_fs_id)
{
char reformatbf[MLX5_MACSEC_TAG_LEN + MACSEC_SCI_LEN];
struct mlx5_pkt_reformat_params reformat_params = {};
- struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
struct net_device *netdev = macsec_fs->netdev;
- union mlx5e_macsec_rule *macsec_rule = NULL;
+ union mlx5_macsec_rule *macsec_rule = NULL;
struct mlx5_flow_destination dest = {};
- struct mlx5e_macsec_tables *tx_tables;
- struct mlx5e_macsec_tx_rule *tx_rule;
+ struct mlx5_macsec_tables *tx_tables;
+ struct mlx5_macsec_tx_rule *tx_rule;
struct mlx5_flow_act flow_act = {};
struct mlx5_flow_handle *rule;
struct mlx5_flow_spec *spec;
@@ -631,11 +631,11 @@ macsec_fs_tx_add_rule(struct mlx5e_macsec_fs *macsec_fs,
return macsec_rule;
}
-static void macsec_fs_tx_cleanup(struct mlx5e_macsec_fs *macsec_fs)
+static void macsec_fs_tx_cleanup(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
- struct mlx5e_macsec_tables *tx_tables;
+ struct mlx5_macsec_tables *tx_tables;
if (!tx_fs)
return;
@@ -664,12 +664,12 @@ static void macsec_fs_tx_cleanup(struct mlx5e_macsec_fs *macsec_fs)
macsec_fs->tx_fs = NULL;
}
-static int macsec_fs_tx_init(struct mlx5e_macsec_fs *macsec_fs)
+static int macsec_fs_tx_init(struct mlx5_macsec_fs *macsec_fs)
{
struct net_device *netdev = macsec_fs->netdev;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
- struct mlx5e_macsec_tables *tx_tables;
- struct mlx5e_macsec_tx *tx_fs;
+ struct mlx5_macsec_tables *tx_tables;
+ struct mlx5_macsec_tx *tx_fs;
struct mlx5_fc *flow_counter;
int err;
@@ -716,10 +716,10 @@ static int macsec_fs_tx_init(struct mlx5e_macsec_fs *macsec_fs)
return err;
}
-static void macsec_fs_rx_destroy(struct mlx5e_macsec_fs *macsec_fs)
+static void macsec_fs_rx_destroy(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
- struct mlx5e_macsec_tables *rx_tables;
+ struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
+ struct mlx5_macsec_tables *rx_tables;
int i;
/* Rx check table */
@@ -844,7 +844,7 @@ static int macsec_fs_rx_create_crypto_table_groups(struct mlx5_macsec_flow_table
return err;
}
-static int macsec_fs_rx_create_check_decap_rule(struct mlx5e_macsec_fs *macsec_fs,
+static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs,
struct mlx5_flow_destination *dest,
struct mlx5_flow_act *flow_act,
struct mlx5_flow_spec *spec,
@@ -853,9 +853,9 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5e_macsec_fs *macsec_f
int rule_index = (reformat_param_size == MLX5_SECTAG_HEADER_SIZE_WITH_SCI) ? 0 : 1;
u8 mlx5_reformat_buf[MLX5_SECTAG_HEADER_SIZE_WITH_SCI];
struct mlx5_pkt_reformat_params reformat_params = {};
- struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+ struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
struct net_device *netdev = macsec_fs->netdev;
- struct mlx5e_macsec_tables *rx_tables;
+ struct mlx5_macsec_tables *rx_tables;
struct mlx5_flow_handle *rule;
int err = 0;
@@ -915,15 +915,15 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5e_macsec_fs *macsec_f
return 0;
}
-static int macsec_fs_rx_create(struct mlx5e_macsec_fs *macsec_fs)
+static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
- struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+ struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
struct net_device *netdev = macsec_fs->netdev;
struct mlx5_macsec_flow_table *ft_crypto;
struct mlx5_flow_table_attr ft_attr = {};
struct mlx5_flow_destination dest = {};
- struct mlx5e_macsec_tables *rx_tables;
+ struct mlx5_macsec_tables *rx_tables;
struct mlx5_flow_table *flow_table;
struct mlx5_flow_group *flow_group;
struct mlx5_flow_act flow_act = {};
@@ -1043,9 +1043,9 @@ static int macsec_fs_rx_create(struct mlx5e_macsec_fs *macsec_fs)
return err;
}
-static int macsec_fs_rx_ft_get(struct mlx5e_macsec_fs *macsec_fs)
+static int macsec_fs_rx_ft_get(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
+ struct mlx5_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
int err = 0;
if (rx_tables->refcnt)
@@ -1060,9 +1060,9 @@ static int macsec_fs_rx_ft_get(struct mlx5e_macsec_fs *macsec_fs)
return err;
}
-static void macsec_fs_rx_ft_put(struct mlx5e_macsec_fs *macsec_fs)
+static void macsec_fs_rx_ft_put(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
+ struct mlx5_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
if (--rx_tables->refcnt)
return;
@@ -1070,8 +1070,8 @@ static void macsec_fs_rx_ft_put(struct mlx5e_macsec_fs *macsec_fs)
macsec_fs_rx_destroy(macsec_fs);
}
-static void macsec_fs_rx_del_rule(struct mlx5e_macsec_fs *macsec_fs,
- struct mlx5e_macsec_rx_rule *rx_rule)
+static void macsec_fs_rx_del_rule(struct mlx5_macsec_fs *macsec_fs,
+ struct mlx5_macsec_rx_rule *rx_rule)
{
int i;
@@ -1138,20 +1138,20 @@ static void macsec_fs_rx_setup_fte(struct mlx5_flow_spec *spec,
crypto_params->obj_id = attrs->macsec_obj_id;
}
-static union mlx5e_macsec_rule *
-macsec_fs_rx_add_rule(struct mlx5e_macsec_fs *macsec_fs,
+static union mlx5_macsec_rule *
+macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
struct mlx5_macsec_rule_attrs *attrs,
u32 fs_id)
{
u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
- struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+ struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
struct net_device *netdev = macsec_fs->netdev;
- union mlx5e_macsec_rule *macsec_rule = NULL;
+ union mlx5_macsec_rule *macsec_rule = NULL;
struct mlx5_modify_hdr *modify_hdr = NULL;
struct mlx5_macsec_flow_table *ft_crypto;
struct mlx5_flow_destination dest = {};
- struct mlx5e_macsec_tables *rx_tables;
- struct mlx5e_macsec_rx_rule *rx_rule;
+ struct mlx5_macsec_tables *rx_tables;
+ struct mlx5_macsec_rx_rule *rx_rule;
struct mlx5_flow_act flow_act = {};
struct mlx5_flow_handle *rule;
struct mlx5_flow_spec *spec;
@@ -1250,12 +1250,12 @@ macsec_fs_rx_add_rule(struct mlx5e_macsec_fs *macsec_fs,
return macsec_rule;
}
-static int macsec_fs_rx_init(struct mlx5e_macsec_fs *macsec_fs)
+static int macsec_fs_rx_init(struct mlx5_macsec_fs *macsec_fs)
{
struct net_device *netdev = macsec_fs->netdev;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
- struct mlx5e_macsec_tables *rx_tables;
- struct mlx5e_macsec_rx *rx_fs;
+ struct mlx5_macsec_tables *rx_tables;
+ struct mlx5_macsec_rx *rx_fs;
struct mlx5_fc *flow_counter;
int err;
@@ -1300,11 +1300,11 @@ static int macsec_fs_rx_init(struct mlx5e_macsec_fs *macsec_fs)
return err;
}
-static void macsec_fs_rx_cleanup(struct mlx5e_macsec_fs *macsec_fs)
+static void macsec_fs_rx_cleanup(struct mlx5_macsec_fs *macsec_fs)
{
- struct mlx5e_macsec_rx *rx_fs = macsec_fs->rx_fs;
+ struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
- struct mlx5e_macsec_tables *rx_tables;
+ struct mlx5_macsec_tables *rx_tables;
if (!rx_fs)
return;
@@ -1332,11 +1332,11 @@ static void macsec_fs_rx_cleanup(struct mlx5e_macsec_fs *macsec_fs)
macsec_fs->rx_fs = NULL;
}
-void mlx5e_macsec_fs_get_stats_fill(struct mlx5e_macsec_fs *macsec_fs, void *macsec_stats)
+void mlx5_macsec_fs_get_stats_fill(struct mlx5_macsec_fs *macsec_fs, void *macsec_stats)
{
- struct mlx5e_macsec_stats *stats = (struct mlx5e_macsec_stats *)macsec_stats;
- struct mlx5e_macsec_tables *tx_tables = &macsec_fs->tx_fs->tables;
- struct mlx5e_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
+ struct mlx5_macsec_stats *stats = (struct mlx5_macsec_stats *)macsec_stats;
+ struct mlx5_macsec_tables *tx_tables = &macsec_fs->tx_fs->tables;
+ struct mlx5_macsec_tables *rx_tables = &macsec_fs->rx_fs->tables;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
if (tx_tables->check_rule_counter)
@@ -1356,38 +1356,38 @@ void mlx5e_macsec_fs_get_stats_fill(struct mlx5e_macsec_fs *macsec_fs, void *mac
&stats->macsec_rx_pkts_drop, &stats->macsec_rx_bytes_drop);
}
-union mlx5e_macsec_rule *
-mlx5e_macsec_fs_add_rule(struct mlx5e_macsec_fs *macsec_fs,
- const struct macsec_context *macsec_ctx,
- struct mlx5_macsec_rule_attrs *attrs,
- u32 *sa_fs_id)
+union mlx5_macsec_rule *
+mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
+ const struct macsec_context *macsec_ctx,
+ struct mlx5_macsec_rule_attrs *attrs,
+ u32 *sa_fs_id)
{
return (attrs->action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
macsec_fs_tx_add_rule(macsec_fs, macsec_ctx, attrs, sa_fs_id) :
macsec_fs_rx_add_rule(macsec_fs, attrs, *sa_fs_id);
}
-void mlx5e_macsec_fs_del_rule(struct mlx5e_macsec_fs *macsec_fs,
- union mlx5e_macsec_rule *macsec_rule,
- int action)
+void mlx5_macsec_fs_del_rule(struct mlx5_macsec_fs *macsec_fs,
+ union mlx5_macsec_rule *macsec_rule,
+ int action)
{
(action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
macsec_fs_tx_del_rule(macsec_fs, &macsec_rule->tx_rule) :
macsec_fs_rx_del_rule(macsec_fs, &macsec_rule->rx_rule);
}
-void mlx5e_macsec_fs_cleanup(struct mlx5e_macsec_fs *macsec_fs)
+void mlx5_macsec_fs_cleanup(struct mlx5_macsec_fs *macsec_fs)
{
macsec_fs_rx_cleanup(macsec_fs);
macsec_fs_tx_cleanup(macsec_fs);
kfree(macsec_fs);
}
-struct mlx5e_macsec_fs *
-mlx5e_macsec_fs_init(struct mlx5_core_dev *mdev,
- struct net_device *netdev)
+struct mlx5_macsec_fs *
+mlx5_macsec_fs_init(struct mlx5_core_dev *mdev,
+ struct net_device *netdev)
{
- struct mlx5e_macsec_fs *macsec_fs;
+ struct mlx5_macsec_fs *macsec_fs;
int err;
macsec_fs = kzalloc(sizeof(*macsec_fs), GFP_KERNEL);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
index b282c0850e16..6a749e036e68 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
@@ -14,8 +14,8 @@
#define MLX5_MACSEC_NUM_OF_SUPPORTED_INTERFACES 16
-struct mlx5e_macsec_fs;
-union mlx5e_macsec_rule;
+struct mlx5_macsec_fs;
+union mlx5_macsec_rule;
struct mlx5_macsec_rule_attrs {
sci_t sci;
@@ -24,7 +24,7 @@ struct mlx5_macsec_rule_attrs {
int action;
};
-struct mlx5e_macsec_stats {
+struct mlx5_macsec_stats {
u64 macsec_rx_pkts;
u64 macsec_rx_bytes;
u64 macsec_rx_pkts_drop;
@@ -40,22 +40,22 @@ enum mlx5_macsec_action {
MLX5_ACCEL_MACSEC_ACTION_DECRYPT,
};
-void mlx5e_macsec_fs_cleanup(struct mlx5e_macsec_fs *macsec_fs);
+void mlx5_macsec_fs_cleanup(struct mlx5_macsec_fs *macsec_fs);
-struct mlx5e_macsec_fs *
-mlx5e_macsec_fs_init(struct mlx5_core_dev *mdev, struct net_device *netdev);
+struct mlx5_macsec_fs *
+mlx5_macsec_fs_init(struct mlx5_core_dev *mdev, struct net_device *netdev);
-union mlx5e_macsec_rule *
-mlx5e_macsec_fs_add_rule(struct mlx5e_macsec_fs *macsec_fs,
- const struct macsec_context *ctx,
- struct mlx5_macsec_rule_attrs *attrs,
- u32 *sa_fs_id);
+union mlx5_macsec_rule *
+mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
+ const struct macsec_context *ctx,
+ struct mlx5_macsec_rule_attrs *attrs,
+ u32 *sa_fs_id);
-void mlx5e_macsec_fs_del_rule(struct mlx5e_macsec_fs *macsec_fs,
- union mlx5e_macsec_rule *macsec_rule,
- int action);
+void mlx5_macsec_fs_del_rule(struct mlx5_macsec_fs *macsec_fs,
+ union mlx5_macsec_rule *macsec_rule,
+ int action);
-void mlx5e_macsec_fs_get_stats_fill(struct mlx5e_macsec_fs *macsec_fs, void *macsec_stats);
+void mlx5_macsec_fs_get_stats_fill(struct mlx5_macsec_fs *macsec_fs, void *macsec_stats);
#endif
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 05/14] net/mlx5e: Move MACsec flow steering and statistics database from ethernet to core
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (3 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 04/14] net/mlx5e: Rename MACsec flow steering functions/parameters to suit core naming style Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 06/14] net/mlx5: Remove netdevice from MACsec steering Leon Romanovsky
` (9 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Since now MACsec flow steering (macsec_fs) and MACsec statistics (stats)
are maintained by the core driver, move their data as well to be saved
inside core structures instead of staying part of ethernet MACsec database.
In addition cleanup all MACsec stats functions from the ethernet MACsec
code and move what's needed to be part of macsec_fs instead.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../mellanox/mlx5/core/en_accel/macsec.c | 25 +++----------------
.../mellanox/mlx5/core/en_accel/macsec.h | 2 --
.../mlx5/core/en_accel/macsec_stats.c | 6 +++--
.../mellanox/mlx5/core/lib/macsec_fs.c | 11 ++++++++
.../mellanox/mlx5/core/lib/macsec_fs.h | 1 +
include/linux/mlx5/driver.h | 3 +++
6 files changed, 23 insertions(+), 25 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
index 831d83094abd..5e8ad355c67f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
@@ -124,7 +124,6 @@ struct mlx5e_macsec_device {
struct mlx5e_macsec {
struct list_head macsec_device_list_head;
int num_of_devices;
- struct mlx5_macsec_fs *macsec_fs;
struct mutex lock; /* Protects mlx5e_macsec internal contexts */
/* Tx sci -> fs id mapping handling */
@@ -135,9 +134,6 @@ struct mlx5e_macsec {
struct mlx5_core_dev *mdev;
- /* Stats manage */
- struct mlx5_macsec_stats stats;
-
/* ASO */
struct mlx5e_macsec_aso aso;
@@ -343,7 +339,7 @@ static void mlx5e_macsec_cleanup_sa(struct mlx5e_macsec *macsec,
if (!sa->macsec_rule)
return;
- mlx5_macsec_fs_del_rule(macsec->macsec_fs, sa->macsec_rule, action);
+ mlx5_macsec_fs_del_rule(macsec->mdev->macsec_fs, sa->macsec_rule, action);
mlx5e_macsec_destroy_object(macsec->mdev, sa->macsec_obj_id);
sa->macsec_rule = NULL;
}
@@ -386,7 +382,7 @@ static int mlx5e_macsec_init_sa(struct macsec_context *ctx,
rule_attrs.action = (is_tx) ? MLX5_ACCEL_MACSEC_ACTION_ENCRYPT :
MLX5_ACCEL_MACSEC_ACTION_DECRYPT;
- macsec_rule = mlx5_macsec_fs_add_rule(macsec->macsec_fs, ctx, &rule_attrs, &sa->fs_id);
+ macsec_rule = mlx5_macsec_fs_add_rule(mdev->macsec_fs, ctx, &rule_attrs, &sa->fs_id);
if (!macsec_rule) {
err = -ENOMEM;
goto destroy_macsec_object;
@@ -1677,19 +1673,6 @@ bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev)
return true;
}
-void mlx5e_macsec_get_stats_fill(struct mlx5e_macsec *macsec, void *macsec_stats)
-{
- mlx5_macsec_fs_get_stats_fill(macsec->macsec_fs, macsec_stats);
-}
-
-struct mlx5_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec)
-{
- if (!macsec)
- return NULL;
-
- return &macsec->stats;
-}
-
static const struct macsec_ops macsec_offload_ops = {
.mdo_add_txsa = mlx5e_macsec_add_txsa,
.mdo_upd_txsa = mlx5e_macsec_upd_txsa,
@@ -1827,7 +1810,7 @@ int mlx5e_macsec_init(struct mlx5e_priv *priv)
goto err_out;
}
- macsec->macsec_fs = macsec_fs;
+ mdev->macsec_fs = macsec_fs;
macsec->nb.notifier_call = macsec_obj_change_event;
mlx5_notifier_register(mdev, &macsec->nb);
@@ -1857,7 +1840,7 @@ void mlx5e_macsec_cleanup(struct mlx5e_priv *priv)
return;
mlx5_notifier_unregister(mdev, &macsec->nb);
- mlx5_macsec_fs_cleanup(macsec->macsec_fs);
+ mlx5_macsec_fs_cleanup(mdev->macsec_fs);
destroy_workqueue(macsec->wq);
mlx5e_macsec_aso_cleanup(&macsec->aso, mdev);
rhashtable_destroy(&macsec->sci_hash);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
index 47dc1d4448d9..2ecd769585f4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
@@ -37,8 +37,6 @@ static inline bool mlx5e_macsec_is_rx_flow(struct mlx5_cqe64 *cqe)
void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev, struct sk_buff *skb,
struct mlx5_cqe64 *cqe);
bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev);
-void mlx5e_macsec_get_stats_fill(struct mlx5e_macsec *macsec, void *macsec_stats);
-struct mlx5_macsec_stats *mlx5e_macsec_get_stats(struct mlx5e_macsec *macsec);
#else
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
index 8326a593b3fb..4559ee16a11a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec_stats.c
@@ -52,6 +52,7 @@ static MLX5E_DECLARE_STATS_GRP_OP_FILL_STRS(macsec_hw)
static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(macsec_hw)
{
+ struct mlx5_macsec_fs *macsec_fs;
int i;
if (!priv->macsec)
@@ -60,9 +61,10 @@ static MLX5E_DECLARE_STATS_GRP_OP_FILL_STATS(macsec_hw)
if (!mlx5e_is_macsec_device(priv->mdev))
return idx;
- mlx5e_macsec_get_stats_fill(priv->macsec, mlx5e_macsec_get_stats(priv->macsec));
+ macsec_fs = priv->mdev->macsec_fs;
+ mlx5_macsec_fs_get_stats_fill(macsec_fs, mlx5_macsec_fs_get_stats(macsec_fs));
for (i = 0; i < NUM_MACSEC_HW_COUNTERS; i++)
- data[idx++] = MLX5E_READ_CTR64_CPU(mlx5e_macsec_get_stats(priv->macsec),
+ data[idx++] = MLX5E_READ_CTR64_CPU(mlx5_macsec_fs_get_stats(macsec_fs),
mlx5e_macsec_hw_stats_desc,
i);
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 8feb7db6a220..8c19c6e16ecf 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -106,6 +106,9 @@ struct mlx5_macsec_fs {
struct net_device *netdev;
struct mlx5_macsec_tx *tx_fs;
struct mlx5_macsec_rx *rx_fs;
+
+ /* Stats manage */
+ struct mlx5_macsec_stats stats;
};
static void macsec_fs_destroy_groups(struct mlx5_macsec_flow_table *ft)
@@ -1356,6 +1359,14 @@ void mlx5_macsec_fs_get_stats_fill(struct mlx5_macsec_fs *macsec_fs, void *macse
&stats->macsec_rx_pkts_drop, &stats->macsec_rx_bytes_drop);
}
+struct mlx5_macsec_stats *mlx5_macsec_fs_get_stats(struct mlx5_macsec_fs *macsec_fs)
+{
+ if (!macsec_fs)
+ return NULL;
+
+ return &macsec_fs->stats;
+}
+
union mlx5_macsec_rule *
mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
const struct macsec_context *macsec_ctx,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
index 6a749e036e68..c0c28d08eae5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
@@ -56,6 +56,7 @@ void mlx5_macsec_fs_del_rule(struct mlx5_macsec_fs *macsec_fs,
int action);
void mlx5_macsec_fs_get_stats_fill(struct mlx5_macsec_fs *macsec_fs, void *macsec_stats);
+struct mlx5_macsec_stats *mlx5_macsec_fs_get_stats(struct mlx5_macsec_fs *macsec_fs);
#endif
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index fa70c25423b2..541c292373e6 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -808,6 +808,9 @@ struct mlx5_core_dev {
struct mlx5_thermal *thermal;
u64 num_block_tc;
u64 num_block_ipsec;
+#ifdef CONFIG_MLX5_MACSEC
+ struct mlx5_macsec_fs *macsec_fs;
+#endif
};
struct mlx5_db {
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 06/14] net/mlx5: Remove netdevice from MACsec steering
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (4 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 05/14] net/mlx5e: Move MACsec flow steering and statistics database from ethernet to core Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 07/14] net/mlx5: Maintain fs_id xarray per MACsec device inside macsec steering Leon Romanovsky
` (8 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Since MACsec steering was moved from ethernet private code to core,
remove the netdevice from the MACsec steering, and use core device
methods for error reporting instead.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../mellanox/mlx5/core/en_accel/macsec.c | 2 +-
.../mellanox/mlx5/core/lib/macsec_fs.c | 144 +++++++++---------
.../mellanox/mlx5/core/lib/macsec_fs.h | 2 +-
3 files changed, 71 insertions(+), 77 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
index 5e8ad355c67f..fb599bacbe30 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
@@ -1804,7 +1804,7 @@ int mlx5e_macsec_init(struct mlx5e_priv *priv)
macsec->mdev = mdev;
- macsec_fs = mlx5_macsec_fs_init(mdev, priv->netdev);
+ macsec_fs = mlx5_macsec_fs_init(mdev);
if (!macsec_fs) {
err = -ENOMEM;
goto err_out;
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 8c19c6e16ecf..afa48b339e4a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -2,7 +2,6 @@
/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved. */
#include <net/macsec.h>
-#include <linux/netdevice.h>
#include <linux/mlx5/qp.h>
#include <linux/if_vlan.h>
#include "fs_core.h"
@@ -103,7 +102,6 @@ union mlx5_macsec_rule {
struct mlx5_macsec_fs {
struct mlx5_core_dev *mdev;
- struct net_device *netdev;
struct mlx5_macsec_tx *tx_fs;
struct mlx5_macsec_rx *rx_fs;
@@ -268,7 +266,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
- struct net_device *netdev = macsec_fs->netdev;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
struct mlx5_flow_table_attr ft_attr = {};
struct mlx5_flow_destination dest = {};
struct mlx5_macsec_tables *tx_tables;
@@ -282,7 +280,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
u32 *flow_group_in;
int err;
- ns = mlx5_get_flow_namespace(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_EGRESS_MACSEC);
+ ns = mlx5_get_flow_namespace(mdev, MLX5_FLOW_NAMESPACE_EGRESS_MACSEC);
if (!ns)
return -ENOMEM;
@@ -307,7 +305,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
flow_table = mlx5_create_flow_table(ns, &ft_attr);
if (IS_ERR(flow_table)) {
err = PTR_ERR(flow_table);
- netdev_err(netdev, "Failed to create MACsec Tx crypto table err(%d)\n", err);
+ mlx5_core_err(mdev, "Failed to create MACsec Tx crypto table err(%d)\n", err);
goto out_flow_group;
}
ft_crypto->t = flow_table;
@@ -315,9 +313,9 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
/* Tx crypto table groups */
err = macsec_fs_tx_create_crypto_table_groups(ft_crypto);
if (err) {
- netdev_err(netdev,
- "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
+ err);
goto err;
}
@@ -331,7 +329,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
rule = mlx5_add_flow_rules(ft_crypto->t, spec, &flow_act, NULL, 0);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev, "Failed to add MACsec TX MKE rule, err=%d\n", err);
+ mlx5_core_err(mdev, "Failed to add MACsec TX MKE rule, err=%d\n", err);
goto err;
}
tx_fs->crypto_mke_rule = rule;
@@ -342,7 +340,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
rule = mlx5_add_flow_rules(ft_crypto->t, NULL, &flow_act, NULL, 0);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev, "Failed to add MACsec Tx table default miss rule %d\n", err);
+ mlx5_core_err(mdev, "Failed to add MACsec Tx table default miss rule %d\n", err);
goto err;
}
tx_tables->crypto_miss_rule = rule;
@@ -352,7 +350,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
TX_CHECK_TABLE_NUM_FTE);
if (IS_ERR(flow_table)) {
err = PTR_ERR(flow_table);
- netdev_err(netdev, "fail to create MACsec TX check table, err(%d)\n", err);
+ mlx5_core_err(mdev, "Fail to create MACsec TX check table, err(%d)\n", err);
goto err;
}
tx_tables->ft_check = flow_table;
@@ -364,9 +362,9 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
flow_group = mlx5_create_flow_group(tx_tables->ft_check, flow_group_in);
if (IS_ERR(flow_group)) {
err = PTR_ERR(flow_group);
- netdev_err(netdev,
- "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
+ err);
goto err;
}
tx_tables->ft_check_group = flow_group;
@@ -380,7 +378,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
rule = mlx5_add_flow_rules(tx_tables->ft_check, NULL, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev, "Failed to added MACsec tx check drop rule, err(%d)\n", err);
+ mlx5_core_err(mdev, "Failed to added MACsec tx check drop rule, err(%d)\n", err);
goto err;
}
tx_tables->check_miss_rule = rule;
@@ -401,7 +399,7 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
rule = mlx5_add_flow_rules(tx_tables->ft_check, spec, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev, "Failed to add MACsec check rule, err=%d\n", err);
+ mlx5_core_err(mdev, "Failed to add MACsec check rule, err=%d\n", err);
goto err;
}
tx_fs->check_rule = rule;
@@ -549,7 +547,7 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
char reformatbf[MLX5_MACSEC_TAG_LEN + MACSEC_SCI_LEN];
struct mlx5_pkt_reformat_params reformat_params = {};
struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
- struct net_device *netdev = macsec_fs->netdev;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
union mlx5_macsec_rule *macsec_rule = NULL;
struct mlx5_flow_destination dest = {};
struct mlx5_macsec_tables *tx_tables;
@@ -589,21 +587,21 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
if (is_vlan_dev(macsec_ctx->netdev))
reformat_params.param_0 = MLX5_REFORMAT_PARAM_ADD_MACSEC_OFFSET_4_BYTES;
- flow_act.pkt_reformat = mlx5_packet_reformat_alloc(macsec_fs->mdev,
+ flow_act.pkt_reformat = mlx5_packet_reformat_alloc(mdev,
&reformat_params,
MLX5_FLOW_NAMESPACE_EGRESS_MACSEC);
if (IS_ERR(flow_act.pkt_reformat)) {
err = PTR_ERR(flow_act.pkt_reformat);
- netdev_err(netdev, "Failed to allocate MACsec Tx reformat context err=%d\n", err);
+ mlx5_core_err(mdev, "Failed to allocate MACsec Tx reformat context err=%d\n", err);
goto err;
}
tx_rule->pkt_reformat = flow_act.pkt_reformat;
err = macsec_fs_tx_setup_fte(macsec_fs, spec, &flow_act, attrs->macsec_obj_id, &fs_id);
if (err) {
- netdev_err(netdev,
- "Failed to add packet reformat for MACsec TX crypto rule, err=%d\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to add packet reformat for MACsec TX crypto rule, err=%d\n",
+ err);
goto err;
}
@@ -618,7 +616,7 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
rule = mlx5_add_flow_rules(tx_tables->ft_crypto.t, spec, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev, "Failed to add MACsec TX crypto rule, err=%d\n", err);
+ mlx5_core_err(mdev, "Failed to add MACsec TX crypto rule, err=%d\n", err);
goto err;
}
tx_rule->rule = rule;
@@ -645,9 +643,9 @@ static void macsec_fs_tx_cleanup(struct mlx5_macsec_fs *macsec_fs)
tx_tables = &tx_fs->tables;
if (tx_tables->refcnt) {
- netdev_err(macsec_fs->netdev,
- "Can't destroy MACsec offload tx_fs, refcnt(%u) isn't 0\n",
- tx_tables->refcnt);
+ mlx5_core_err(mdev,
+ "Can't destroy MACsec offload tx_fs, refcnt(%u) isn't 0\n",
+ tx_tables->refcnt);
return;
}
@@ -669,7 +667,6 @@ static void macsec_fs_tx_cleanup(struct mlx5_macsec_fs *macsec_fs)
static int macsec_fs_tx_init(struct mlx5_macsec_fs *macsec_fs)
{
- struct net_device *netdev = macsec_fs->netdev;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
struct mlx5_macsec_tables *tx_tables;
struct mlx5_macsec_tx *tx_fs;
@@ -685,9 +682,9 @@ static int macsec_fs_tx_init(struct mlx5_macsec_fs *macsec_fs)
flow_counter = mlx5_fc_create(mdev, false);
if (IS_ERR(flow_counter)) {
err = PTR_ERR(flow_counter);
- netdev_err(netdev,
- "Failed to create MACsec Tx encrypt flow counter, err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create MACsec Tx encrypt flow counter, err(%d)\n",
+ err);
goto err_encrypt_counter;
}
tx_tables->check_rule_counter = flow_counter;
@@ -695,9 +692,9 @@ static int macsec_fs_tx_init(struct mlx5_macsec_fs *macsec_fs)
flow_counter = mlx5_fc_create(mdev, false);
if (IS_ERR(flow_counter)) {
err = PTR_ERR(flow_counter);
- netdev_err(netdev,
- "Failed to create MACsec Tx drop flow counter, err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create MACsec Tx drop flow counter, err(%d)\n",
+ err);
goto err_drop_counter;
}
tx_tables->check_miss_rule_counter = flow_counter;
@@ -857,7 +854,7 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs
u8 mlx5_reformat_buf[MLX5_SECTAG_HEADER_SIZE_WITH_SCI];
struct mlx5_pkt_reformat_params reformat_params = {};
struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
- struct net_device *netdev = macsec_fs->netdev;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
struct mlx5_macsec_tables *rx_tables;
struct mlx5_flow_handle *rule;
int err = 0;
@@ -872,12 +869,12 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs
reformat_params.type = MLX5_REFORMAT_TYPE_DEL_MACSEC;
reformat_params.size = reformat_param_size;
reformat_params.data = mlx5_reformat_buf;
- flow_act->pkt_reformat = mlx5_packet_reformat_alloc(macsec_fs->mdev,
+ flow_act->pkt_reformat = mlx5_packet_reformat_alloc(mdev,
&reformat_params,
MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC);
if (IS_ERR(flow_act->pkt_reformat)) {
err = PTR_ERR(flow_act->pkt_reformat);
- netdev_err(netdev, "Failed to allocate MACsec Rx reformat context err=%d\n", err);
+ mlx5_core_err(mdev, "Failed to allocate MACsec Rx reformat context err=%d\n", err);
return err;
}
rx_fs->check_rule_pkt_reformat[rule_index] = flow_act->pkt_reformat;
@@ -909,7 +906,7 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs
rule = mlx5_add_flow_rules(rx_tables->ft_check, spec, flow_act, dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev, "Failed to add MACsec Rx check rule, err=%d\n", err);
+ mlx5_core_err(mdev, "Failed to add MACsec Rx check rule, err=%d\n", err);
return err;
}
@@ -922,7 +919,7 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
- struct net_device *netdev = macsec_fs->netdev;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
struct mlx5_macsec_flow_table *ft_crypto;
struct mlx5_flow_table_attr ft_attr = {};
struct mlx5_flow_destination dest = {};
@@ -936,7 +933,7 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
u32 *flow_group_in;
int err;
- ns = mlx5_get_flow_namespace(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC);
+ ns = mlx5_get_flow_namespace(mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC);
if (!ns)
return -ENOMEM;
@@ -960,7 +957,7 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
flow_table = mlx5_create_flow_table(ns, &ft_attr);
if (IS_ERR(flow_table)) {
err = PTR_ERR(flow_table);
- netdev_err(netdev, "Failed to create MACsec Rx crypto table err(%d)\n", err);
+ mlx5_core_err(mdev, "Failed to create MACsec Rx crypto table err(%d)\n", err);
goto out_flow_group;
}
ft_crypto->t = flow_table;
@@ -968,9 +965,9 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
/* Rx crypto table groups */
err = macsec_fs_rx_create_crypto_table_groups(ft_crypto);
if (err) {
- netdev_err(netdev,
- "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create default flow group for MACsec Tx crypto table err(%d)\n",
+ err);
goto err;
}
@@ -978,9 +975,9 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
rule = mlx5_add_flow_rules(ft_crypto->t, NULL, &flow_act, NULL, 0);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev,
- "Failed to add MACsec Rx crypto table default miss rule %d\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to add MACsec Rx crypto table default miss rule %d\n",
+ err);
goto err;
}
rx_tables->crypto_miss_rule = rule;
@@ -992,7 +989,7 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
RX_CHECK_TABLE_NUM_FTE);
if (IS_ERR(flow_table)) {
err = PTR_ERR(flow_table);
- netdev_err(netdev, "fail to create MACsec RX check table, err(%d)\n", err);
+ mlx5_core_err(mdev, "Fail to create MACsec RX check table, err(%d)\n", err);
goto err;
}
rx_tables->ft_check = flow_table;
@@ -1003,9 +1000,9 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
flow_group = mlx5_create_flow_group(rx_tables->ft_check, flow_group_in);
if (IS_ERR(flow_group)) {
err = PTR_ERR(flow_group);
- netdev_err(netdev,
- "Failed to create default flow group for MACsec Rx check table err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create default flow group for MACsec Rx check table err(%d)\n",
+ err);
goto err;
}
rx_tables->ft_check_group = flow_group;
@@ -1019,7 +1016,7 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
rule = mlx5_add_flow_rules(rx_tables->ft_check, NULL, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev, "Failed to added MACsec Rx check drop rule, err(%d)\n", err);
+ mlx5_core_err(mdev, "Failed to added MACsec Rx check drop rule, err(%d)\n", err);
goto err;
}
rx_tables->check_miss_rule = rule;
@@ -1148,7 +1145,7 @@ macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
{
u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
- struct net_device *netdev = macsec_fs->netdev;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
union mlx5_macsec_rule *macsec_rule = NULL;
struct mlx5_modify_hdr *modify_hdr = NULL;
struct mlx5_macsec_flow_table *ft_crypto;
@@ -1186,11 +1183,11 @@ macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
MLX5_SET(set_action_in, action, offset, 0);
MLX5_SET(set_action_in, action, length, 32);
- modify_hdr = mlx5_modify_header_alloc(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC,
+ modify_hdr = mlx5_modify_header_alloc(mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC,
1, action);
if (IS_ERR(modify_hdr)) {
err = PTR_ERR(modify_hdr);
- netdev_err(netdev, "fail to alloc MACsec set modify_header_id err=%d\n", err);
+ mlx5_core_err(mdev, "Fail to alloc MACsec set modify_header_id err=%d\n", err);
modify_hdr = NULL;
goto err;
}
@@ -1209,9 +1206,9 @@ macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
rule = mlx5_add_flow_rules(ft_crypto->t, spec, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev,
- "Failed to add SA with SCI rule to Rx crypto rule, err=%d\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to add SA with SCI rule to Rx crypto rule, err=%d\n",
+ err);
goto err;
}
rx_rule->rule[0] = rule;
@@ -1234,9 +1231,9 @@ macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
rule = mlx5_add_flow_rules(ft_crypto->t, spec, &flow_act, &dest, 1);
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
- netdev_err(netdev,
- "Failed to add SA without SCI rule to Rx crypto rule, err=%d\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to add SA without SCI rule to Rx crypto rule, err=%d\n",
+ err);
goto err;
}
rx_rule->rule[1] = rule;
@@ -1255,7 +1252,6 @@ macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
static int macsec_fs_rx_init(struct mlx5_macsec_fs *macsec_fs)
{
- struct net_device *netdev = macsec_fs->netdev;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
struct mlx5_macsec_tables *rx_tables;
struct mlx5_macsec_rx *rx_fs;
@@ -1269,9 +1265,9 @@ static int macsec_fs_rx_init(struct mlx5_macsec_fs *macsec_fs)
flow_counter = mlx5_fc_create(mdev, false);
if (IS_ERR(flow_counter)) {
err = PTR_ERR(flow_counter);
- netdev_err(netdev,
- "Failed to create MACsec Rx encrypt flow counter, err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create MACsec Rx encrypt flow counter, err(%d)\n",
+ err);
goto err_encrypt_counter;
}
@@ -1281,9 +1277,9 @@ static int macsec_fs_rx_init(struct mlx5_macsec_fs *macsec_fs)
flow_counter = mlx5_fc_create(mdev, false);
if (IS_ERR(flow_counter)) {
err = PTR_ERR(flow_counter);
- netdev_err(netdev,
- "Failed to create MACsec Rx drop flow counter, err(%d)\n",
- err);
+ mlx5_core_err(mdev,
+ "Failed to create MACsec Rx drop flow counter, err(%d)\n",
+ err);
goto err_drop_counter;
}
rx_tables->check_miss_rule_counter = flow_counter;
@@ -1315,9 +1311,9 @@ static void macsec_fs_rx_cleanup(struct mlx5_macsec_fs *macsec_fs)
rx_tables = &rx_fs->tables;
if (rx_tables->refcnt) {
- netdev_err(macsec_fs->netdev,
- "Can't destroy MACsec offload rx_fs, refcnt(%u) isn't 0\n",
- rx_tables->refcnt);
+ mlx5_core_err(mdev,
+ "Can't destroy MACsec offload rx_fs, refcnt(%u) isn't 0\n",
+ rx_tables->refcnt);
return;
}
@@ -1395,8 +1391,7 @@ void mlx5_macsec_fs_cleanup(struct mlx5_macsec_fs *macsec_fs)
}
struct mlx5_macsec_fs *
-mlx5_macsec_fs_init(struct mlx5_core_dev *mdev,
- struct net_device *netdev)
+mlx5_macsec_fs_init(struct mlx5_core_dev *mdev)
{
struct mlx5_macsec_fs *macsec_fs;
int err;
@@ -1406,17 +1401,16 @@ mlx5_macsec_fs_init(struct mlx5_core_dev *mdev,
return NULL;
macsec_fs->mdev = mdev;
- macsec_fs->netdev = netdev;
err = macsec_fs_tx_init(macsec_fs);
if (err) {
- netdev_err(netdev, "MACsec offload: Failed to init tx_fs, err=%d\n", err);
+ mlx5_core_err(mdev, "MACsec offload: Failed to init tx_fs, err=%d\n", err);
goto err;
}
err = macsec_fs_rx_init(macsec_fs);
if (err) {
- netdev_err(netdev, "MACsec offload: Failed to init tx_fs, err=%d\n", err);
+ mlx5_core_err(mdev, "MACsec offload: Failed to init tx_fs, err=%d\n", err);
goto tx_cleanup;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
index c0c28d08eae5..f007c33369c2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
@@ -43,7 +43,7 @@ enum mlx5_macsec_action {
void mlx5_macsec_fs_cleanup(struct mlx5_macsec_fs *macsec_fs);
struct mlx5_macsec_fs *
-mlx5_macsec_fs_init(struct mlx5_core_dev *mdev, struct net_device *netdev);
+mlx5_macsec_fs_init(struct mlx5_core_dev *mdev);
union mlx5_macsec_rule *
mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 07/14] net/mlx5: Maintain fs_id xarray per MACsec device inside macsec steering
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (5 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 06/14] net/mlx5: Remove netdevice from MACsec steering Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 08/14] RDMA/mlx5: Implement MACsec gid addition and deletion Leon Romanovsky
` (7 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Remove fs_id from the MACsec SA, since it has no real usage there and
instead maintain with the MACsec steering data inside the core.
Downstream patches requires this change to facilitate IB driver accesses
to the fs_ids to avoid RoCE MACsec dependency on EN driver.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../mellanox/mlx5/core/en_accel/macsec.c | 115 +++-----
.../mellanox/mlx5/core/lib/macsec_fs.c | 247 +++++++++++++++++-
.../mellanox/mlx5/core/lib/macsec_fs.h | 3 +-
3 files changed, 271 insertions(+), 94 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
index fb599bacbe30..b4f3f4f10af3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
@@ -65,8 +65,6 @@ struct mlx5e_macsec_sa {
ssci_t ssci;
salt_t salt;
- struct rhash_head hash;
- u32 fs_id;
union mlx5_macsec_rule *macsec_rule;
struct rcu_head rcu_head;
struct mlx5e_macsec_epn_state epn_state;
@@ -105,14 +103,6 @@ struct mlx5e_macsec_aso {
u32 pdn;
};
-static const struct rhashtable_params rhash_sci = {
- .key_len = sizeof_field(struct mlx5e_macsec_sa, sci),
- .key_offset = offsetof(struct mlx5e_macsec_sa, sci),
- .head_offset = offsetof(struct mlx5e_macsec_sa, hash),
- .automatic_shrinking = true,
- .min_size = 1,
-};
-
struct mlx5e_macsec_device {
const struct net_device *netdev;
struct mlx5e_macsec_sa *tx_sa[MACSEC_NUM_AN];
@@ -126,9 +116,6 @@ struct mlx5e_macsec {
int num_of_devices;
struct mutex lock; /* Protects mlx5e_macsec internal contexts */
- /* Tx sci -> fs id mapping handling */
- struct rhashtable sci_hash; /* sci -> mlx5e_macsec_sa */
-
/* Rx fs_id -> rx_sc mapping */
struct xarray sc_xarray;
@@ -325,29 +312,23 @@ static void mlx5e_macsec_destroy_object(struct mlx5_core_dev *mdev, u32 macsec_o
static void mlx5e_macsec_cleanup_sa(struct mlx5e_macsec *macsec,
struct mlx5e_macsec_sa *sa,
- bool is_tx)
+ bool is_tx, struct net_device *netdev, u32 fs_id)
{
int action = (is_tx) ? MLX5_ACCEL_MACSEC_ACTION_ENCRYPT :
MLX5_ACCEL_MACSEC_ACTION_DECRYPT;
- if ((is_tx) && sa->fs_id) {
- /* Make sure ongoing datapath readers sees a valid SA */
- rhashtable_remove_fast(&macsec->sci_hash, &sa->hash, rhash_sci);
- sa->fs_id = 0;
- }
-
if (!sa->macsec_rule)
return;
- mlx5_macsec_fs_del_rule(macsec->mdev->macsec_fs, sa->macsec_rule, action);
+ mlx5_macsec_fs_del_rule(macsec->mdev->macsec_fs, sa->macsec_rule, action, netdev,
+ fs_id);
mlx5e_macsec_destroy_object(macsec->mdev, sa->macsec_obj_id);
sa->macsec_rule = NULL;
}
static int mlx5e_macsec_init_sa(struct macsec_context *ctx,
struct mlx5e_macsec_sa *sa,
- bool encrypt,
- bool is_tx)
+ bool encrypt, bool is_tx, u32 *fs_id)
{
struct mlx5e_priv *priv = macsec_netdev_priv(ctx->netdev);
struct mlx5e_macsec *macsec = priv->macsec;
@@ -382,7 +363,7 @@ static int mlx5e_macsec_init_sa(struct macsec_context *ctx,
rule_attrs.action = (is_tx) ? MLX5_ACCEL_MACSEC_ACTION_ENCRYPT :
MLX5_ACCEL_MACSEC_ACTION_DECRYPT;
- macsec_rule = mlx5_macsec_fs_add_rule(mdev->macsec_fs, ctx, &rule_attrs, &sa->fs_id);
+ macsec_rule = mlx5_macsec_fs_add_rule(mdev->macsec_fs, ctx, &rule_attrs, fs_id);
if (!macsec_rule) {
err = -ENOMEM;
goto destroy_macsec_object;
@@ -390,16 +371,8 @@ static int mlx5e_macsec_init_sa(struct macsec_context *ctx,
sa->macsec_rule = macsec_rule;
- if (is_tx) {
- err = rhashtable_insert_fast(&macsec->sci_hash, &sa->hash, rhash_sci);
- if (err)
- goto destroy_macsec_object_and_rule;
- }
-
return 0;
-destroy_macsec_object_and_rule:
- mlx5e_macsec_cleanup_sa(macsec, sa, is_tx);
destroy_macsec_object:
mlx5e_macsec_destroy_object(mdev, sa->macsec_obj_id);
@@ -421,7 +394,7 @@ mlx5e_macsec_get_rx_sc_from_sc_list(const struct list_head *list, sci_t sci)
static int macsec_rx_sa_active_update(struct macsec_context *ctx,
struct mlx5e_macsec_sa *rx_sa,
- bool active)
+ bool active, u32 *fs_id)
{
struct mlx5e_priv *priv = macsec_netdev_priv(ctx->netdev);
struct mlx5e_macsec *macsec = priv->macsec;
@@ -432,11 +405,11 @@ static int macsec_rx_sa_active_update(struct macsec_context *ctx,
rx_sa->active = active;
if (!active) {
- mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+ mlx5e_macsec_cleanup_sa(macsec, rx_sa, false, ctx->secy->netdev, *fs_id);
return 0;
}
- err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false);
+ err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false, fs_id);
if (err)
rx_sa->active = false;
@@ -558,7 +531,7 @@ static int mlx5e_macsec_add_txsa(struct macsec_context *ctx)
!tx_sa->active)
goto out;
- err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true);
+ err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true, NULL);
if (err)
goto destroy_encryption_key;
@@ -622,7 +595,7 @@ static int mlx5e_macsec_upd_txsa(struct macsec_context *ctx)
goto out;
if (ctx_tx_sa->active) {
- err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true);
+ err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true, NULL);
if (err)
goto out;
} else {
@@ -631,7 +604,7 @@ static int mlx5e_macsec_upd_txsa(struct macsec_context *ctx)
goto out;
}
- mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+ mlx5e_macsec_cleanup_sa(macsec, tx_sa, true, ctx->secy->netdev, 0);
}
out:
mutex_unlock(&macsec->lock);
@@ -664,7 +637,7 @@ static int mlx5e_macsec_del_txsa(struct macsec_context *ctx)
goto out;
}
- mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+ mlx5e_macsec_cleanup_sa(macsec, tx_sa, true, ctx->secy->netdev, 0);
mlx5_destroy_encryption_key(macsec->mdev, tx_sa->enc_key_id);
kfree_rcu_mightsleep(tx_sa);
macsec_device->tx_sa[assoc_num] = NULL;
@@ -675,20 +648,6 @@ static int mlx5e_macsec_del_txsa(struct macsec_context *ctx)
return err;
}
-static u32 mlx5e_macsec_get_sa_from_hashtable(struct rhashtable *sci_hash, sci_t *sci)
-{
- struct mlx5e_macsec_sa *macsec_sa;
- u32 fs_id = 0;
-
- rcu_read_lock();
- macsec_sa = rhashtable_lookup(sci_hash, sci, rhash_sci);
- if (macsec_sa)
- fs_id = macsec_sa->fs_id;
- rcu_read_unlock();
-
- return fs_id;
-}
-
static int mlx5e_macsec_add_rxsc(struct macsec_context *ctx)
{
struct mlx5e_macsec_rx_sc_xarray_element *sc_xarray_element;
@@ -808,7 +767,8 @@ static int mlx5e_macsec_upd_rxsc(struct macsec_context *ctx)
if (!rx_sa)
continue;
- err = macsec_rx_sa_active_update(ctx, rx_sa, rx_sa->active && ctx_rx_sc->active);
+ err = macsec_rx_sa_active_update(ctx, rx_sa, rx_sa->active && ctx_rx_sc->active,
+ &rx_sc->sc_xarray_element->fs_id);
if (err)
goto out;
}
@@ -819,7 +779,8 @@ static int mlx5e_macsec_upd_rxsc(struct macsec_context *ctx)
return err;
}
-static void macsec_del_rxsc_ctx(struct mlx5e_macsec *macsec, struct mlx5e_macsec_rx_sc *rx_sc)
+static void macsec_del_rxsc_ctx(struct mlx5e_macsec *macsec, struct mlx5e_macsec_rx_sc *rx_sc,
+ struct net_device *netdev)
{
struct mlx5e_macsec_sa *rx_sa;
int i;
@@ -829,7 +790,8 @@ static void macsec_del_rxsc_ctx(struct mlx5e_macsec *macsec, struct mlx5e_macsec
if (!rx_sa)
continue;
- mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+ mlx5e_macsec_cleanup_sa(macsec, rx_sa, false, netdev,
+ rx_sc->sc_xarray_element->fs_id);
mlx5_destroy_encryption_key(macsec->mdev, rx_sa->enc_key_id);
kfree(rx_sa);
@@ -877,7 +839,7 @@ static int mlx5e_macsec_del_rxsc(struct macsec_context *ctx)
goto out;
}
- macsec_del_rxsc_ctx(macsec, rx_sc);
+ macsec_del_rxsc_ctx(macsec, rx_sc, ctx->secy->netdev);
out:
mutex_unlock(&macsec->lock);
@@ -936,7 +898,6 @@ static int mlx5e_macsec_add_rxsa(struct macsec_context *ctx)
rx_sa->next_pn = ctx_rx_sa->next_pn;
rx_sa->sci = sci;
rx_sa->assoc_num = assoc_num;
- rx_sa->fs_id = rx_sc->sc_xarray_element->fs_id;
if (ctx->secy->xpn)
update_macsec_epn(rx_sa, &ctx_rx_sa->key, &ctx_rx_sa->next_pn_halves,
@@ -953,7 +914,7 @@ static int mlx5e_macsec_add_rxsa(struct macsec_context *ctx)
goto out;
//TODO - add support for both authentication and encryption flows
- err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false);
+ err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false, &rx_sc->sc_xarray_element->fs_id);
if (err)
goto destroy_encryption_key;
@@ -1020,7 +981,8 @@ static int mlx5e_macsec_upd_rxsa(struct macsec_context *ctx)
goto out;
}
- err = macsec_rx_sa_active_update(ctx, rx_sa, ctx_rx_sa->active);
+ err = macsec_rx_sa_active_update(ctx, rx_sa, ctx_rx_sa->active,
+ &rx_sc->sc_xarray_element->fs_id);
out:
mutex_unlock(&macsec->lock);
@@ -1068,7 +1030,8 @@ static int mlx5e_macsec_del_rxsa(struct macsec_context *ctx)
goto out;
}
- mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+ mlx5e_macsec_cleanup_sa(macsec, rx_sa, false, ctx->secy->netdev,
+ rx_sc->sc_xarray_element->fs_id);
mlx5_destroy_encryption_key(macsec->mdev, rx_sa->enc_key_id);
kfree(rx_sa);
rx_sc->rx_sa[assoc_num] = NULL;
@@ -1149,7 +1112,8 @@ static int macsec_upd_secy_hw_address(struct macsec_context *ctx,
if (!rx_sa || !rx_sa->macsec_rule)
continue;
- mlx5e_macsec_cleanup_sa(macsec, rx_sa, false);
+ mlx5e_macsec_cleanup_sa(macsec, rx_sa, false, ctx->secy->netdev,
+ rx_sc->sc_xarray_element->fs_id);
}
}
@@ -1160,7 +1124,8 @@ static int macsec_upd_secy_hw_address(struct macsec_context *ctx,
continue;
if (rx_sa->active) {
- err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false);
+ err = mlx5e_macsec_init_sa(ctx, rx_sa, true, false,
+ &rx_sc->sc_xarray_element->fs_id);
if (err)
goto out;
}
@@ -1213,7 +1178,7 @@ static int mlx5e_macsec_upd_secy(struct macsec_context *ctx)
if (!tx_sa)
continue;
- mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+ mlx5e_macsec_cleanup_sa(macsec, tx_sa, true, ctx->secy->netdev, 0);
}
for (i = 0; i < MACSEC_NUM_AN; ++i) {
@@ -1222,7 +1187,7 @@ static int mlx5e_macsec_upd_secy(struct macsec_context *ctx)
continue;
if (tx_sa->assoc_num == tx_sc->encoding_sa && tx_sa->active) {
- err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true);
+ err = mlx5e_macsec_init_sa(ctx, tx_sa, tx_sc->encrypt, true, NULL);
if (err)
goto out;
}
@@ -1260,7 +1225,7 @@ static int mlx5e_macsec_del_secy(struct macsec_context *ctx)
if (!tx_sa)
continue;
- mlx5e_macsec_cleanup_sa(macsec, tx_sa, true);
+ mlx5e_macsec_cleanup_sa(macsec, tx_sa, true, ctx->secy->netdev, 0);
mlx5_destroy_encryption_key(macsec->mdev, tx_sa->enc_key_id);
kfree(tx_sa);
macsec_device->tx_sa[i] = NULL;
@@ -1268,7 +1233,7 @@ static int mlx5e_macsec_del_secy(struct macsec_context *ctx)
list = &macsec_device->macsec_rx_sc_list_head;
list_for_each_entry_safe(rx_sc, tmp, list, rx_sc_list_element)
- macsec_del_rxsc_ctx(macsec, rx_sc);
+ macsec_del_rxsc_ctx(macsec, rx_sc, ctx->secy->netdev);
kfree(macsec_device->dev_addr);
macsec_device->dev_addr = NULL;
@@ -1693,7 +1658,8 @@ bool mlx5e_macsec_handle_tx_skb(struct mlx5e_macsec *macsec, struct sk_buff *skb
struct metadata_dst *md_dst = skb_metadata_dst(skb);
u32 fs_id;
- fs_id = mlx5e_macsec_get_sa_from_hashtable(&macsec->sci_hash, &md_dst->u.macsec_info.sci);
+ fs_id = mlx5_macsec_fs_get_fs_id_from_hashtable(macsec->mdev->macsec_fs,
+ &md_dst->u.macsec_info.sci);
if (!fs_id)
goto err_out;
@@ -1711,7 +1677,8 @@ void mlx5e_macsec_tx_build_eseg(struct mlx5e_macsec *macsec,
struct metadata_dst *md_dst = skb_metadata_dst(skb);
u32 fs_id;
- fs_id = mlx5e_macsec_get_sa_from_hashtable(&macsec->sci_hash, &md_dst->u.macsec_info.sci);
+ fs_id = mlx5_macsec_fs_get_fs_id_from_hashtable(macsec->mdev->macsec_fs,
+ &md_dst->u.macsec_info.sci);
if (!fs_id)
return;
@@ -1779,13 +1746,6 @@ int mlx5e_macsec_init(struct mlx5e_priv *priv)
INIT_LIST_HEAD(&macsec->macsec_device_list_head);
mutex_init(&macsec->lock);
- err = rhashtable_init(&macsec->sci_hash, &rhash_sci);
- if (err) {
- mlx5_core_err(mdev, "MACsec offload: Failed to init SCI hash table, err=%d\n",
- err);
- goto err_hash;
- }
-
err = mlx5e_macsec_aso_init(&macsec->aso, priv->mdev);
if (err) {
mlx5_core_err(mdev, "MACsec offload: Failed to init aso, err=%d\n", err);
@@ -1824,8 +1784,6 @@ int mlx5e_macsec_init(struct mlx5e_priv *priv)
err_wq:
mlx5e_macsec_aso_cleanup(&macsec->aso, priv->mdev);
err_aso:
- rhashtable_destroy(&macsec->sci_hash);
-err_hash:
kfree(macsec);
priv->macsec = NULL;
return err;
@@ -1843,7 +1801,6 @@ void mlx5e_macsec_cleanup(struct mlx5e_priv *priv)
mlx5_macsec_fs_cleanup(mdev->macsec_fs);
destroy_workqueue(macsec->wq);
mlx5e_macsec_aso_cleanup(&macsec->aso, mdev);
- rhashtable_destroy(&macsec->sci_hash);
mutex_destroy(&macsec->lock);
kfree(macsec);
}
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 afa48b339e4a..d39ca7c66542 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -74,6 +74,20 @@ struct mlx5_macsec_tables {
u32 refcnt;
};
+struct mlx5_fs_id {
+ u32 id;
+ refcount_t refcnt;
+ sci_t sci;
+ struct rhash_head hash;
+};
+
+struct mlx5_macsec_device {
+ struct list_head macsec_devices_list_entry;
+ void *macdev;
+ struct xarray tx_id_xa;
+ struct xarray rx_id_xa;
+};
+
struct mlx5_macsec_tx {
struct mlx5_flow_handle *crypto_mke_rule;
struct mlx5_flow_handle *check_rule;
@@ -100,6 +114,22 @@ union mlx5_macsec_rule {
struct mlx5_macsec_rx_rule rx_rule;
};
+static const struct rhashtable_params rhash_sci = {
+ .key_len = sizeof_field(struct mlx5_fs_id, sci),
+ .key_offset = offsetof(struct mlx5_fs_id, sci),
+ .head_offset = offsetof(struct mlx5_fs_id, hash),
+ .automatic_shrinking = true,
+ .min_size = 1,
+};
+
+static const struct rhashtable_params rhash_fs_id = {
+ .key_len = sizeof_field(struct mlx5_fs_id, id),
+ .key_offset = offsetof(struct mlx5_fs_id, id),
+ .head_offset = offsetof(struct mlx5_fs_id, hash),
+ .automatic_shrinking = true,
+ .min_size = 1,
+};
+
struct mlx5_macsec_fs {
struct mlx5_core_dev *mdev;
struct mlx5_macsec_tx *tx_fs;
@@ -107,6 +137,15 @@ struct mlx5_macsec_fs {
/* Stats manage */
struct mlx5_macsec_stats stats;
+
+ /* Tx sci -> fs id mapping handling */
+ struct rhashtable sci_hash; /* sci -> mlx5_fs_id */
+
+ /* RX fs_id -> mlx5_fs_id mapping handling */
+ struct rhashtable fs_id_hash; /* fs_id -> mlx5_fs_id */
+
+ /* TX & RX fs_id lists per macsec device */
+ struct list_head macsec_devices_list;
};
static void macsec_fs_destroy_groups(struct mlx5_macsec_flow_table *ft)
@@ -513,9 +552,137 @@ static void macsec_fs_tx_create_sectag_header(const struct macsec_context *ctx,
memcpy(reformatbf, §ag, *reformat_size);
}
+static bool macsec_fs_is_macsec_device_empty(struct mlx5_macsec_device *macsec_device)
+{
+ if (xa_empty(&macsec_device->tx_id_xa) &&
+ xa_empty(&macsec_device->rx_id_xa))
+ return true;
+
+ return false;
+}
+
+static void macsec_fs_id_del(struct list_head *macsec_devices_list, u32 fs_id,
+ void *macdev, struct rhashtable *hash_table, bool is_tx)
+{
+ const struct rhashtable_params *rhash = (is_tx) ? &rhash_sci : &rhash_fs_id;
+ struct mlx5_macsec_device *iter, *macsec_device = NULL;
+ struct mlx5_fs_id *fs_id_found;
+ struct xarray *fs_id_xa;
+
+ list_for_each_entry(iter, macsec_devices_list, macsec_devices_list_entry) {
+ if (iter->macdev == macdev) {
+ macsec_device = iter;
+ break;
+ }
+ }
+ WARN_ON(!macsec_device);
+
+ fs_id_xa = (is_tx) ? &macsec_device->tx_id_xa :
+ &macsec_device->rx_id_xa;
+ xa_lock(fs_id_xa);
+ fs_id_found = xa_load(fs_id_xa, fs_id);
+ WARN_ON(!fs_id_found);
+
+ if (!refcount_dec_and_test(&fs_id_found->refcnt)) {
+ xa_unlock(fs_id_xa);
+ return;
+ }
+
+ if (fs_id_found->id) {
+ /* Make sure ongoing datapath readers sees a valid SA */
+ rhashtable_remove_fast(hash_table, &fs_id_found->hash, *rhash);
+ fs_id_found->id = 0;
+ }
+ xa_unlock(fs_id_xa);
+
+ xa_erase(fs_id_xa, fs_id);
+
+ kfree(fs_id_found);
+
+ if (macsec_fs_is_macsec_device_empty(macsec_device)) {
+ list_del(&macsec_device->macsec_devices_list_entry);
+ kfree(macsec_device);
+ }
+}
+
+static int macsec_fs_id_add(struct list_head *macsec_devices_list, u32 fs_id,
+ void *macdev, struct rhashtable *hash_table, sci_t sci,
+ bool is_tx)
+{
+ const struct rhashtable_params *rhash = (is_tx) ? &rhash_sci : &rhash_fs_id;
+ struct mlx5_macsec_device *iter, *macsec_device = NULL;
+ struct mlx5_fs_id *fs_id_iter;
+ struct xarray *fs_id_xa;
+ int err;
+
+ if (!is_tx) {
+ rcu_read_lock();
+ fs_id_iter = rhashtable_lookup(hash_table, &fs_id, rhash_fs_id);
+ if (fs_id_iter) {
+ refcount_inc(&fs_id_iter->refcnt);
+ rcu_read_unlock();
+ return 0;
+ }
+ rcu_read_unlock();
+ }
+
+ fs_id_iter = kzalloc(sizeof(*fs_id_iter), GFP_KERNEL);
+ if (!fs_id_iter)
+ return -ENOMEM;
+
+ list_for_each_entry(iter, macsec_devices_list, macsec_devices_list_entry) {
+ if (iter->macdev == macdev) {
+ macsec_device = iter;
+ break;
+ }
+ }
+
+ if (!macsec_device) { /* first time adding a SA to that device */
+ macsec_device = kzalloc(sizeof(*macsec_device), GFP_KERNEL);
+ if (!macsec_device) {
+ err = -ENOMEM;
+ goto err_alloc_dev;
+ }
+ macsec_device->macdev = macdev;
+ xa_init(&macsec_device->tx_id_xa);
+ xa_init(&macsec_device->rx_id_xa);
+ list_add(&macsec_device->macsec_devices_list_entry, macsec_devices_list);
+ }
+
+ fs_id_xa = (is_tx) ? &macsec_device->tx_id_xa :
+ &macsec_device->rx_id_xa;
+ fs_id_iter->id = fs_id;
+ refcount_set(&fs_id_iter->refcnt, 1);
+ fs_id_iter->sci = sci;
+ err = xa_err(xa_store(fs_id_xa, fs_id, fs_id_iter, GFP_KERNEL));
+ if (err)
+ goto err_store_id;
+
+ err = rhashtable_insert_fast(hash_table, &fs_id_iter->hash, *rhash);
+ if (err)
+ goto err_hash_insert;
+
+ return 0;
+
+err_hash_insert:
+ xa_erase(fs_id_xa, fs_id);
+err_store_id:
+ if (macsec_fs_is_macsec_device_empty(macsec_device)) {
+ list_del(&macsec_device->macsec_devices_list_entry);
+ kfree(macsec_device);
+ }
+err_alloc_dev:
+ kfree(fs_id_iter);
+ return err;
+}
+
static void macsec_fs_tx_del_rule(struct mlx5_macsec_fs *macsec_fs,
- struct mlx5_macsec_tx_rule *tx_rule)
+ struct mlx5_macsec_tx_rule *tx_rule,
+ void *macdev)
{
+ macsec_fs_id_del(&macsec_fs->macsec_devices_list, tx_rule->fs_id, macdev,
+ &macsec_fs->sci_hash, true);
+
if (tx_rule->rule) {
mlx5_del_flow_rules(tx_rule->rule);
tx_rule->rule = NULL;
@@ -541,8 +708,7 @@ static void macsec_fs_tx_del_rule(struct mlx5_macsec_fs *macsec_fs,
static union mlx5_macsec_rule *
macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
const struct macsec_context *macsec_ctx,
- struct mlx5_macsec_rule_attrs *attrs,
- u32 *sa_fs_id)
+ struct mlx5_macsec_rule_attrs *attrs)
{
char reformatbf[MLX5_MACSEC_TAG_LEN + MACSEC_SCI_LEN];
struct mlx5_pkt_reformat_params reformat_params = {};
@@ -606,7 +772,6 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
}
tx_rule->fs_id = fs_id;
- *sa_fs_id = fs_id;
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
MLX5_FLOW_CONTEXT_ACTION_CRYPTO_ENCRYPT |
@@ -621,10 +786,17 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
}
tx_rule->rule = rule;
+ err = macsec_fs_id_add(&macsec_fs->macsec_devices_list, fs_id, macsec_ctx->secy->netdev,
+ &macsec_fs->sci_hash, attrs->sci, true);
+ if (err) {
+ mlx5_core_err(mdev, "Failed to save fs_id, err=%d\n", err);
+ goto err;
+ }
+
goto out_spec;
err:
- macsec_fs_tx_del_rule(macsec_fs, tx_rule);
+ macsec_fs_tx_del_rule(macsec_fs, tx_rule, macsec_ctx->secy->netdev);
macsec_rule = NULL;
out_spec:
kvfree(spec);
@@ -700,6 +872,7 @@ static int macsec_fs_tx_init(struct mlx5_macsec_fs *macsec_fs)
tx_tables->check_miss_rule_counter = flow_counter;
ida_init(&tx_fs->tx_halloc);
+ INIT_LIST_HEAD(&macsec_fs->macsec_devices_list);
macsec_fs->tx_fs = tx_fs;
@@ -1071,10 +1244,14 @@ static void macsec_fs_rx_ft_put(struct mlx5_macsec_fs *macsec_fs)
}
static void macsec_fs_rx_del_rule(struct mlx5_macsec_fs *macsec_fs,
- struct mlx5_macsec_rx_rule *rx_rule)
+ struct mlx5_macsec_rx_rule *rx_rule,
+ void *macdev, u32 fs_id)
{
int i;
+ macsec_fs_id_del(&macsec_fs->macsec_devices_list, fs_id, macdev,
+ &macsec_fs->fs_id_hash, false);
+
for (i = 0; i < RX_NUM_OF_RULES_PER_SA; ++i) {
if (rx_rule->rule[i]) {
mlx5_del_flow_rules(rx_rule->rule[i]);
@@ -1140,6 +1317,7 @@ static void macsec_fs_rx_setup_fte(struct mlx5_flow_spec *spec,
static union mlx5_macsec_rule *
macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
+ const struct macsec_context *macsec_ctx,
struct mlx5_macsec_rule_attrs *attrs,
u32 fs_id)
{
@@ -1239,11 +1417,18 @@ macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
rx_rule->rule[1] = rule;
}
+ err = macsec_fs_id_add(&macsec_fs->macsec_devices_list, fs_id, macsec_ctx->secy->netdev,
+ &macsec_fs->fs_id_hash, attrs->sci, false);
+ if (err) {
+ mlx5_core_err(mdev, "Failed to save fs_id, err=%d\n", err);
+ goto err;
+ }
+
kvfree(spec);
return macsec_rule;
err:
- macsec_fs_rx_del_rule(macsec_fs, rx_rule);
+ macsec_fs_rx_del_rule(macsec_fs, rx_rule, macsec_ctx->secy->netdev, fs_id);
macsec_rule = NULL;
out_spec:
kvfree(spec);
@@ -1363,6 +1548,20 @@ struct mlx5_macsec_stats *mlx5_macsec_fs_get_stats(struct mlx5_macsec_fs *macsec
return &macsec_fs->stats;
}
+u32 mlx5_macsec_fs_get_fs_id_from_hashtable(struct mlx5_macsec_fs *macsec_fs, sci_t *sci)
+{
+ struct mlx5_fs_id *mlx5_fs_id;
+ u32 fs_id = 0;
+
+ rcu_read_lock();
+ mlx5_fs_id = rhashtable_lookup(&macsec_fs->sci_hash, sci, rhash_sci);
+ if (mlx5_fs_id)
+ fs_id = mlx5_fs_id->id;
+ rcu_read_unlock();
+
+ return fs_id;
+}
+
union mlx5_macsec_rule *
mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
const struct macsec_context *macsec_ctx,
@@ -1370,23 +1569,25 @@ mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
u32 *sa_fs_id)
{
return (attrs->action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
- macsec_fs_tx_add_rule(macsec_fs, macsec_ctx, attrs, sa_fs_id) :
- macsec_fs_rx_add_rule(macsec_fs, attrs, *sa_fs_id);
+ macsec_fs_tx_add_rule(macsec_fs, macsec_ctx, attrs) :
+ macsec_fs_rx_add_rule(macsec_fs, macsec_ctx, attrs, *sa_fs_id);
}
void mlx5_macsec_fs_del_rule(struct mlx5_macsec_fs *macsec_fs,
union mlx5_macsec_rule *macsec_rule,
- int action)
+ int action, void *macdev, u32 sa_fs_id)
{
(action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
- macsec_fs_tx_del_rule(macsec_fs, &macsec_rule->tx_rule) :
- macsec_fs_rx_del_rule(macsec_fs, &macsec_rule->rx_rule);
+ macsec_fs_tx_del_rule(macsec_fs, &macsec_rule->tx_rule, macdev) :
+ macsec_fs_rx_del_rule(macsec_fs, &macsec_rule->rx_rule, macdev, sa_fs_id);
}
void mlx5_macsec_fs_cleanup(struct mlx5_macsec_fs *macsec_fs)
{
macsec_fs_rx_cleanup(macsec_fs);
macsec_fs_tx_cleanup(macsec_fs);
+ rhashtable_destroy(&macsec_fs->fs_id_hash);
+ rhashtable_destroy(&macsec_fs->sci_hash);
kfree(macsec_fs);
}
@@ -1402,10 +1603,24 @@ mlx5_macsec_fs_init(struct mlx5_core_dev *mdev)
macsec_fs->mdev = mdev;
+ err = rhashtable_init(&macsec_fs->sci_hash, &rhash_sci);
+ if (err) {
+ mlx5_core_err(mdev, "MACsec offload: Failed to init SCI hash table, err=%d\n",
+ err);
+ goto err_hash;
+ }
+
+ err = rhashtable_init(&macsec_fs->fs_id_hash, &rhash_fs_id);
+ if (err) {
+ mlx5_core_err(mdev, "MACsec offload: Failed to init FS_ID hash table, err=%d\n",
+ err);
+ goto sci_hash_cleanup;
+ }
+
err = macsec_fs_tx_init(macsec_fs);
if (err) {
mlx5_core_err(mdev, "MACsec offload: Failed to init tx_fs, err=%d\n", err);
- goto err;
+ goto fs_id_hash_cleanup;
}
err = macsec_fs_rx_init(macsec_fs);
@@ -1418,7 +1633,11 @@ mlx5_macsec_fs_init(struct mlx5_core_dev *mdev)
tx_cleanup:
macsec_fs_tx_cleanup(macsec_fs);
-err:
+fs_id_hash_cleanup:
+ rhashtable_destroy(&macsec_fs->fs_id_hash);
+sci_hash_cleanup:
+ rhashtable_destroy(&macsec_fs->sci_hash);
+err_hash:
kfree(macsec_fs);
return NULL;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
index f007c33369c2..34b80c3ef6a5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.h
@@ -53,10 +53,11 @@ mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
void mlx5_macsec_fs_del_rule(struct mlx5_macsec_fs *macsec_fs,
union mlx5_macsec_rule *macsec_rule,
- int action);
+ int action, void *macdev, u32 sa_fs_id);
void mlx5_macsec_fs_get_stats_fill(struct mlx5_macsec_fs *macsec_fs, void *macsec_stats);
struct mlx5_macsec_stats *mlx5_macsec_fs_get_stats(struct mlx5_macsec_fs *macsec_fs);
+u32 mlx5_macsec_fs_get_fs_id_from_hashtable(struct mlx5_macsec_fs *macsec_fs, sci_t *sci);
#endif
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 08/14] RDMA/mlx5: Implement MACsec gid addition and deletion
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (6 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 07/14] net/mlx5: Maintain fs_id xarray per MACsec device inside macsec steering Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 09/14] net/mlx5: Add MACsec priorities in RDMA namespaces Leon Romanovsky
` (6 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Handle MACsec IP ambiguity issue, since mlx5 hw can't support
programming both the MACsec and the physical gid when they have the same
IP address, because it wouldn't know to whom to steer the traffic.
Hence in such case we delete the physical gid from the hw gid table,
which would then cause all traffic sent over it to fail, and we'll only
be able to send traffic over the MACsec gid.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Raed Salem <raeds@nvidia.com>
Reviewed-by: Mark Zhang <markzhang@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
drivers/infiniband/hw/mlx5/Makefile | 1 +
drivers/infiniband/hw/mlx5/macsec.c | 155 ++++++++++++++++++
drivers/infiniband/hw/mlx5/macsec.h | 25 +++
drivers/infiniband/hw/mlx5/main.c | 37 ++++-
drivers/infiniband/hw/mlx5/mlx5_ib.h | 7 +
.../mellanox/mlx5/core/en_accel/macsec.c | 31 ----
.../mellanox/mlx5/core/en_accel/macsec.h | 2 -
include/linux/mlx5/driver.h | 44 +++++
8 files changed, 260 insertions(+), 42 deletions(-)
create mode 100644 drivers/infiniband/hw/mlx5/macsec.c
create mode 100644 drivers/infiniband/hw/mlx5/macsec.h
diff --git a/drivers/infiniband/hw/mlx5/Makefile b/drivers/infiniband/hw/mlx5/Makefile
index 612ee8190a2d..72a526236c2e 100644
--- a/drivers/infiniband/hw/mlx5/Makefile
+++ b/drivers/infiniband/hw/mlx5/Makefile
@@ -28,3 +28,4 @@ mlx5_ib-$(CONFIG_INFINIBAND_USER_ACCESS) += devx.o \
fs.o \
qos.o \
std_types.o
+mlx5_ib-$(CONFIG_MLX5_MACSEC) += macsec.o
diff --git a/drivers/infiniband/hw/mlx5/macsec.c b/drivers/infiniband/hw/mlx5/macsec.c
new file mode 100644
index 000000000000..349ad13af75d
--- /dev/null
+++ b/drivers/infiniband/hw/mlx5/macsec.c
@@ -0,0 +1,155 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. */
+
+#include "macsec.h"
+
+struct mlx5_reserved_gids {
+ int macsec_index;
+ const struct ib_gid_attr *physical_gid;
+};
+
+int mlx5r_macsec_alloc_gids(struct mlx5_ib_dev *dev)
+{
+ int i, j, max_gids;
+
+ if (!mlx5_is_macsec_roce_supported(dev->mdev)) {
+ mlx5_ib_dbg(dev, "RoCE MACsec not supported due to capabilities\n");
+ return 0;
+ }
+
+ max_gids = MLX5_CAP_ROCE(dev->mdev, roce_address_table_size);
+ for (i = 0; i < dev->num_ports; i++) {
+ dev->port[i].reserved_gids = kcalloc(max_gids,
+ sizeof(*dev->port[i].reserved_gids),
+ GFP_KERNEL);
+ if (!dev->port[i].reserved_gids)
+ goto err;
+
+ for (j = 0; j < max_gids; j++)
+ dev->port[i].reserved_gids[j].macsec_index = -1;
+ }
+
+ return 0;
+err:
+ while (i >= 0) {
+ kfree(dev->port[i].reserved_gids);
+ i--;
+ }
+ return -ENOMEM;
+}
+
+void mlx5r_macsec_dealloc_gids(struct mlx5_ib_dev *dev)
+{
+ int i;
+
+ if (!mlx5_is_macsec_roce_supported(dev->mdev))
+ mlx5_ib_dbg(dev, "RoCE MACsec not supported due to capabilities\n");
+
+ for (i = 0; i < dev->num_ports; i++)
+ kfree(dev->port[i].reserved_gids);
+}
+
+int mlx5r_add_gid_macsec_operations(const struct ib_gid_attr *attr)
+{
+ struct mlx5_ib_dev *dev = to_mdev(attr->device);
+ const struct ib_gid_attr *physical_gid;
+ struct mlx5_reserved_gids *mgids;
+ struct net_device *ndev;
+ int ret = 0;
+
+ if (attr->gid_type != IB_GID_TYPE_ROCE_UDP_ENCAP)
+ return 0;
+
+ if (!mlx5_is_macsec_roce_supported(dev->mdev)) {
+ mlx5_ib_dbg(dev, "RoCE MACsec not supported due to capabilities\n");
+ return 0;
+ }
+
+ rcu_read_lock();
+ ndev = rcu_dereference(attr->ndev);
+ if (!ndev) {
+ rcu_read_unlock();
+ return -ENODEV;
+ }
+
+ if (!netif_is_macsec(ndev) || !macsec_netdev_is_offloaded(ndev)) {
+ rcu_read_unlock();
+ return 0;
+ }
+ rcu_read_unlock();
+
+ physical_gid = rdma_find_gid(attr->device, &attr->gid,
+ attr->gid_type, NULL);
+ if (IS_ERR(physical_gid))
+ return 0;
+
+ ret = set_roce_addr(to_mdev(physical_gid->device),
+ physical_gid->port_num,
+ physical_gid->index, NULL,
+ physical_gid);
+ if (ret)
+ goto gid_err;
+
+ mgids = &dev->port[attr->port_num - 1].reserved_gids[physical_gid->index];
+ mgids->macsec_index = attr->index;
+ mgids->physical_gid = physical_gid;
+
+ return 0;
+
+gid_err:
+ rdma_put_gid_attr(physical_gid);
+ return ret;
+}
+
+void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr)
+{
+ struct mlx5_ib_dev *dev = to_mdev(attr->device);
+ struct mlx5_reserved_gids *mgids;
+ struct net_device *ndev;
+ int i, max_gids;
+
+ if (attr->gid_type != IB_GID_TYPE_ROCE_UDP_ENCAP)
+ return;
+
+ if (!mlx5_is_macsec_roce_supported(dev->mdev)) {
+ mlx5_ib_dbg(dev, "RoCE MACsec not supported due to capabilities\n");
+ return;
+ }
+
+ mgids = &dev->port[attr->port_num - 1].reserved_gids[attr->index];
+ if (mgids->macsec_index != -1) { /* Checking if physical gid has ambiguous IP */
+ rdma_put_gid_attr(mgids->physical_gid);
+ mgids->macsec_index = -1;
+ return;
+ }
+
+ rcu_read_lock();
+ ndev = rcu_dereference(attr->ndev);
+ if (!ndev) {
+ rcu_read_unlock();
+ return;
+ }
+
+ if (!netif_is_macsec(ndev) || !macsec_netdev_is_offloaded(ndev)) {
+ rcu_read_unlock();
+ return;
+ }
+ rcu_read_unlock();
+
+ max_gids = MLX5_CAP_ROCE(dev->mdev, roce_address_table_size);
+ for (i = 0; i < max_gids; i++) { /* Checking if macsec gid has ambiguous IP */
+ mgids = &dev->port[attr->port_num - 1].reserved_gids[i];
+ if (mgids->macsec_index == attr->index) {
+ const struct ib_gid_attr *physical_gid = mgids->physical_gid;
+
+ set_roce_addr(to_mdev(physical_gid->device),
+ physical_gid->port_num,
+ physical_gid->index,
+ &physical_gid->gid, physical_gid);
+
+ rdma_put_gid_attr(physical_gid);
+ mgids->macsec_index = -1;
+ break;
+ }
+ }
+}
diff --git a/drivers/infiniband/hw/mlx5/macsec.h b/drivers/infiniband/hw/mlx5/macsec.h
new file mode 100644
index 000000000000..b60f8f046d6f
--- /dev/null
+++ b/drivers/infiniband/hw/mlx5/macsec.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. */
+
+#ifndef __MLX5_MACSEC_H__
+#define __MLX5_MACSEC_H__
+
+#include <net/macsec.h>
+#include <rdma/ib_cache.h>
+#include <rdma/ib_addr.h>
+#include "mlx5_ib.h"
+
+#ifdef CONFIG_MLX5_MACSEC
+struct mlx5_reserved_gids;
+
+int mlx5r_add_gid_macsec_operations(const struct ib_gid_attr *attr);
+void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr);
+int mlx5r_macsec_alloc_gids(struct mlx5_ib_dev *dev);
+void mlx5r_macsec_dealloc_gids(struct mlx5_ib_dev *dev);
+#else
+static inline int mlx5r_add_gid_macsec_operations(const struct ib_gid_attr *attr) { return 0; }
+static inline void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr) {}
+static inline int mlx5r_macsec_alloc_gids(struct mlx5_ib_dev *dev) { return 0; }
+static inline void mlx5r_macsec_dealloc_gids(struct mlx5_ib_dev *dev) {}
+#endif
+#endif /* __MLX5_MACSEC_H__ */
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index f0b394ed7452..f463cf8b7501 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -46,6 +46,7 @@
#include <rdma/uverbs_ioctl.h>
#include <rdma/mlx5_user_ioctl_verbs.h>
#include <rdma/mlx5_user_ioctl_cmds.h>
+#include "macsec.h"
#define UVERBS_MODULE_NAME mlx5_ib
#include <rdma/uverbs_named_ioctl.h>
@@ -564,9 +565,9 @@ static int mlx5_query_port_roce(struct ib_device *device, u32 port_num,
return err;
}
-static int set_roce_addr(struct mlx5_ib_dev *dev, u32 port_num,
- unsigned int index, const union ib_gid *gid,
- const struct ib_gid_attr *attr)
+int set_roce_addr(struct mlx5_ib_dev *dev, u32 port_num,
+ unsigned int index, const union ib_gid *gid,
+ const struct ib_gid_attr *attr)
{
enum ib_gid_type gid_type;
u16 vlan_id = 0xffff;
@@ -607,6 +608,12 @@ static int set_roce_addr(struct mlx5_ib_dev *dev, u32 port_num,
static int mlx5_ib_add_gid(const struct ib_gid_attr *attr,
__always_unused void **context)
{
+ int ret;
+
+ ret = mlx5r_add_gid_macsec_operations(attr);
+ if (ret)
+ return ret;
+
return set_roce_addr(to_mdev(attr->device), attr->port_num,
attr->index, &attr->gid, attr);
}
@@ -614,8 +621,15 @@ static int mlx5_ib_add_gid(const struct ib_gid_attr *attr,
static int mlx5_ib_del_gid(const struct ib_gid_attr *attr,
__always_unused void **context)
{
- return set_roce_addr(to_mdev(attr->device), attr->port_num,
- attr->index, NULL, attr);
+ int ret;
+
+ ret = set_roce_addr(to_mdev(attr->device), attr->port_num,
+ attr->index, NULL, attr);
+ if (ret)
+ return ret;
+
+ mlx5r_del_gid_macsec_operations(attr);
+ return 0;
}
__be16 mlx5_get_roce_udp_sport_min(const struct mlx5_ib_dev *dev,
@@ -3644,13 +3658,13 @@ static void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev)
mutex_destroy(&dev->cap_mask_mutex);
WARN_ON(!xa_empty(&dev->sig_mrs));
WARN_ON(!bitmap_empty(dev->dm.memic_alloc_pages, MLX5_MAX_MEMIC_PAGES));
+ mlx5r_macsec_dealloc_gids(dev);
}
static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
{
struct mlx5_core_dev *mdev = dev->mdev;
- int err;
- int i;
+ int err, i;
dev->ib_dev.node_type = RDMA_NODE_IB_CA;
dev->ib_dev.local_dma_lkey = 0 /* not supported for now */;
@@ -3670,10 +3684,14 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
if (err)
return err;
- err = mlx5_ib_init_multiport_master(dev);
+ err = mlx5r_macsec_alloc_gids(dev);
if (err)
return err;
+ err = mlx5_ib_init_multiport_master(dev);
+ if (err)
+ goto err;
+
err = set_has_smi_cap(dev);
if (err)
goto err_mp;
@@ -3697,7 +3715,8 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
spin_lock_init(&dev->dm.lock);
dev->dm.dev = mdev;
return 0;
-
+err:
+ mlx5r_macsec_dealloc_gids(dev);
err_mp:
mlx5_ib_cleanup_multiport_master(dev);
return err;
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index 9c33d960af3c..a4b940b5035f 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -26,6 +26,7 @@
#include "srq.h"
#include "qp.h"
+#include "macsec.h"
#define mlx5_ib_dbg(_dev, format, arg...) \
dev_dbg(&(_dev)->ib_dev.dev, "%s:%d:(pid %d): " format, __func__, \
@@ -870,6 +871,9 @@ struct mlx5_ib_port {
struct mlx5_ib_dbg_cc_params *dbg_cc_params;
struct mlx5_roce roce;
struct mlx5_eswitch_rep *rep;
+#ifdef CONFIG_MLX5_MACSEC
+ struct mlx5_reserved_gids *reserved_gids;
+#endif
};
struct mlx5_ib_dbg_param {
@@ -1648,4 +1652,7 @@ static inline bool mlx5_umem_needs_ats(struct mlx5_ib_dev *dev,
return access_flags & IB_ACCESS_RELAXED_ORDERING;
}
+int set_roce_addr(struct mlx5_ib_dev *dev, u32 port_num,
+ unsigned int index, const union ib_gid *gid,
+ const struct ib_gid_attr *attr);
#endif /* MLX5_IB_H */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
index b4f3f4f10af3..c9c1db971652 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.c
@@ -1607,37 +1607,6 @@ static void mlx5e_macsec_aso_cleanup(struct mlx5e_macsec_aso *aso, struct mlx5_c
mlx5_core_dealloc_pd(mdev, aso->pdn);
}
-bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev)
-{
- if (!(MLX5_CAP_GEN_64(mdev, general_obj_types) &
- MLX5_GENERAL_OBJ_TYPES_CAP_MACSEC_OFFLOAD))
- return false;
-
- if (!MLX5_CAP_GEN(mdev, log_max_dek))
- return false;
-
- if (!MLX5_CAP_MACSEC(mdev, log_max_macsec_offload))
- return false;
-
- if (!MLX5_CAP_FLOWTABLE_NIC_RX(mdev, macsec_decrypt) ||
- !MLX5_CAP_FLOWTABLE_NIC_RX(mdev, reformat_remove_macsec))
- return false;
-
- if (!MLX5_CAP_FLOWTABLE_NIC_TX(mdev, macsec_encrypt) ||
- !MLX5_CAP_FLOWTABLE_NIC_TX(mdev, reformat_add_macsec))
- return false;
-
- if (!MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_128_encrypt) &&
- !MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_256_encrypt))
- return false;
-
- if (!MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_128_decrypt) &&
- !MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_256_decrypt))
- return false;
-
- return true;
-}
-
static const struct macsec_ops macsec_offload_ops = {
.mdo_add_txsa = mlx5e_macsec_add_txsa,
.mdo_upd_txsa = mlx5e_macsec_upd_txsa,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
index 2ecd769585f4..27df72e23106 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/macsec.h
@@ -36,7 +36,6 @@ static inline bool mlx5e_macsec_is_rx_flow(struct mlx5_cqe64 *cqe)
void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev, struct sk_buff *skb,
struct mlx5_cqe64 *cqe);
-bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev);
#else
@@ -49,7 +48,6 @@ static inline void mlx5e_macsec_offload_handle_rx_skb(struct net_device *netdev,
struct sk_buff *skb,
struct mlx5_cqe64 *cqe)
{}
-static inline bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev) { return false; }
#endif /* CONFIG_MLX5_MACSEC */
#endif /* __MLX5_ACCEL_EN_MACSEC_H__ */
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 541c292373e6..c1b88a1112cc 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -1326,6 +1326,50 @@ static inline bool mlx5_get_roce_state(struct mlx5_core_dev *dev)
return mlx5_is_roce_on(dev);
}
+static inline bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev)
+{
+ if (!(MLX5_CAP_GEN_64(mdev, general_obj_types) &
+ MLX5_GENERAL_OBJ_TYPES_CAP_MACSEC_OFFLOAD))
+ return false;
+
+ if (!MLX5_CAP_GEN(mdev, log_max_dek))
+ return false;
+
+ if (!MLX5_CAP_MACSEC(mdev, log_max_macsec_offload))
+ return false;
+
+ if (!MLX5_CAP_FLOWTABLE_NIC_RX(mdev, macsec_decrypt) ||
+ !MLX5_CAP_FLOWTABLE_NIC_RX(mdev, reformat_remove_macsec))
+ return false;
+
+ if (!MLX5_CAP_FLOWTABLE_NIC_TX(mdev, macsec_encrypt) ||
+ !MLX5_CAP_FLOWTABLE_NIC_TX(mdev, reformat_add_macsec))
+ return false;
+
+ if (!MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_128_encrypt) &&
+ !MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_256_encrypt))
+ return false;
+
+ if (!MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_128_decrypt) &&
+ !MLX5_CAP_MACSEC(mdev, macsec_crypto_esp_aes_gcm_256_decrypt))
+ return false;
+
+ return true;
+}
+
+#define NIC_RDMA_BOTH_DIRS_CAPS (MLX5_FT_NIC_RX_2_NIC_RX_RDMA | MLX5_FT_NIC_TX_RDMA_2_NIC_TX)
+
+static inline bool mlx5_is_macsec_roce_supported(struct mlx5_core_dev *mdev)
+{
+ if (((MLX5_CAP_GEN_2(mdev, flow_table_type_2_type) &
+ NIC_RDMA_BOTH_DIRS_CAPS) != NIC_RDMA_BOTH_DIRS_CAPS) ||
+ !MLX5_CAP_FLOWTABLE_RDMA_TX(mdev, max_modify_header_actions) ||
+ !mlx5e_is_macsec_device(mdev))
+ return false;
+
+ return true;
+}
+
enum {
MLX5_OCTWORD = 16,
};
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 09/14] net/mlx5: Add MACsec priorities in RDMA namespaces
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (7 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 08/14] RDMA/mlx5: Implement MACsec gid addition and deletion Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 10/14] IB/core: Reorder GID delete code for RoCE Leon Romanovsky
` (5 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Add MACsec flow steering priorities in RDMA namespaces. This allows
adding tables/rules to forward RoCEv2 traffic to the MACsec crypto
tables in NIC_TX domain, and accept RoCEv2 traffic from NIC_RX domain.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/fs_core.c | 35 +++++++++++++++++--
include/linux/mlx5/fs.h | 2 ++
2 files changed, 34 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index a3228502f866..815fe6393c4b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -224,6 +224,7 @@ static struct init_tree_node egress_root_fs = {
enum {
RDMA_RX_IPSEC_PRIO,
+ RDMA_RX_MACSEC_PRIO,
RDMA_RX_COUNTERS_PRIO,
RDMA_RX_BYPASS_PRIO,
RDMA_RX_KERNEL_PRIO,
@@ -237,9 +238,13 @@ enum {
#define RDMA_RX_KERNEL_MIN_LEVEL (RDMA_RX_BYPASS_MIN_LEVEL + 1)
#define RDMA_RX_COUNTERS_MIN_LEVEL (RDMA_RX_KERNEL_MIN_LEVEL + 2)
+#define RDMA_RX_MACSEC_NUM_PRIOS 1
+#define RDMA_RX_MACSEC_PRIO_NUM_LEVELS 2
+#define RDMA_RX_MACSEC_MIN_LEVEL (RDMA_RX_COUNTERS_MIN_LEVEL + RDMA_RX_MACSEC_NUM_PRIOS)
+
static struct init_tree_node rdma_rx_root_fs = {
.type = FS_TYPE_NAMESPACE,
- .ar_size = 4,
+ .ar_size = 5,
.children = (struct init_tree_node[]) {
[RDMA_RX_IPSEC_PRIO] =
ADD_PRIO(0, RDMA_RX_IPSEC_MIN_LEVEL, 0,
@@ -247,6 +252,12 @@ static struct init_tree_node rdma_rx_root_fs = {
ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
ADD_MULTIPLE_PRIO(RDMA_RX_IPSEC_NUM_PRIOS,
RDMA_RX_IPSEC_NUM_LEVELS))),
+ [RDMA_RX_MACSEC_PRIO] =
+ ADD_PRIO(0, RDMA_RX_MACSEC_MIN_LEVEL, 0,
+ FS_CHAINING_CAPS,
+ ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+ ADD_MULTIPLE_PRIO(RDMA_RX_MACSEC_NUM_PRIOS,
+ RDMA_RX_MACSEC_PRIO_NUM_LEVELS))),
[RDMA_RX_COUNTERS_PRIO] =
ADD_PRIO(0, RDMA_RX_COUNTERS_MIN_LEVEL, 0,
FS_CHAINING_CAPS,
@@ -270,6 +281,7 @@ static struct init_tree_node rdma_rx_root_fs = {
enum {
RDMA_TX_COUNTERS_PRIO,
RDMA_TX_IPSEC_PRIO,
+ RDMA_TX_MACSEC_PRIO,
RDMA_TX_BYPASS_PRIO,
};
@@ -280,9 +292,13 @@ enum {
#define RDMA_TX_IPSEC_PRIO_NUM_LEVELS 1
#define RDMA_TX_IPSEC_MIN_LEVEL (RDMA_TX_COUNTERS_MIN_LEVEL + RDMA_TX_IPSEC_NUM_PRIOS)
+#define RDMA_TX_MACSEC_NUM_PRIOS 1
+#define RDMA_TX_MACESC_PRIO_NUM_LEVELS 1
+#define RDMA_TX_MACSEC_MIN_LEVEL (RDMA_TX_COUNTERS_MIN_LEVEL + RDMA_TX_MACSEC_NUM_PRIOS)
+
static struct init_tree_node rdma_tx_root_fs = {
.type = FS_TYPE_NAMESPACE,
- .ar_size = 3,
+ .ar_size = 4,
.children = (struct init_tree_node[]) {
[RDMA_TX_COUNTERS_PRIO] =
ADD_PRIO(0, RDMA_TX_COUNTERS_MIN_LEVEL, 0,
@@ -296,7 +312,12 @@ static struct init_tree_node rdma_tx_root_fs = {
ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
ADD_MULTIPLE_PRIO(RDMA_TX_IPSEC_NUM_PRIOS,
RDMA_TX_IPSEC_PRIO_NUM_LEVELS))),
-
+ [RDMA_TX_MACSEC_PRIO] =
+ ADD_PRIO(0, RDMA_TX_MACSEC_MIN_LEVEL, 0,
+ FS_CHAINING_CAPS,
+ ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
+ ADD_MULTIPLE_PRIO(RDMA_TX_MACSEC_NUM_PRIOS,
+ RDMA_TX_MACESC_PRIO_NUM_LEVELS))),
[RDMA_TX_BYPASS_PRIO] =
ADD_PRIO(0, RDMA_TX_BYPASS_MIN_LEVEL, 0,
FS_CHAINING_CAPS_RDMA_TX,
@@ -2466,6 +2487,14 @@ struct mlx5_flow_namespace *mlx5_get_flow_namespace(struct mlx5_core_dev *dev,
root_ns = steering->rdma_tx_root_ns;
prio = RDMA_TX_IPSEC_PRIO;
break;
+ case MLX5_FLOW_NAMESPACE_RDMA_RX_MACSEC:
+ root_ns = steering->rdma_rx_root_ns;
+ prio = RDMA_RX_MACSEC_PRIO;
+ break;
+ case MLX5_FLOW_NAMESPACE_RDMA_TX_MACSEC:
+ root_ns = steering->rdma_tx_root_ns;
+ prio = RDMA_TX_MACSEC_PRIO;
+ break;
default: /* Must be NIC RX */
WARN_ON(!is_nic_rx_ns(type));
root_ns = steering->root_ns;
diff --git a/include/linux/mlx5/fs.h b/include/linux/mlx5/fs.h
index c302ec34255b..1e00c2436377 100644
--- a/include/linux/mlx5/fs.h
+++ b/include/linux/mlx5/fs.h
@@ -105,6 +105,8 @@ enum mlx5_flow_namespace_type {
MLX5_FLOW_NAMESPACE_RDMA_TX_COUNTERS,
MLX5_FLOW_NAMESPACE_RDMA_RX_IPSEC,
MLX5_FLOW_NAMESPACE_RDMA_TX_IPSEC,
+ MLX5_FLOW_NAMESPACE_RDMA_RX_MACSEC,
+ MLX5_FLOW_NAMESPACE_RDMA_TX_MACSEC,
};
enum {
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 10/14] IB/core: Reorder GID delete code for RoCE
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (8 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 09/14] net/mlx5: Add MACsec priorities in RDMA namespaces Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 11/14] net/mlx5: Configure MACsec steering for egress RoCEv2 traffic Leon Romanovsky
` (4 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Reorder GID delete code so that the driver del_gid operation is executed
before nullifying the gid attribute ndev parameter, this allows drivers
to access the ndev during their gid delete operation, which makes more
sense since they had access to it during the gid addition operation.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
drivers/infiniband/core/cache.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c
index 2e91d8879326..73f913cbd146 100644
--- a/drivers/infiniband/core/cache.c
+++ b/drivers/infiniband/core/cache.c
@@ -400,6 +400,9 @@ static void del_gid(struct ib_device *ib_dev, u32 port,
table->data_vec[ix] = NULL;
write_unlock_irq(&table->rwlock);
+ if (rdma_cap_roce_gid_table(ib_dev, port))
+ ib_dev->ops.del_gid(&entry->attr, &entry->context);
+
ndev_storage = entry->ndev_storage;
if (ndev_storage) {
entry->ndev_storage = NULL;
@@ -407,9 +410,6 @@ static void del_gid(struct ib_device *ib_dev, u32 port,
call_rcu(&ndev_storage->rcu_head, put_gid_ndev);
}
- if (rdma_cap_roce_gid_table(ib_dev, port))
- ib_dev->ops.del_gid(&entry->attr, &entry->context);
-
put_gid_entry_locked(entry);
}
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 11/14] net/mlx5: Configure MACsec steering for egress RoCEv2 traffic
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (9 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 10/14] IB/core: Reorder GID delete code for RoCE Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 12/14] net/mlx5: Configure MACsec steering for ingress " Leon Romanovsky
` (3 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Add steering table in RDMA_TX domain, to forward MACsec traffic
to MACsec crypto table in NIC domain.
The tables are created in a lazy manner when the first TX SA is
being created, and destroyed upon the destruction of the last SA.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../mellanox/mlx5/core/lib/macsec_fs.c | 46 ++++++++++++++++++-
1 file changed, 45 insertions(+), 1 deletion(-)
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 d39ca7c66542..15e7ea3ed79f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -95,6 +95,8 @@ struct mlx5_macsec_tx {
struct ida tx_halloc;
struct mlx5_macsec_tables tables;
+
+ struct mlx5_flow_table *ft_rdma_tx;
};
struct mlx5_macsec_rx_rule {
@@ -173,6 +175,9 @@ static void macsec_fs_tx_destroy(struct mlx5_macsec_fs *macsec_fs)
struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
struct mlx5_macsec_tables *tx_tables;
+ if (mlx5_is_macsec_roce_supported(macsec_fs->mdev))
+ mlx5_destroy_flow_table(tx_fs->ft_rdma_tx);
+
tx_tables = &tx_fs->tables;
/* Tx check table */
@@ -301,6 +306,39 @@ static struct mlx5_flow_table
return fdb;
}
+enum {
+ RDMA_TX_MACSEC_LEVEL = 0,
+};
+
+static int macsec_fs_tx_roce_create(struct mlx5_macsec_fs *macsec_fs)
+{
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
+ struct mlx5_flow_namespace *ns;
+ struct mlx5_flow_table *ft;
+ int err;
+
+ if (!mlx5_is_macsec_roce_supported(mdev)) {
+ mlx5_core_dbg(mdev, "Failed to init RoCE MACsec, capabilities not supported\n");
+ return 0;
+ }
+
+ ns = mlx5_get_flow_namespace(mdev, MLX5_FLOW_NAMESPACE_RDMA_TX_MACSEC);
+ if (!ns)
+ return -ENOMEM;
+
+ /* Tx RoCE crypto table */
+ ft = macsec_fs_auto_group_table_create(ns, 0, RDMA_TX_MACSEC_LEVEL, CRYPTO_NUM_MAXSEC_FTE);
+ if (IS_ERR(ft)) {
+ err = PTR_ERR(ft);
+ mlx5_core_err(mdev, "Failed to create MACsec RoCE Tx crypto table err(%d)\n", err);
+ return err;
+ }
+ tx_fs->ft_rdma_tx = ft;
+
+ return 0;
+}
+
static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
@@ -443,7 +481,13 @@ static int macsec_fs_tx_create(struct mlx5_macsec_fs *macsec_fs)
}
tx_fs->check_rule = rule;
- goto out_flow_group;
+ err = macsec_fs_tx_roce_create(macsec_fs);
+ if (err)
+ goto err;
+
+ kvfree(flow_group_in);
+ kvfree(spec);
+ return 0;
err:
macsec_fs_tx_destroy(macsec_fs);
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 12/14] net/mlx5: Configure MACsec steering for ingress RoCEv2 traffic
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (10 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 11/14] net/mlx5: Configure MACsec steering for egress RoCEv2 traffic Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 13/14] net/mlx5: Add RoCE MACsec steering infrastructure in core Leon Romanovsky
` (2 subsequent siblings)
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Add steering tables/rules to check if the decrypted traffic is RoCEv2,
if so copy reg_b_metadata to a temp reg and forward it to RDMA_RX domain.
The rules are added once the MACsec device is assigned an IP address
where we verify that the packet ip is for MACsec device and that the temp
reg has MACsec operation and a valid SCI inside.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/fs_core.c | 2 +-
.../mellanox/mlx5/core/lib/macsec_fs.c | 359 +++++++++++++++++-
2 files changed, 352 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index 815fe6393c4b..a13b9c2bd144 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -107,7 +107,7 @@
LEFTOVERS_NUM_PRIOS)
#define KERNEL_RX_MACSEC_NUM_PRIOS 1
-#define KERNEL_RX_MACSEC_NUM_LEVELS 2
+#define KERNEL_RX_MACSEC_NUM_LEVELS 3
#define KERNEL_RX_MACSEC_MIN_LEVEL (BY_PASS_MIN_LEVEL + KERNEL_RX_MACSEC_NUM_PRIOS)
#define ETHTOOL_PRIO_NUM_LEVELS 1
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 15e7ea3ed79f..be909b613288 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -22,7 +22,9 @@
#define TX_CHECK_TABLE_NUM_FTE 2
#define RX_CRYPTO_TABLE_LEVEL 0
#define RX_CHECK_TABLE_LEVEL 1
+#define RX_ROCE_TABLE_LEVEL 2
#define RX_CHECK_TABLE_NUM_FTE 3
+#define RX_ROCE_TABLE_NUM_FTE 2
#define RX_CRYPTO_TABLE_NUM_GROUPS 3
#define RX_CRYPTO_TABLE_SA_RULE_WITH_SCI_GROUP_SIZE \
((CRYPTO_NUM_MAXSEC_FTE - CRYPTO_TABLE_DEFAULT_RULE_GROUP_SIZE) / 2)
@@ -30,6 +32,9 @@
(CRYPTO_NUM_MAXSEC_FTE - RX_CRYPTO_TABLE_SA_RULE_WITH_SCI_GROUP_SIZE)
#define RX_NUM_OF_RULES_PER_SA 2
+#define RDMA_RX_ROCE_IP_TABLE_LEVEL 0
+#define RDMA_RX_ROCE_MACSEC_OP_TABLE_LEVEL 1
+
#define MLX5_MACSEC_TAG_LEN 8 /* SecTAG length with ethertype and without the optional SCI */
#define MLX5_MACSEC_SECTAG_TCI_AN_FIELD_BITMASK 0x23
#define MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET 0x8
@@ -104,11 +109,31 @@ struct mlx5_macsec_rx_rule {
struct mlx5_modify_hdr *meta_modhdr;
};
+struct mlx5_macsec_miss {
+ struct mlx5_flow_group *g;
+ struct mlx5_flow_handle *rule;
+};
+
+struct mlx5_macsec_rx_roce {
+ /* Flow table/rules in NIC domain, to check if it's a RoCE packet */
+ struct mlx5_flow_group *g;
+ struct mlx5_flow_table *ft;
+ struct mlx5_flow_handle *rule;
+ struct mlx5_modify_hdr *copy_modify_hdr;
+ struct mlx5_macsec_miss nic_miss;
+
+ /* Flow table/rule in RDMA domain, to check dgid */
+ struct mlx5_flow_table *ft_ip_check;
+ struct mlx5_flow_table *ft_macsec_op_check;
+ struct mlx5_macsec_miss miss;
+};
+
struct mlx5_macsec_rx {
struct mlx5_flow_handle *check_rule[2];
struct mlx5_pkt_reformat *check_rule_pkt_reformat[2];
struct mlx5_macsec_tables tables;
+ struct mlx5_macsec_rx_roce roce;
};
union mlx5_macsec_rule {
@@ -933,6 +958,29 @@ static int macsec_fs_tx_init(struct mlx5_macsec_fs *macsec_fs)
return err;
}
+static void macsec_fs_rx_roce_miss_destroy(struct mlx5_macsec_miss *miss)
+{
+ mlx5_del_flow_rules(miss->rule);
+ mlx5_destroy_flow_group(miss->g);
+}
+
+static void macsec_fs_rdma_rx_destroy(struct mlx5_macsec_rx_roce *roce, struct mlx5_core_dev *mdev)
+{
+ if (!mlx5_is_macsec_roce_supported(mdev))
+ return;
+
+ mlx5_del_flow_rules(roce->nic_miss.rule);
+ mlx5_del_flow_rules(roce->rule);
+ mlx5_modify_header_dealloc(mdev, roce->copy_modify_hdr);
+ mlx5_destroy_flow_group(roce->nic_miss.g);
+ mlx5_destroy_flow_group(roce->g);
+ mlx5_destroy_flow_table(roce->ft);
+
+ macsec_fs_rx_roce_miss_destroy(&roce->miss);
+ mlx5_destroy_flow_table(roce->ft_macsec_op_check);
+ mlx5_destroy_flow_table(roce->ft_ip_check);
+}
+
static void macsec_fs_rx_destroy(struct mlx5_macsec_fs *macsec_fs)
{
struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
@@ -977,6 +1025,8 @@ static void macsec_fs_rx_destroy(struct mlx5_macsec_fs *macsec_fs)
}
macsec_fs_destroy_flow_table(&rx_tables->ft_crypto);
+
+ macsec_fs_rdma_rx_destroy(&macsec_fs->rx_fs->roce, macsec_fs->mdev);
}
static int macsec_fs_rx_create_crypto_table_groups(struct mlx5_macsec_flow_table *ft)
@@ -1072,9 +1122,10 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs
struct mlx5_pkt_reformat_params reformat_params = {};
struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
struct mlx5_core_dev *mdev = macsec_fs->mdev;
+ struct mlx5_flow_destination roce_dest[2];
struct mlx5_macsec_tables *rx_tables;
struct mlx5_flow_handle *rule;
- int err = 0;
+ int err = 0, dstn = 0;
rx_tables = &rx_fs->tables;
@@ -1115,12 +1166,22 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs
MLX5_MACSEC_SECTAG_TCI_AN_FIELD_OFFSET);
flow_act->flags = FLOW_ACT_NO_APPEND;
- flow_act->action = MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO |
- MLX5_FLOW_CONTEXT_ACTION_PACKET_REFORMAT |
- MLX5_FLOW_CONTEXT_ACTION_COUNT;
- dest->type = MLX5_FLOW_DESTINATION_TYPE_COUNTER;
- dest->counter_id = mlx5_fc_id(rx_tables->check_rule_counter);
- rule = mlx5_add_flow_rules(rx_tables->ft_check, spec, flow_act, dest, 1);
+
+ if (rx_fs->roce.ft) {
+ flow_act->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+ roce_dest[dstn].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+ roce_dest[dstn].ft = rx_fs->roce.ft;
+ dstn++;
+ } else {
+ flow_act->action = MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
+ }
+
+ 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);
+ rule = mlx5_add_flow_rules(rx_tables->ft_check, spec, flow_act, roce_dest, dstn + 1);
+
if (IS_ERR(rule)) {
err = PTR_ERR(rule);
mlx5_core_err(mdev, "Failed to add MACsec Rx check rule, err=%d\n", err);
@@ -1132,6 +1193,284 @@ static int macsec_fs_rx_create_check_decap_rule(struct mlx5_macsec_fs *macsec_fs
return 0;
}
+static int macsec_fs_rx_roce_miss_create(struct mlx5_core_dev *mdev,
+ struct mlx5_macsec_rx_roce *roce)
+{
+ struct mlx5_flow_act flow_act = {};
+ struct mlx5_flow_group *flow_group;
+ struct mlx5_flow_handle *rule;
+ u32 *flow_group_in;
+ int err;
+
+ flow_group_in = kvzalloc(MLX5_ST_SZ_BYTES(create_flow_group_in), GFP_KERNEL);
+ if (!flow_group_in)
+ return -ENOMEM;
+
+ /* IP check ft has no miss rule since we use default miss action which is go to next PRIO */
+ MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index,
+ roce->ft_macsec_op_check->max_fte - 1);
+ MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index,
+ roce->ft_macsec_op_check->max_fte - 1);
+ flow_group = mlx5_create_flow_group(roce->ft_macsec_op_check, flow_group_in);
+ if (IS_ERR(flow_group)) {
+ err = PTR_ERR(flow_group);
+ mlx5_core_err(mdev,
+ "Failed to create miss flow group for MACsec RoCE operation check table err(%d)\n",
+ err);
+ goto err_macsec_op_miss_group;
+ }
+ roce->miss.g = flow_group;
+
+ flow_act.action = MLX5_FLOW_CONTEXT_ACTION_DROP;
+ rule = mlx5_add_flow_rules(roce->ft_macsec_op_check, NULL, &flow_act, NULL, 0);
+ if (IS_ERR(rule)) {
+ err = PTR_ERR(rule);
+ mlx5_core_err(mdev, "Failed to add miss rule to MACsec RoCE operation check table err(%d)\n",
+ err);
+ goto err_macsec_op_rule;
+ }
+ roce->miss.rule = rule;
+
+ kvfree(flow_group_in);
+ return 0;
+
+err_macsec_op_rule:
+ mlx5_destroy_flow_group(roce->miss.g);
+err_macsec_op_miss_group:
+ kvfree(flow_group_in);
+ return err;
+}
+
+#define MLX5_RX_ROCE_GROUP_SIZE BIT(0)
+
+static int macsec_fs_rx_roce_jump_to_rdma_groups_create(struct mlx5_core_dev *mdev,
+ struct mlx5_macsec_rx_roce *roce)
+{
+ struct mlx5_flow_group *g;
+ void *outer_headers_c;
+ int ix = 0;
+ u32 *in;
+ int err;
+ u8 *mc;
+
+ in = kvzalloc(MLX5_ST_SZ_BYTES(create_flow_group_in), GFP_KERNEL);
+ if (!in)
+ return -ENOMEM;
+
+ mc = MLX5_ADDR_OF(create_flow_group_in, in, match_criteria);
+ outer_headers_c = MLX5_ADDR_OF(fte_match_param, mc, outer_headers);
+ MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, ip_protocol);
+ MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, outer_headers_c, udp_dport);
+
+ MLX5_SET_CFG(in, match_criteria_enable, MLX5_MATCH_OUTER_HEADERS);
+ MLX5_SET_CFG(in, start_flow_index, ix);
+ ix += MLX5_RX_ROCE_GROUP_SIZE;
+ MLX5_SET_CFG(in, end_flow_index, ix - 1);
+ g = mlx5_create_flow_group(roce->ft, in);
+ if (IS_ERR(g)) {
+ err = PTR_ERR(g);
+ mlx5_core_err(mdev, "Failed to create main flow group for MACsec RoCE NIC UDP table err(%d)\n",
+ err);
+ goto err_udp_group;
+ }
+ roce->g = g;
+
+ memset(in, 0, MLX5_ST_SZ_BYTES(create_flow_group_in));
+ MLX5_SET_CFG(in, start_flow_index, ix);
+ ix += MLX5_RX_ROCE_GROUP_SIZE;
+ MLX5_SET_CFG(in, end_flow_index, ix - 1);
+ g = mlx5_create_flow_group(roce->ft, in);
+ if (IS_ERR(g)) {
+ err = PTR_ERR(g);
+ mlx5_core_err(mdev, "Failed to create miss flow group for MACsec RoCE NIC UDP table err(%d)\n",
+ err);
+ goto err_udp_miss_group;
+ }
+ roce->nic_miss.g = g;
+
+ kvfree(in);
+ return 0;
+
+err_udp_miss_group:
+ mlx5_destroy_flow_group(roce->g);
+err_udp_group:
+ kvfree(in);
+ return err;
+}
+
+static int macsec_fs_rx_roce_jump_to_rdma_rules_create(struct mlx5_macsec_fs *macsec_fs,
+ struct mlx5_macsec_rx_roce *roce)
+{
+ u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
+ struct mlx5_flow_destination dst = {};
+ struct mlx5_modify_hdr *modify_hdr;
+ MLX5_DECLARE_FLOW_ACT(flow_act);
+ struct mlx5_flow_handle *rule;
+ struct mlx5_flow_spec *spec;
+ int err;
+
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+ if (!spec)
+ return -ENOMEM;
+
+ spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+ MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_protocol);
+ MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_protocol, IPPROTO_UDP);
+ MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.udp_dport);
+ MLX5_SET(fte_match_param, spec->match_value, outer_headers.udp_dport, ROCE_V2_UDP_DPORT);
+
+ MLX5_SET(copy_action_in, action, action_type, MLX5_ACTION_TYPE_COPY);
+ MLX5_SET(copy_action_in, action, src_field, MLX5_ACTION_IN_FIELD_METADATA_REG_B);
+ MLX5_SET(copy_action_in, action, src_offset, 0);
+ MLX5_SET(copy_action_in, action, length, 32);
+ MLX5_SET(copy_action_in, action, dst_field, MLX5_ACTION_IN_FIELD_METADATA_REG_C_5);
+ MLX5_SET(copy_action_in, action, dst_offset, 0);
+
+ modify_hdr = mlx5_modify_header_alloc(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC,
+ 1, action);
+
+ if (IS_ERR(modify_hdr)) {
+ err = PTR_ERR(modify_hdr);
+ mlx5_core_err(mdev,
+ "Failed to alloc macsec copy modify_header_id err(%d)\n", err);
+ goto err_alloc_hdr;
+ }
+
+ flow_act.action = MLX5_FLOW_CONTEXT_ACTION_MOD_HDR | MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+ flow_act.modify_hdr = modify_hdr;
+ dst.type = MLX5_FLOW_DESTINATION_TYPE_TABLE_TYPE;
+ dst.ft = roce->ft_ip_check;
+ rule = mlx5_add_flow_rules(roce->ft, spec, &flow_act, &dst, 1);
+ if (IS_ERR(rule)) {
+ err = PTR_ERR(rule);
+ mlx5_core_err(mdev, "Failed to add rule to MACsec RoCE NIC UDP table err(%d)\n",
+ err);
+ goto err_add_rule;
+ }
+ roce->rule = rule;
+ roce->copy_modify_hdr = modify_hdr;
+
+ memset(&flow_act, 0, sizeof(flow_act));
+ flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
+ rule = mlx5_add_flow_rules(roce->ft, NULL, &flow_act, NULL, 0);
+ if (IS_ERR(rule)) {
+ err = PTR_ERR(rule);
+ mlx5_core_err(mdev, "Failed to add miss rule to MACsec RoCE NIC UDP table err(%d)\n",
+ err);
+ goto err_add_rule2;
+ }
+ roce->nic_miss.rule = rule;
+
+ kvfree(spec);
+ return 0;
+
+err_add_rule2:
+ mlx5_del_flow_rules(roce->rule);
+err_add_rule:
+ mlx5_modify_header_dealloc(macsec_fs->mdev, modify_hdr);
+err_alloc_hdr:
+ kvfree(spec);
+ return err;
+}
+
+static int macsec_fs_rx_roce_jump_to_rdma_create(struct mlx5_macsec_fs *macsec_fs,
+ struct mlx5_macsec_rx_roce *roce)
+{
+ int err;
+
+ err = macsec_fs_rx_roce_jump_to_rdma_groups_create(macsec_fs->mdev, roce);
+ if (err)
+ return err;
+
+ err = macsec_fs_rx_roce_jump_to_rdma_rules_create(macsec_fs, roce);
+ if (err)
+ goto err;
+
+ return 0;
+err:
+ mlx5_destroy_flow_group(roce->nic_miss.g);
+ mlx5_destroy_flow_group(roce->g);
+ return err;
+}
+
+static int macsec_fs_rx_roce_create(struct mlx5_macsec_fs *macsec_fs)
+{
+ struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
+ struct mlx5_flow_table_attr ft_attr = {};
+ struct mlx5_flow_namespace *ns;
+ struct mlx5_flow_table *ft;
+ int err = 0;
+
+ if (!mlx5_is_macsec_roce_supported(macsec_fs->mdev)) {
+ mlx5_core_dbg(mdev, "Failed to init RoCE MACsec, capabilities not supported\n");
+ return 0;
+ }
+
+ ns = mlx5_get_flow_namespace(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_RDMA_RX_MACSEC);
+ if (!ns)
+ return -ENOMEM;
+
+ ft = macsec_fs_auto_group_table_create(ns, 0, RDMA_RX_ROCE_IP_TABLE_LEVEL,
+ CRYPTO_NUM_MAXSEC_FTE);
+ if (IS_ERR(ft)) {
+ err = PTR_ERR(ft);
+ mlx5_core_err(mdev,
+ "Failed to create MACsec IP check RoCE table err(%d)\n", err);
+ return err;
+ }
+ rx_fs->roce.ft_ip_check = ft;
+
+ ft = macsec_fs_auto_group_table_create(ns, 0, RDMA_RX_ROCE_MACSEC_OP_TABLE_LEVEL,
+ CRYPTO_NUM_MAXSEC_FTE);
+ if (IS_ERR(ft)) {
+ err = PTR_ERR(ft);
+ mlx5_core_err(mdev,
+ "Failed to create MACsec operation check RoCE table err(%d)\n",
+ err);
+ goto err_macsec_op;
+ }
+ rx_fs->roce.ft_macsec_op_check = ft;
+
+ err = macsec_fs_rx_roce_miss_create(mdev, &rx_fs->roce);
+ if (err)
+ goto err_miss_create;
+
+ ns = mlx5_get_flow_namespace(macsec_fs->mdev, MLX5_FLOW_NAMESPACE_KERNEL_RX_MACSEC);
+ if (!ns) {
+ err = -EOPNOTSUPP;
+ goto err_ns;
+ }
+
+ ft_attr.level = RX_ROCE_TABLE_LEVEL;
+ ft_attr.max_fte = RX_ROCE_TABLE_NUM_FTE;
+ ft = mlx5_create_flow_table(ns, &ft_attr);
+ if (IS_ERR(ft)) {
+ err = PTR_ERR(ft);
+ mlx5_core_err(mdev,
+ "Failed to create MACsec jump to RX RoCE, NIC table err(%d)\n", err);
+ goto err_ns;
+ }
+ rx_fs->roce.ft = ft;
+
+ err = macsec_fs_rx_roce_jump_to_rdma_create(macsec_fs, &rx_fs->roce);
+ if (err)
+ goto err_udp_ft;
+
+ return 0;
+
+err_udp_ft:
+ mlx5_destroy_flow_table(rx_fs->roce.ft);
+err_ns:
+ macsec_fs_rx_roce_miss_destroy(&rx_fs->roce.miss);
+err_miss_create:
+ mlx5_destroy_flow_table(rx_fs->roce.ft_macsec_op_check);
+err_macsec_op:
+ mlx5_destroy_flow_table(rx_fs->roce.ft_ip_check);
+ return err;
+}
+
static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
@@ -1167,6 +1506,10 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
rx_tables = &rx_fs->tables;
ft_crypto = &rx_tables->ft_crypto;
+ err = macsec_fs_rx_roce_create(macsec_fs);
+ if (err)
+ goto out_flow_group;
+
/* Rx crypto table */
ft_attr.level = RX_CRYPTO_TABLE_LEVEL;
ft_attr.max_fte = CRYPTO_NUM_MAXSEC_FTE;
@@ -1175,7 +1518,7 @@ static int macsec_fs_rx_create(struct mlx5_macsec_fs *macsec_fs)
if (IS_ERR(flow_table)) {
err = PTR_ERR(flow_table);
mlx5_core_err(mdev, "Failed to create MACsec Rx crypto table err(%d)\n", err);
- goto out_flow_group;
+ goto err;
}
ft_crypto->t = flow_table;
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 13/14] net/mlx5: Add RoCE MACsec steering infrastructure in core
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (11 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 12/14] net/mlx5: Configure MACsec steering for ingress " Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 8:29 ` [PATCH mlx5-next v1 14/14] RDMA/mlx5: Handles RoCE MACsec steering rules addition and deletion Leon Romanovsky
2023-08-09 23:09 ` [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Jakub Kicinski
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Adds all the core steering helper functions that are needed in order
to setup RoCE steering rules which includes both the RX and TX rules
addition and deletion.
As well as exporting the function to be ready to use from the IB driver
where we expose functions to allow deletion of all rules, which is
needed when a GID is deleted, or a deletion of a specific rule when an SA
is deleted, and a similar manner for the rules addition.
These functions are used in a later patch by IB driver to trigger the
rules addition/deletion when needed.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/fs_cmd.c | 1 +
.../mellanox/mlx5/core/lib/macsec_fs.c | 399 +++++++++++++++++-
include/linux/mlx5/device.h | 2 +
include/linux/mlx5/driver.h | 2 +
include/linux/mlx5/macsec.h | 32 ++
5 files changed, 427 insertions(+), 9 deletions(-)
create mode 100644 include/linux/mlx5/macsec.h
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
index aab7059bf6e9..51c13d8cc55e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
@@ -967,6 +967,7 @@ static int mlx5_cmd_modify_header_alloc(struct mlx5_flow_root_namespace *ns,
max_actions = MLX5_CAP_ESW_INGRESS_ACL(dev, max_modify_header_actions);
table_type = FS_FT_ESW_INGRESS_ACL;
break;
+ case MLX5_FLOW_NAMESPACE_RDMA_TX_MACSEC:
case MLX5_FLOW_NAMESPACE_RDMA_TX:
max_actions = MLX5_CAP_FLOWTABLE_RDMA_TX(dev, max_modify_header_actions);
table_type = FS_FT_RDMA_TX;
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 be909b613288..4a078113e292 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/macsec_fs.c
@@ -4,6 +4,8 @@
#include <net/macsec.h>
#include <linux/mlx5/qp.h>
#include <linux/if_vlan.h>
+#include <linux/mlx5/fs_helpers.h>
+#include <linux/mlx5/macsec.h>
#include "fs_core.h"
#include "lib/macsec_fs.h"
#include "mlx5_core.h"
@@ -46,6 +48,10 @@
/* MACsec RX flow steering */
#define MLX5_ETH_WQE_FT_META_MACSEC_MASK 0x3E
+/* MACsec fs_id handling for steering */
+#define macsec_fs_set_tx_fs_id(fs_id) (MLX5_ETH_WQE_FT_META_MACSEC | (fs_id) << 2)
+#define macsec_fs_set_rx_fs_id(fs_id) ((fs_id) | BIT(30))
+
struct mlx5_sectag_header {
__be16 ethertype;
u8 tci_an;
@@ -54,6 +60,14 @@ struct mlx5_sectag_header {
u8 sci[MACSEC_SCI_LEN]; /* optional */
} __packed;
+struct mlx5_roce_macsec_tx_rule {
+ u32 fs_id;
+ u16 gid_idx;
+ struct list_head entry;
+ struct mlx5_flow_handle *rule;
+ struct mlx5_modify_hdr *meta_modhdr;
+};
+
struct mlx5_macsec_tx_rule {
struct mlx5_flow_handle *rule;
struct mlx5_pkt_reformat *pkt_reformat;
@@ -104,6 +118,14 @@ struct mlx5_macsec_tx {
struct mlx5_flow_table *ft_rdma_tx;
};
+struct mlx5_roce_macsec_rx_rule {
+ u32 fs_id;
+ u16 gid_idx;
+ struct mlx5_flow_handle *op;
+ struct mlx5_flow_handle *ip;
+ struct list_head entry;
+};
+
struct mlx5_macsec_rx_rule {
struct mlx5_flow_handle *rule[RX_NUM_OF_RULES_PER_SA];
struct mlx5_modify_hdr *meta_modhdr;
@@ -575,7 +597,7 @@ static int macsec_fs_tx_setup_fte(struct mlx5_macsec_fs *macsec_fs,
MLX5_SET(fte_match_param, spec->match_criteria, misc_parameters_2.metadata_reg_a,
MLX5_ETH_WQE_FT_META_MACSEC_MASK);
MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_a,
- MLX5_ETH_WQE_FT_META_MACSEC | id << 2);
+ macsec_fs_set_tx_fs_id(id));
*fs_id = id;
flow_act->crypto.type = MLX5_FLOW_CONTEXT_ENCRYPT_DECRYPT_TYPE_MACSEC;
@@ -777,7 +799,7 @@ static void macsec_fs_tx_del_rule(struct mlx5_macsec_fs *macsec_fs,
static union mlx5_macsec_rule *
macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
const struct macsec_context *macsec_ctx,
- struct mlx5_macsec_rule_attrs *attrs)
+ struct mlx5_macsec_rule_attrs *attrs, u32 *fs_id)
{
char reformatbf[MLX5_MACSEC_TAG_LEN + MACSEC_SCI_LEN];
struct mlx5_pkt_reformat_params reformat_params = {};
@@ -792,7 +814,6 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
struct mlx5_flow_spec *spec;
size_t reformat_size;
int err = 0;
- u32 fs_id;
tx_tables = &tx_fs->tables;
@@ -832,7 +853,7 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
}
tx_rule->pkt_reformat = flow_act.pkt_reformat;
- err = macsec_fs_tx_setup_fte(macsec_fs, spec, &flow_act, attrs->macsec_obj_id, &fs_id);
+ err = macsec_fs_tx_setup_fte(macsec_fs, spec, &flow_act, attrs->macsec_obj_id, fs_id);
if (err) {
mlx5_core_err(mdev,
"Failed to add packet reformat for MACsec TX crypto rule, err=%d\n",
@@ -840,7 +861,7 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
goto err;
}
- tx_rule->fs_id = fs_id;
+ tx_rule->fs_id = *fs_id;
flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST |
MLX5_FLOW_CONTEXT_ACTION_CRYPTO_ENCRYPT |
@@ -855,7 +876,7 @@ macsec_fs_tx_add_rule(struct mlx5_macsec_fs *macsec_fs,
}
tx_rule->rule = rule;
- err = macsec_fs_id_add(&macsec_fs->macsec_devices_list, fs_id, macsec_ctx->secy->netdev,
+ err = macsec_fs_id_add(&macsec_fs->macsec_devices_list, *fs_id, macsec_ctx->secy->netdev,
&macsec_fs->sci_hash, attrs->sci, true);
if (err) {
mlx5_core_err(mdev, "Failed to save fs_id, err=%d\n", err);
@@ -1744,7 +1765,7 @@ macsec_fs_rx_add_rule(struct mlx5_macsec_fs *macsec_fs,
/* Set bit[15-0] fs id */
MLX5_SET(set_action_in, action, action_type, MLX5_ACTION_TYPE_SET);
MLX5_SET(set_action_in, action, field, MLX5_ACTION_IN_FIELD_METADATA_REG_B);
- MLX5_SET(set_action_in, action, data, MLX5_MACSEC_RX_METADAT_HANDLE(fs_id) | BIT(30));
+ MLX5_SET(set_action_in, action, data, macsec_fs_set_rx_fs_id(fs_id));
MLX5_SET(set_action_in, action, offset, 0);
MLX5_SET(set_action_in, action, length, 32);
@@ -1903,6 +1924,114 @@ static void macsec_fs_rx_cleanup(struct mlx5_macsec_fs *macsec_fs)
macsec_fs->rx_fs = NULL;
}
+static void set_ipaddr_spec_v4(struct sockaddr_in *in, struct mlx5_flow_spec *spec, bool is_dst_ip)
+{
+ MLX5_SET(fte_match_param, spec->match_value,
+ outer_headers.ip_version, MLX5_FS_IPV4_VERSION);
+
+ if (is_dst_ip) {
+ MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+ outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
+ memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+ outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+ &in->sin_addr.s_addr, 4);
+ } else {
+ MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+ outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4);
+ memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+ outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4),
+ &in->sin_addr.s_addr, 4);
+ }
+}
+
+static void set_ipaddr_spec_v6(struct sockaddr_in6 *in6, struct mlx5_flow_spec *spec,
+ bool is_dst_ip)
+{
+ MLX5_SET(fte_match_param, spec->match_value,
+ outer_headers.ip_version, MLX5_FS_IPV6_VERSION);
+
+ if (is_dst_ip) {
+ memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+ outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+ 0xff, 16);
+ memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+ outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+ &in6->sin6_addr, 16);
+ } else {
+ memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+ outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+ 0xff, 16);
+ memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
+ outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
+ &in6->sin6_addr, 16);
+ }
+}
+
+static void set_ipaddr_spec(const struct sockaddr *addr,
+ struct mlx5_flow_spec *spec, bool is_dst_ip)
+{
+ struct sockaddr_in6 *in6;
+
+ spec->match_criteria_enable |= MLX5_MATCH_OUTER_HEADERS;
+ MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+ outer_headers.ip_version);
+
+ if (addr->sa_family == AF_INET) {
+ struct sockaddr_in *in = (struct sockaddr_in *)addr;
+
+ set_ipaddr_spec_v4(in, spec, is_dst_ip);
+ return;
+ }
+
+ in6 = (struct sockaddr_in6 *)addr;
+ set_ipaddr_spec_v6(in6, spec, is_dst_ip);
+}
+
+static void macsec_fs_del_roce_rule_rx(struct mlx5_roce_macsec_rx_rule *rx_rule)
+{
+ mlx5_del_flow_rules(rx_rule->op);
+ mlx5_del_flow_rules(rx_rule->ip);
+ list_del(&rx_rule->entry);
+ kfree(rx_rule);
+}
+
+static void macsec_fs_del_roce_rules_rx(struct mlx5_macsec_fs *macsec_fs, u32 fs_id,
+ struct list_head *rx_rules_list)
+{
+ struct mlx5_roce_macsec_rx_rule *rx_rule, *next;
+
+ if (!mlx5_is_macsec_roce_supported(macsec_fs->mdev))
+ return;
+
+ list_for_each_entry_safe(rx_rule, next, rx_rules_list, entry) {
+ if (rx_rule->fs_id == fs_id)
+ macsec_fs_del_roce_rule_rx(rx_rule);
+ }
+}
+
+static void macsec_fs_del_roce_rule_tx(struct mlx5_core_dev *mdev,
+ struct mlx5_roce_macsec_tx_rule *tx_rule)
+{
+ mlx5_del_flow_rules(tx_rule->rule);
+ mlx5_modify_header_dealloc(mdev, tx_rule->meta_modhdr);
+ list_del(&tx_rule->entry);
+ kfree(tx_rule);
+}
+
+static void macsec_fs_del_roce_rules_tx(struct mlx5_macsec_fs *macsec_fs, u32 fs_id,
+ struct list_head *tx_rules_list)
+{
+ struct mlx5_roce_macsec_tx_rule *tx_rule, *next;
+
+ if (!mlx5_is_macsec_roce_supported(macsec_fs->mdev))
+ return;
+
+ list_for_each_entry_safe(tx_rule, next, tx_rules_list, entry) {
+ if (tx_rule->fs_id == fs_id)
+ macsec_fs_del_roce_rule_tx(macsec_fs->mdev, tx_rule);
+ }
+}
+
void mlx5_macsec_fs_get_stats_fill(struct mlx5_macsec_fs *macsec_fs, void *macsec_stats)
{
struct mlx5_macsec_stats *stats = (struct mlx5_macsec_stats *)macsec_stats;
@@ -1955,20 +2084,270 @@ mlx5_macsec_fs_add_rule(struct mlx5_macsec_fs *macsec_fs,
struct mlx5_macsec_rule_attrs *attrs,
u32 *sa_fs_id)
{
- return (attrs->action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
- macsec_fs_tx_add_rule(macsec_fs, macsec_ctx, attrs) :
+ struct mlx5_macsec_event_data data = {.macsec_fs = macsec_fs,
+ .macdev = macsec_ctx->secy->netdev,
+ .is_tx =
+ (attrs->action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT)
+ };
+ union mlx5_macsec_rule *macsec_rule;
+ u32 tx_new_fs_id;
+
+ macsec_rule = (attrs->action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
+ macsec_fs_tx_add_rule(macsec_fs, macsec_ctx, attrs, &tx_new_fs_id) :
macsec_fs_rx_add_rule(macsec_fs, macsec_ctx, attrs, *sa_fs_id);
+
+ data.fs_id = (data.is_tx) ? tx_new_fs_id : *sa_fs_id;
+ if (macsec_rule)
+ blocking_notifier_call_chain(&macsec_fs->mdev->macsec_nh,
+ MLX5_DRIVER_EVENT_MACSEC_SA_ADDED,
+ &data);
+
+ return macsec_rule;
}
void mlx5_macsec_fs_del_rule(struct mlx5_macsec_fs *macsec_fs,
union mlx5_macsec_rule *macsec_rule,
int action, void *macdev, u32 sa_fs_id)
{
+ struct mlx5_macsec_event_data data = {.macsec_fs = macsec_fs,
+ .macdev = macdev,
+ .is_tx = (action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT)
+ };
+
+ data.fs_id = (data.is_tx) ? macsec_rule->tx_rule.fs_id : sa_fs_id;
+ blocking_notifier_call_chain(&macsec_fs->mdev->macsec_nh,
+ MLX5_DRIVER_EVENT_MACSEC_SA_DELETED,
+ &data);
+
(action == MLX5_ACCEL_MACSEC_ACTION_ENCRYPT) ?
macsec_fs_tx_del_rule(macsec_fs, &macsec_rule->tx_rule, macdev) :
macsec_fs_rx_del_rule(macsec_fs, &macsec_rule->rx_rule, macdev, sa_fs_id);
}
+static int mlx5_macsec_fs_add_roce_rule_rx(struct mlx5_macsec_fs *macsec_fs, u32 fs_id, u16 gid_idx,
+ const struct sockaddr *addr,
+ struct list_head *rx_rules_list)
+{
+ struct mlx5_macsec_rx *rx_fs = macsec_fs->rx_fs;
+ struct mlx5_roce_macsec_rx_rule *rx_rule;
+ struct mlx5_flow_destination dest = {};
+ struct mlx5_flow_act flow_act = {};
+ struct mlx5_flow_handle *new_rule;
+ struct mlx5_flow_spec *spec;
+ int err = 0;
+
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+ if (!spec)
+ return -ENOMEM;
+
+ rx_rule = kzalloc(sizeof(*rx_rule), GFP_KERNEL);
+ if (!rx_rule) {
+ err = -ENOMEM;
+ goto out;
+ }
+
+ set_ipaddr_spec(addr, spec, true);
+
+ flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+ dest.ft = rx_fs->roce.ft_macsec_op_check;
+ dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
+ new_rule = mlx5_add_flow_rules(rx_fs->roce.ft_ip_check, spec, &flow_act,
+ &dest, 1);
+ if (IS_ERR(new_rule)) {
+ err = PTR_ERR(new_rule);
+ goto ip_rule_err;
+ }
+ rx_rule->ip = new_rule;
+
+ memset(&flow_act, 0, sizeof(flow_act));
+ memset(spec, 0, sizeof(*spec));
+
+ spec->match_criteria_enable |= MLX5_MATCH_MISC_PARAMETERS_2;
+ MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, misc_parameters_2.metadata_reg_c_5);
+ MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_c_5,
+ macsec_fs_set_rx_fs_id(fs_id));
+ flow_act.action = MLX5_FLOW_CONTEXT_ACTION_ALLOW;
+ new_rule = mlx5_add_flow_rules(rx_fs->roce.ft_macsec_op_check, spec, &flow_act,
+ NULL, 0);
+ if (IS_ERR(new_rule)) {
+ err = PTR_ERR(new_rule);
+ goto op_rule_err;
+ }
+ rx_rule->op = new_rule;
+ rx_rule->gid_idx = gid_idx;
+ rx_rule->fs_id = fs_id;
+ list_add_tail(&rx_rule->entry, rx_rules_list);
+
+ goto out;
+
+op_rule_err:
+ mlx5_del_flow_rules(rx_rule->ip);
+ rx_rule->ip = NULL;
+ip_rule_err:
+ kfree(rx_rule);
+out:
+ kvfree(spec);
+ return err;
+}
+
+static int mlx5_macsec_fs_add_roce_rule_tx(struct mlx5_macsec_fs *macsec_fs, u32 fs_id, u16 gid_idx,
+ const struct sockaddr *addr,
+ struct list_head *tx_rules_list)
+{
+ u8 action[MLX5_UN_SZ_BYTES(set_add_copy_action_in_auto)] = {};
+ struct mlx5_macsec_tx *tx_fs = macsec_fs->tx_fs;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
+ struct mlx5_modify_hdr *modify_hdr = NULL;
+ struct mlx5_roce_macsec_tx_rule *tx_rule;
+ struct mlx5_flow_destination dest = {};
+ struct mlx5_flow_act flow_act = {};
+ struct mlx5_flow_handle *new_rule;
+ struct mlx5_flow_spec *spec;
+ int err = 0;
+
+ spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+ if (!spec)
+ return -ENOMEM;
+
+ tx_rule = kzalloc(sizeof(*tx_rule), GFP_KERNEL);
+ if (!tx_rule) {
+ err = -ENOMEM;
+ goto out;
+ }
+
+ set_ipaddr_spec(addr, spec, false);
+
+ MLX5_SET(set_action_in, action, action_type, MLX5_ACTION_TYPE_SET);
+ MLX5_SET(set_action_in, action, field, MLX5_ACTION_IN_FIELD_METADATA_REG_A);
+ MLX5_SET(set_action_in, action, data, macsec_fs_set_tx_fs_id(fs_id));
+ MLX5_SET(set_action_in, action, offset, 0);
+ MLX5_SET(set_action_in, action, length, 32);
+
+ modify_hdr = mlx5_modify_header_alloc(mdev, MLX5_FLOW_NAMESPACE_RDMA_TX_MACSEC,
+ 1, action);
+ if (IS_ERR(modify_hdr)) {
+ err = PTR_ERR(modify_hdr);
+ mlx5_core_err(mdev, "Fail to alloc ROCE MACsec set modify_header_id err=%d\n",
+ err);
+ modify_hdr = NULL;
+ goto modify_hdr_err;
+ }
+ tx_rule->meta_modhdr = modify_hdr;
+
+ flow_act.modify_hdr = modify_hdr;
+ flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | MLX5_FLOW_CONTEXT_ACTION_MOD_HDR;
+
+ dest.type = MLX5_FLOW_DESTINATION_TYPE_TABLE_TYPE;
+ dest.ft = tx_fs->tables.ft_crypto.t;
+ new_rule = mlx5_add_flow_rules(tx_fs->ft_rdma_tx, spec, &flow_act, &dest, 1);
+ if (IS_ERR(new_rule)) {
+ err = PTR_ERR(new_rule);
+ mlx5_core_err(mdev, "Failed to add ROCE TX rule, err=%d\n", err);
+ goto rule_err;
+ }
+ tx_rule->rule = new_rule;
+ tx_rule->gid_idx = gid_idx;
+ tx_rule->fs_id = fs_id;
+ list_add_tail(&tx_rule->entry, tx_rules_list);
+
+ goto out;
+
+rule_err:
+ mlx5_modify_header_dealloc(mdev, tx_rule->meta_modhdr);
+modify_hdr_err:
+ kfree(tx_rule);
+out:
+ kvfree(spec);
+ return err;
+}
+
+void mlx5_macsec_del_roce_rule(u16 gid_idx, struct mlx5_macsec_fs *macsec_fs,
+ struct list_head *tx_rules_list, struct list_head *rx_rules_list)
+{
+ struct mlx5_roce_macsec_rx_rule *rx_rule, *next_rx;
+ struct mlx5_roce_macsec_tx_rule *tx_rule, *next_tx;
+
+ list_for_each_entry_safe(tx_rule, next_tx, tx_rules_list, entry) {
+ if (tx_rule->gid_idx == gid_idx)
+ macsec_fs_del_roce_rule_tx(macsec_fs->mdev, tx_rule);
+ }
+
+ list_for_each_entry_safe(rx_rule, next_rx, rx_rules_list, entry) {
+ if (rx_rule->gid_idx == gid_idx)
+ macsec_fs_del_roce_rule_rx(rx_rule);
+ }
+}
+EXPORT_SYMBOL_GPL(mlx5_macsec_del_roce_rule);
+
+int mlx5_macsec_add_roce_rule(void *macdev, const struct sockaddr *addr, u16 gid_idx,
+ struct list_head *tx_rules_list, struct list_head *rx_rules_list,
+ struct mlx5_macsec_fs *macsec_fs)
+{
+ struct mlx5_macsec_device *iter, *macsec_device = NULL;
+ struct mlx5_core_dev *mdev = macsec_fs->mdev;
+ struct mlx5_fs_id *fs_id_iter;
+ unsigned long index = 0;
+ int err;
+
+ list_for_each_entry(iter, &macsec_fs->macsec_devices_list, macsec_devices_list_entry) {
+ if (iter->macdev == macdev) {
+ macsec_device = iter;
+ break;
+ }
+ }
+
+ if (!macsec_device)
+ return 0;
+
+ xa_for_each(&macsec_device->tx_id_xa, index, fs_id_iter) {
+ err = mlx5_macsec_fs_add_roce_rule_tx(macsec_fs, fs_id_iter->id, gid_idx, addr,
+ tx_rules_list);
+ if (err) {
+ mlx5_core_err(mdev, "MACsec offload: Failed to add roce TX rule\n");
+ goto out;
+ }
+ }
+
+ index = 0;
+ xa_for_each(&macsec_device->rx_id_xa, index, fs_id_iter) {
+ err = mlx5_macsec_fs_add_roce_rule_rx(macsec_fs, fs_id_iter->id, gid_idx, addr,
+ rx_rules_list);
+ if (err) {
+ mlx5_core_err(mdev, "MACsec offload: Failed to add roce TX rule\n");
+ goto out;
+ }
+ }
+
+ return 0;
+out:
+ mlx5_macsec_del_roce_rule(gid_idx, macsec_fs, tx_rules_list, rx_rules_list);
+ return err;
+}
+EXPORT_SYMBOL_GPL(mlx5_macsec_add_roce_rule);
+
+void mlx5_macsec_add_roce_sa_rules(u32 fs_id, const struct sockaddr *addr, u16 gid_idx,
+ struct list_head *tx_rules_list,
+ struct list_head *rx_rules_list,
+ struct mlx5_macsec_fs *macsec_fs, bool is_tx)
+{
+ (is_tx) ?
+ mlx5_macsec_fs_add_roce_rule_tx(macsec_fs, fs_id, gid_idx, addr,
+ tx_rules_list) :
+ mlx5_macsec_fs_add_roce_rule_rx(macsec_fs, fs_id, gid_idx, addr,
+ rx_rules_list);
+}
+EXPORT_SYMBOL_GPL(mlx5_macsec_add_roce_sa_rules);
+
+void mlx5_macsec_del_roce_sa_rules(u32 fs_id, struct mlx5_macsec_fs *macsec_fs,
+ struct list_head *tx_rules_list,
+ struct list_head *rx_rules_list, bool is_tx)
+{
+ (is_tx) ?
+ macsec_fs_del_roce_rules_tx(macsec_fs, fs_id, tx_rules_list) :
+ macsec_fs_del_roce_rules_rx(macsec_fs, fs_id, rx_rules_list);
+}
+EXPORT_SYMBOL_GPL(mlx5_macsec_del_roce_sa_rules);
+
void mlx5_macsec_fs_cleanup(struct mlx5_macsec_fs *macsec_fs)
{
macsec_fs_rx_cleanup(macsec_fs);
@@ -2016,6 +2395,8 @@ mlx5_macsec_fs_init(struct mlx5_core_dev *mdev)
goto tx_cleanup;
}
+ BLOCKING_INIT_NOTIFIER_HEAD(&mdev->macsec_nh);
+
return macsec_fs;
tx_cleanup:
diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h
index 80cc12a9a531..ca93a5ef9dac 100644
--- a/include/linux/mlx5/device.h
+++ b/include/linux/mlx5/device.h
@@ -364,6 +364,8 @@ enum mlx5_event {
enum mlx5_driver_event {
MLX5_DRIVER_EVENT_TYPE_TRAP = 0,
MLX5_DRIVER_EVENT_UPLINK_NETDEV,
+ MLX5_DRIVER_EVENT_MACSEC_SA_ADDED,
+ MLX5_DRIVER_EVENT_MACSEC_SA_DELETED,
};
enum {
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index c1b88a1112cc..848993823ba5 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -810,6 +810,8 @@ struct mlx5_core_dev {
u64 num_block_ipsec;
#ifdef CONFIG_MLX5_MACSEC
struct mlx5_macsec_fs *macsec_fs;
+ /* MACsec notifier chain to sync MACsec core and IB database */
+ struct blocking_notifier_head macsec_nh;
#endif
};
diff --git a/include/linux/mlx5/macsec.h b/include/linux/mlx5/macsec.h
new file mode 100644
index 000000000000..f7ff4c2a95d0
--- /dev/null
+++ b/include/linux/mlx5/macsec.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
+/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. */
+
+#ifndef MLX5_MACSEC_H
+#define MLX5_MACSEC_H
+
+#ifdef CONFIG_MLX5_MACSEC
+struct mlx5_macsec_event_data {
+ struct mlx5_macsec_fs *macsec_fs;
+ void *macdev;
+ u32 fs_id;
+ bool is_tx;
+};
+
+int mlx5_macsec_add_roce_rule(void *macdev, const struct sockaddr *addr, u16 gid_idx,
+ struct list_head *tx_rules_list, struct list_head *rx_rules_list,
+ struct mlx5_macsec_fs *macsec_fs);
+
+void mlx5_macsec_del_roce_rule(u16 gid_idx, struct mlx5_macsec_fs *macsec_fs,
+ struct list_head *tx_rules_list, struct list_head *rx_rules_list);
+
+void mlx5_macsec_add_roce_sa_rules(u32 fs_id, const struct sockaddr *addr, u16 gid_idx,
+ struct list_head *tx_rules_list,
+ struct list_head *rx_rules_list,
+ struct mlx5_macsec_fs *macsec_fs, bool is_tx);
+
+void mlx5_macsec_del_roce_sa_rules(u32 fs_id, struct mlx5_macsec_fs *macsec_fs,
+ struct list_head *tx_rules_list,
+ struct list_head *rx_rules_list, bool is_tx);
+
+#endif
+#endif /* MLX5_MACSEC_H */
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* [PATCH mlx5-next v1 14/14] RDMA/mlx5: Handles RoCE MACsec steering rules addition and deletion
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (12 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 13/14] net/mlx5: Add RoCE MACsec steering infrastructure in core Leon Romanovsky
@ 2023-08-09 8:29 ` Leon Romanovsky
2023-08-09 23:09 ` [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Jakub Kicinski
14 siblings, 0 replies; 18+ messages in thread
From: Leon Romanovsky @ 2023-08-09 8:29 UTC (permalink / raw)
To: Jason Gunthorpe, Jakub Kicinski
Cc: Patrisious Haddad, Leon Romanovsky, David S . Miller,
Eric Dumazet, linux-rdma, Maor Gottlieb, Mark Zhang, netdev,
Paolo Abeni, Raed Salem, Saeed Mahameed
From: Patrisious Haddad <phaddad@nvidia.com>
Add RoCE MACsec rules when a gid is added for the MACsec netdevice and
handle their cleanup when the gid is removed or the MACsec SA is deleted.
Also support alias IP for the MACsec device, as long as we don't have
more ips than what the gid table can hold.
In addition handle the case where a gid is added but there are still no
SAs added for the MACsec device, so the rules are added later on when
the SAs are added.
Signed-off-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
drivers/infiniband/hw/mlx5/macsec.c | 235 +++++++++++++++++++++++++--
drivers/infiniband/hw/mlx5/macsec.h | 8 +-
drivers/infiniband/hw/mlx5/main.c | 6 +-
drivers/infiniband/hw/mlx5/mlx5_ib.h | 10 ++
include/linux/mlx5/driver.h | 4 +-
5 files changed, 246 insertions(+), 17 deletions(-)
diff --git a/drivers/infiniband/hw/mlx5/macsec.c b/drivers/infiniband/hw/mlx5/macsec.c
index 349ad13af75d..3c56eb5eddf3 100644
--- a/drivers/infiniband/hw/mlx5/macsec.c
+++ b/drivers/infiniband/hw/mlx5/macsec.c
@@ -2,13 +2,174 @@
/* Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. */
#include "macsec.h"
+#include <linux/mlx5/macsec.h>
struct mlx5_reserved_gids {
int macsec_index;
const struct ib_gid_attr *physical_gid;
};
-int mlx5r_macsec_alloc_gids(struct mlx5_ib_dev *dev)
+struct mlx5_roce_gids {
+ struct list_head roce_gid_list_entry;
+ u16 gid_idx;
+ union {
+ struct sockaddr_in sockaddr_in;
+ struct sockaddr_in6 sockaddr_in6;
+ } addr;
+};
+
+struct mlx5_macsec_device {
+ struct list_head macsec_devices_list_entry;
+ void *macdev;
+ struct list_head macsec_roce_gids;
+ struct list_head tx_rules_list;
+ struct list_head rx_rules_list;
+};
+
+static void cleanup_macsec_device(struct mlx5_macsec_device *macsec_device)
+{
+ if (!list_empty(&macsec_device->tx_rules_list) ||
+ !list_empty(&macsec_device->rx_rules_list) ||
+ !list_empty(&macsec_device->macsec_roce_gids))
+ return;
+
+ list_del(&macsec_device->macsec_devices_list_entry);
+ kfree(macsec_device);
+}
+
+static struct mlx5_macsec_device *get_macsec_device(void *macdev,
+ struct list_head *macsec_devices_list)
+{
+ struct mlx5_macsec_device *iter, *macsec_device = NULL;
+
+ list_for_each_entry(iter, macsec_devices_list, macsec_devices_list_entry) {
+ if (iter->macdev == macdev) {
+ macsec_device = iter;
+ break;
+ }
+ }
+
+ if (macsec_device)
+ return macsec_device;
+
+ macsec_device = kzalloc(sizeof(*macsec_device), GFP_KERNEL);
+ if (!macsec_device)
+ return NULL;
+
+ macsec_device->macdev = macdev;
+ INIT_LIST_HEAD(&macsec_device->tx_rules_list);
+ INIT_LIST_HEAD(&macsec_device->rx_rules_list);
+ INIT_LIST_HEAD(&macsec_device->macsec_roce_gids);
+ list_add(&macsec_device->macsec_devices_list_entry, macsec_devices_list);
+
+ return macsec_device;
+}
+
+static void mlx5_macsec_del_roce_gid(struct mlx5_macsec_device *macsec_device, u16 gid_idx)
+{
+ struct mlx5_roce_gids *current_gid, *next_gid;
+
+ list_for_each_entry_safe(current_gid, next_gid, &macsec_device->macsec_roce_gids,
+ roce_gid_list_entry)
+ if (current_gid->gid_idx == gid_idx) {
+ list_del(¤t_gid->roce_gid_list_entry);
+ kfree(current_gid);
+ }
+}
+
+static void mlx5_macsec_save_roce_gid(struct mlx5_macsec_device *macsec_device,
+ const struct sockaddr *addr, u16 gid_idx)
+{
+ struct mlx5_roce_gids *roce_gids;
+
+ roce_gids = kzalloc(sizeof(*roce_gids), GFP_KERNEL);
+ if (!roce_gids)
+ return;
+
+ roce_gids->gid_idx = gid_idx;
+ if (addr->sa_family == AF_INET)
+ memcpy(&roce_gids->addr.sockaddr_in, addr, sizeof(roce_gids->addr.sockaddr_in));
+ else
+ memcpy(&roce_gids->addr.sockaddr_in6, addr, sizeof(roce_gids->addr.sockaddr_in6));
+
+ list_add_tail(&roce_gids->roce_gid_list_entry, &macsec_device->macsec_roce_gids);
+}
+
+static void handle_macsec_gids(struct list_head *macsec_devices_list,
+ struct mlx5_macsec_event_data *data)
+{
+ struct mlx5_macsec_device *macsec_device;
+ struct mlx5_roce_gids *gid;
+
+ macsec_device = get_macsec_device(data->macdev, macsec_devices_list);
+ if (!macsec_device)
+ return;
+
+ list_for_each_entry(gid, &macsec_device->macsec_roce_gids, roce_gid_list_entry) {
+ mlx5_macsec_add_roce_sa_rules(data->fs_id, (struct sockaddr *)&gid->addr,
+ gid->gid_idx, &macsec_device->tx_rules_list,
+ &macsec_device->rx_rules_list, data->macsec_fs,
+ data->is_tx);
+ }
+}
+
+static void del_sa_roce_rule(struct list_head *macsec_devices_list,
+ struct mlx5_macsec_event_data *data)
+{
+ struct mlx5_macsec_device *macsec_device;
+
+ macsec_device = get_macsec_device(data->macdev, macsec_devices_list);
+ WARN_ON(!macsec_device);
+
+ mlx5_macsec_del_roce_sa_rules(data->fs_id, data->macsec_fs,
+ &macsec_device->tx_rules_list,
+ &macsec_device->rx_rules_list, data->is_tx);
+}
+
+static int macsec_event(struct notifier_block *nb, unsigned long event, void *data)
+{
+ struct mlx5_macsec *macsec = container_of(nb, struct mlx5_macsec, blocking_events_nb);
+
+ mutex_lock(&macsec->lock);
+ switch (event) {
+ case MLX5_DRIVER_EVENT_MACSEC_SA_ADDED:
+ handle_macsec_gids(&macsec->macsec_devices_list, data);
+ break;
+ case MLX5_DRIVER_EVENT_MACSEC_SA_DELETED:
+ del_sa_roce_rule(&macsec->macsec_devices_list, data);
+ break;
+ default:
+ mutex_unlock(&macsec->lock);
+ return NOTIFY_DONE;
+ }
+ mutex_unlock(&macsec->lock);
+ return NOTIFY_OK;
+}
+
+void mlx5r_macsec_event_register(struct mlx5_ib_dev *dev)
+{
+ if (!mlx5_is_macsec_roce_supported(dev->mdev)) {
+ mlx5_ib_dbg(dev, "RoCE MACsec not supported due to capabilities\n");
+ return;
+ }
+
+ dev->macsec.blocking_events_nb.notifier_call = macsec_event;
+ blocking_notifier_chain_register(&dev->mdev->macsec_nh,
+ &dev->macsec.blocking_events_nb);
+}
+
+void mlx5r_macsec_event_unregister(struct mlx5_ib_dev *dev)
+{
+ if (!mlx5_is_macsec_roce_supported(dev->mdev)) {
+ mlx5_ib_dbg(dev, "RoCE MACsec not supported due to capabilities\n");
+ return;
+ }
+
+ blocking_notifier_chain_unregister(&dev->mdev->macsec_nh,
+ &dev->macsec.blocking_events_nb);
+}
+
+int mlx5r_macsec_init_gids_and_devlist(struct mlx5_ib_dev *dev)
{
int i, j, max_gids;
@@ -29,6 +190,9 @@ int mlx5r_macsec_alloc_gids(struct mlx5_ib_dev *dev)
dev->port[i].reserved_gids[j].macsec_index = -1;
}
+ INIT_LIST_HEAD(&dev->macsec.macsec_devices_list);
+ mutex_init(&dev->macsec.lock);
+
return 0;
err:
while (i >= 0) {
@@ -47,15 +211,22 @@ void mlx5r_macsec_dealloc_gids(struct mlx5_ib_dev *dev)
for (i = 0; i < dev->num_ports; i++)
kfree(dev->port[i].reserved_gids);
+
+ mutex_destroy(&dev->macsec.lock);
}
int mlx5r_add_gid_macsec_operations(const struct ib_gid_attr *attr)
{
struct mlx5_ib_dev *dev = to_mdev(attr->device);
+ struct mlx5_macsec_device *macsec_device;
const struct ib_gid_attr *physical_gid;
struct mlx5_reserved_gids *mgids;
struct net_device *ndev;
int ret = 0;
+ union {
+ struct sockaddr_in sockaddr_in;
+ struct sockaddr_in6 sockaddr_in6;
+ } addr;
if (attr->gid_type != IB_GID_TYPE_ROCE_UDP_ENCAP)
return 0;
@@ -76,34 +247,62 @@ int mlx5r_add_gid_macsec_operations(const struct ib_gid_attr *attr)
rcu_read_unlock();
return 0;
}
+ dev_hold(ndev);
rcu_read_unlock();
+ mutex_lock(&dev->macsec.lock);
+ macsec_device = get_macsec_device(ndev, &dev->macsec.macsec_devices_list);
+ if (!macsec_device) {
+ ret = -ENOMEM;
+ goto dev_err;
+ }
+
physical_gid = rdma_find_gid(attr->device, &attr->gid,
attr->gid_type, NULL);
- if (IS_ERR(physical_gid))
- return 0;
+ if (!IS_ERR(physical_gid)) {
+ ret = set_roce_addr(to_mdev(physical_gid->device),
+ physical_gid->port_num,
+ physical_gid->index, NULL,
+ physical_gid);
+ if (ret)
+ goto gid_err;
- ret = set_roce_addr(to_mdev(physical_gid->device),
- physical_gid->port_num,
- physical_gid->index, NULL,
- physical_gid);
- if (ret)
- goto gid_err;
+ mgids = &dev->port[attr->port_num - 1].reserved_gids[physical_gid->index];
+ mgids->macsec_index = attr->index;
+ mgids->physical_gid = physical_gid;
+ }
- mgids = &dev->port[attr->port_num - 1].reserved_gids[physical_gid->index];
- mgids->macsec_index = attr->index;
- mgids->physical_gid = physical_gid;
+ /* Proceed with adding steering rules, regardless if there was gid ambiguity or not.*/
+ rdma_gid2ip((struct sockaddr *)&addr, &attr->gid);
+ ret = mlx5_macsec_add_roce_rule(ndev, (struct sockaddr *)&addr, attr->index,
+ &macsec_device->tx_rules_list,
+ &macsec_device->rx_rules_list, dev->mdev->macsec_fs);
+ if (ret && !IS_ERR(physical_gid))
+ goto rule_err;
- return 0;
+ mlx5_macsec_save_roce_gid(macsec_device, (struct sockaddr *)&addr, attr->index);
+
+ dev_put(ndev);
+ mutex_unlock(&dev->macsec.lock);
+ return ret;
+rule_err:
+ set_roce_addr(to_mdev(physical_gid->device), physical_gid->port_num,
+ physical_gid->index, &physical_gid->gid, physical_gid);
+ mgids->macsec_index = -1;
gid_err:
rdma_put_gid_attr(physical_gid);
+ cleanup_macsec_device(macsec_device);
+dev_err:
+ dev_put(ndev);
+ mutex_unlock(&dev->macsec.lock);
return ret;
}
void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr)
{
struct mlx5_ib_dev *dev = to_mdev(attr->device);
+ struct mlx5_macsec_device *macsec_device;
struct mlx5_reserved_gids *mgids;
struct net_device *ndev;
int i, max_gids;
@@ -134,8 +333,10 @@ void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr)
rcu_read_unlock();
return;
}
+ dev_hold(ndev);
rcu_read_unlock();
+ mutex_lock(&dev->macsec.lock);
max_gids = MLX5_CAP_ROCE(dev->mdev, roce_address_table_size);
for (i = 0; i < max_gids; i++) { /* Checking if macsec gid has ambiguous IP */
mgids = &dev->port[attr->port_num - 1].reserved_gids[i];
@@ -152,4 +353,12 @@ void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr)
break;
}
}
+ macsec_device = get_macsec_device(ndev, &dev->macsec.macsec_devices_list);
+ mlx5_macsec_del_roce_rule(attr->index, dev->mdev->macsec_fs,
+ &macsec_device->tx_rules_list, &macsec_device->rx_rules_list);
+ mlx5_macsec_del_roce_gid(macsec_device, attr->index);
+ cleanup_macsec_device(macsec_device);
+
+ dev_put(ndev);
+ mutex_unlock(&dev->macsec.lock);
}
diff --git a/drivers/infiniband/hw/mlx5/macsec.h b/drivers/infiniband/hw/mlx5/macsec.h
index b60f8f046d6f..9b77ba90f0f4 100644
--- a/drivers/infiniband/hw/mlx5/macsec.h
+++ b/drivers/infiniband/hw/mlx5/macsec.h
@@ -14,12 +14,16 @@ struct mlx5_reserved_gids;
int mlx5r_add_gid_macsec_operations(const struct ib_gid_attr *attr);
void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr);
-int mlx5r_macsec_alloc_gids(struct mlx5_ib_dev *dev);
+int mlx5r_macsec_init_gids_and_devlist(struct mlx5_ib_dev *dev);
void mlx5r_macsec_dealloc_gids(struct mlx5_ib_dev *dev);
+void mlx5r_macsec_event_register(struct mlx5_ib_dev *dev);
+void mlx5r_macsec_event_unregister(struct mlx5_ib_dev *dev);
#else
static inline int mlx5r_add_gid_macsec_operations(const struct ib_gid_attr *attr) { return 0; }
static inline void mlx5r_del_gid_macsec_operations(const struct ib_gid_attr *attr) {}
-static inline int mlx5r_macsec_alloc_gids(struct mlx5_ib_dev *dev) { return 0; }
+static inline int mlx5r_macsec_init_gids_and_devlist(struct mlx5_ib_dev *dev) { return 0; }
static inline void mlx5r_macsec_dealloc_gids(struct mlx5_ib_dev *dev) {}
+static inline void mlx5r_macsec_event_register(struct mlx5_ib_dev *dev) {}
+static inline void mlx5r_macsec_event_unregister(struct mlx5_ib_dev *dev) {}
#endif
#endif /* __MLX5_MACSEC_H__ */
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index f463cf8b7501..b4cc16a19757 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -3684,7 +3684,7 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev)
if (err)
return err;
- err = mlx5r_macsec_alloc_gids(dev);
+ err = mlx5r_macsec_init_gids_and_devlist(dev);
if (err)
return err;
@@ -4125,11 +4125,15 @@ static int mlx5_ib_stage_dev_notifier_init(struct mlx5_ib_dev *dev)
{
dev->mdev_events.notifier_call = mlx5_ib_event;
mlx5_notifier_register(dev->mdev, &dev->mdev_events);
+
+ mlx5r_macsec_event_register(dev);
+
return 0;
}
static void mlx5_ib_stage_dev_notifier_cleanup(struct mlx5_ib_dev *dev)
{
+ mlx5r_macsec_event_unregister(dev);
mlx5_notifier_unregister(dev->mdev, &dev->mdev_events);
}
diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h
index a4b940b5035f..16713baf0d06 100644
--- a/drivers/infiniband/hw/mlx5/mlx5_ib.h
+++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h
@@ -1090,6 +1090,12 @@ struct mlx5_special_mkeys {
__be32 terminate_scatter_list_mkey;
};
+struct mlx5_macsec {
+ struct mutex lock; /* Protects mlx5_macsec internal contexts */
+ struct list_head macsec_devices_list;
+ struct notifier_block blocking_events_nb;
+};
+
struct mlx5_ib_dev {
struct ib_device ib_dev;
struct mlx5_core_dev *mdev;
@@ -1149,6 +1155,10 @@ struct mlx5_ib_dev {
u16 pkey_table_len;
u8 lag_ports;
struct mlx5_special_mkeys mkeys;
+
+#ifdef CONFIG_MLX5_MACSEC
+ struct mlx5_macsec macsec;
+#endif
};
static inline struct mlx5_ib_cq *to_mibcq(struct mlx5_core_cq *mcq)
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 848993823ba5..38412457c617 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -1328,6 +1328,7 @@ static inline bool mlx5_get_roce_state(struct mlx5_core_dev *dev)
return mlx5_is_roce_on(dev);
}
+#ifdef CONFIG_MLX5_MACSEC
static inline bool mlx5e_is_macsec_device(const struct mlx5_core_dev *mdev)
{
if (!(MLX5_CAP_GEN_64(mdev, general_obj_types) &
@@ -1366,11 +1367,12 @@ static inline bool mlx5_is_macsec_roce_supported(struct mlx5_core_dev *mdev)
if (((MLX5_CAP_GEN_2(mdev, flow_table_type_2_type) &
NIC_RDMA_BOTH_DIRS_CAPS) != NIC_RDMA_BOTH_DIRS_CAPS) ||
!MLX5_CAP_FLOWTABLE_RDMA_TX(mdev, max_modify_header_actions) ||
- !mlx5e_is_macsec_device(mdev))
+ !mlx5e_is_macsec_device(mdev) || !mdev->macsec_fs)
return false;
return true;
}
+#endif
enum {
MLX5_OCTWORD = 16,
--
2.41.0
^ permalink raw reply related [flat|nested] 18+ messages in thread* Re: [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support
2023-08-09 8:29 [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Leon Romanovsky
` (13 preceding siblings ...)
2023-08-09 8:29 ` [PATCH mlx5-next v1 14/14] RDMA/mlx5: Handles RoCE MACsec steering rules addition and deletion Leon Romanovsky
@ 2023-08-09 23:09 ` Jakub Kicinski
2023-08-09 23:10 ` Jakub Kicinski
2023-08-09 23:53 ` Jason Gunthorpe
14 siblings, 2 replies; 18+ messages in thread
From: Jakub Kicinski @ 2023-08-09 23:09 UTC (permalink / raw)
To: Leon Romanovsky
Cc: Jason Gunthorpe, Leon Romanovsky, David S . Miller, Eric Dumazet,
linux-rdma, Maor Gottlieb, Mark Zhang, netdev, Paolo Abeni,
Patrisious Haddad, Raed Salem, Saeed Mahameed, Simon Horman
On Wed, 9 Aug 2023 11:29:12 +0300 Leon Romanovsky wrote:
> This series extends previously added MACsec offload support
> to cover RoCE traffic either.
>
> In order to achieve that, we need configure MACsec with offload between
> the two endpoints, like below:
>
> REMOTE_MAC=10:70:fd:43:71:c0
>
> * ip addr add 1.1.1.1/16 dev eth2
> * ip link set dev eth2 up
> * ip link add link eth2 macsec0 type macsec encrypt on
> * ip macsec offload macsec0 mac
> * ip macsec add macsec0 tx sa 0 pn 1 on key 00 dffafc8d7b9a43d5b9a3dfbbf6a30c16
> * ip macsec add macsec0 rx port 1 address $REMOTE_MAC
> * ip macsec add macsec0 rx port 1 address $REMOTE_MAC sa 0 pn 1 on key 01 ead3664f508eb06c40ac7104cdae4ce5
> * ip addr add 10.1.0.1/16 dev macsec0
> * ip link set dev macsec0 up
>
> And in a similar manner on the other machine, while noting the keys order
> would be reversed and the MAC address of the other machine.
>
> RDMA traffic is separated through relevant GID entries and in case of IP ambiguity
> issue - meaning we have a physical GIDs and a MACsec GIDs with the same IP/GID, we
> disable our physical GID in order to force the user to only use the MACsec GID.
Can you explain why you need special code to handle this?
MACsec is L2, RDMA is L4.
^ permalink raw reply [flat|nested] 18+ messages in thread* Re: [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support
2023-08-09 23:09 ` [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Jakub Kicinski
@ 2023-08-09 23:10 ` Jakub Kicinski
2023-08-09 23:53 ` Jason Gunthorpe
1 sibling, 0 replies; 18+ messages in thread
From: Jakub Kicinski @ 2023-08-09 23:10 UTC (permalink / raw)
To: Leon Romanovsky
Cc: Jason Gunthorpe, Leon Romanovsky, David S . Miller, Eric Dumazet,
linux-rdma, Maor Gottlieb, Mark Zhang, netdev, Paolo Abeni,
Patrisious Haddad, Raed Salem, Saeed Mahameed, Simon Horman
On Wed, 9 Aug 2023 16:09:45 -0700 Jakub Kicinski wrote:
> On Wed, 9 Aug 2023 11:29:12 +0300 Leon Romanovsky wrote:
> > This series extends previously added MACsec offload support
> > to cover RoCE traffic either.
> >
> > In order to achieve that, we need configure MACsec with offload between
> > the two endpoints, like below:
> >
> > REMOTE_MAC=10:70:fd:43:71:c0
> >
> > * ip addr add 1.1.1.1/16 dev eth2
> > * ip link set dev eth2 up
> > * ip link add link eth2 macsec0 type macsec encrypt on
> > * ip macsec offload macsec0 mac
> > * ip macsec add macsec0 tx sa 0 pn 1 on key 00 dffafc8d7b9a43d5b9a3dfbbf6a30c16
> > * ip macsec add macsec0 rx port 1 address $REMOTE_MAC
> > * ip macsec add macsec0 rx port 1 address $REMOTE_MAC sa 0 pn 1 on key 01 ead3664f508eb06c40ac7104cdae4ce5
> > * ip addr add 10.1.0.1/16 dev macsec0
> > * ip link set dev macsec0 up
> >
> > And in a similar manner on the other machine, while noting the keys order
> > would be reversed and the MAC address of the other machine.
> >
> > RDMA traffic is separated through relevant GID entries and in case of IP ambiguity
> > issue - meaning we have a physical GIDs and a MACsec GIDs with the same IP/GID, we
> > disable our physical GID in order to force the user to only use the MACsec GID.
>
> Can you explain why you need special code to handle this?
> MACsec is L2, RDMA is L4.
Ah, because you need to support "offload" on device that's not yours.
^ permalink raw reply [flat|nested] 18+ messages in thread
* Re: [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support
2023-08-09 23:09 ` [PATCH mlx5-next v1 00/14] mlx5 MACsec RoCEv2 support Jakub Kicinski
2023-08-09 23:10 ` Jakub Kicinski
@ 2023-08-09 23:53 ` Jason Gunthorpe
1 sibling, 0 replies; 18+ messages in thread
From: Jason Gunthorpe @ 2023-08-09 23:53 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Leon Romanovsky, Leon Romanovsky, David S . Miller, Eric Dumazet,
linux-rdma, Maor Gottlieb, Mark Zhang, netdev, Paolo Abeni,
Patrisious Haddad, Raed Salem, Saeed Mahameed, Simon Horman
On Wed, Aug 09, 2023 at 04:09:45PM -0700, Jakub Kicinski wrote:
> On Wed, 9 Aug 2023 11:29:12 +0300 Leon Romanovsky wrote:
> > This series extends previously added MACsec offload support
> > to cover RoCE traffic either.
> >
> > In order to achieve that, we need configure MACsec with offload between
> > the two endpoints, like below:
> >
> > REMOTE_MAC=10:70:fd:43:71:c0
> >
> > * ip addr add 1.1.1.1/16 dev eth2
> > * ip link set dev eth2 up
> > * ip link add link eth2 macsec0 type macsec encrypt on
> > * ip macsec offload macsec0 mac
> > * ip macsec add macsec0 tx sa 0 pn 1 on key 00 dffafc8d7b9a43d5b9a3dfbbf6a30c16
> > * ip macsec add macsec0 rx port 1 address $REMOTE_MAC
> > * ip macsec add macsec0 rx port 1 address $REMOTE_MAC sa 0 pn 1 on key 01 ead3664f508eb06c40ac7104cdae4ce5
> > * ip addr add 10.1.0.1/16 dev macsec0
> > * ip link set dev macsec0 up
> >
> > And in a similar manner on the other machine, while noting the keys order
> > would be reversed and the MAC address of the other machine.
> >
> > RDMA traffic is separated through relevant GID entries and in case of IP ambiguity
> > issue - meaning we have a physical GIDs and a MACsec GIDs with the same IP/GID, we
> > disable our physical GID in order to force the user to only use the MACsec GID.
>
> Can you explain why you need special code to handle this?
> MACsec is L2, RDMA is L4.
It is similar in concept to how TCP validates the 5 tuple, and
optionally Rx netdev of every packet.
The RDMA Rx layer uses a hardware handle for the entire L2/L3 path
called the 'GID Index'. It has the interface IP, VLAN information and
so on. Logically for MACSEC the 'GID Index' should include the L2
MADSEC association too.
For security, at the L4 layer, the RDMA engines enforce that queues
can only process packets that are matching the correct 'GID Index'. Ie
you cannot Rx a packet on VLAN 10 and have it be delivered to a queue
that thinks it is working on VLAN 11. Or Rx on IP A to a queue working
on IP B. Same basic principle for MADSEC, rx unencrypted/wrong SA
cannot be delivered to a queue that is expecting a certain SA.
This HW generation is not able to directly associate the MADSEC L2
information with the GID index. Thus we cannot create distinct GID
Indexes for the same IP, same VLAN but differing in MADSEC.
Thus if there is an attempt to create two identical GID indexes the
driver will prioritize the encrypted one under the idea that is the
more secure option in case of misconfiguration.
Jason
^ permalink raw reply [flat|nested] 18+ messages in thread