[pull request][net-next 00/16] mlx5 updates 2021-04-02

[pull request][net-next 00/16] mlx5 updates 2021-04-02

[Saeed Mahameed]

From: Saeed Mahameed <saeedm@nvidia.com>

Hi Dave, Jakub,

This series provfides misc updates to mlx5.
For more information please see tag log below.

Please pull and let me know if there is any problem.

Thanks,
Saeed.

---
The following changes since commit bd78980be1a68d14524c51c4b4170782fada622b:

  net: usb: ax88179_178a: initialize local variables before use (2021-04-01 16:09:37 -0700)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-updates-2021-04-02

for you to fetch changes up to 6783f0a21a3ce60e2ac9e56814afcbb01f98cc2e:

  net/mlx5e: Dynamic alloc vlan table for netdev when needed (2021-04-02 16:13:08 -0700)

----------------------------------------------------------------
mlx5-updates-2021-04-02

This series provides trivial updates and cleanup to mlx5 driver

1) Support for matching on ct_state inv and rel flag in connection tracking
2) Reject TC rules that redirect from a VF to itself
3) Parav provided some E-Switch cleanups that could be summarized to:
  3.1) Packing and Reduce structure sizes
  3.2) Dynamic allocation of rate limit tables and structures
4) Vu Makes the netdev arfs and vlan tables allocation dynamic.

----------------------------------------------------------------
Ariel Levkovich (2):
      net/mlx5: CT: Add support for matching on ct_state inv and rel flags
      net/mlx5e: Reject tc rules which redirect from a VF to itself

Parav Pandit (11):
      net/mlx5: E-Switch, cut down mlx5_vport_info structure size by 8 bytes
      net/mlx5: E-Switch, move QoS specific fields to existing qos struct
      net/mlx5: Use unsigned int for free_count
      net/mlx5: Pack mlx5_rl_entry structure
      net/mlx5: Do not hold mutex while reading table constants
      net/mlx5: Use helpers to allocate and free rl table entries
      net/mlx5: Use helper to increment, decrement rate entry refcount
      net/mlx5: Allocate rate limit table when rate is configured
      net/mlx5: Pair mutex_destory with mutex_init for rate limit table
      net/mlx5: E-Switch, cut down mlx5_vport_info structure size by 8 bytes
      net/mlx5: E-Switch, move QoS specific fields to existing qos struct

Roi Dayan (1):
      net/mlx5: Use ida_alloc_range() instead of ida_simple_alloc()

Vu Pham (2):
      net/mlx5e: Dynamic alloc arfs table for netdev when needed
      net/mlx5e: Dynamic alloc vlan table for netdev when needed

 drivers/net/ethernet/mellanox/mlx5/core/en/fs.h    |  46 ++-----
 drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c |  26 +++-
 drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c  |  95 +++++++++-----
 drivers/net/ethernet/mellanox/mlx5/core/en_fs.c    | 132 +++++++++++--------
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c  |   3 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c    |  17 ++-
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.c  |  26 ++--
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h  |  10 +-
 .../net/ethernet/mellanox/mlx5/core/fpga/ipsec.c   |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c  |   6 +-
 drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c  |  10 +-
 .../net/ethernet/mellanox/mlx5/core/pagealloc.c    |   2 +-
 drivers/net/ethernet/mellanox/mlx5/core/rl.c       | 139 ++++++++++++++-------
 include/linux/mlx5/driver.h                        |   3 +-
 14 files changed, 323 insertions(+), 198 deletions(-)

[net-next 01/16] net/mlx5: CT: Add support for matching on ct_state inv and rel flags

[Saeed Mahameed]

From: Ariel Levkovich <lariel@nvidia.com>

Add support for matching on ct_state inv and rel flags.

Currently the support is only for match on -inv and -rel.
Matching on +inv and +rel will be rejected.

Example:
$ tc filter add dev ens1f0_0 ingress prio 1 chain 1 proto ip flower \
  ct_state -est-rel+trk \
  action mirred egress redirect dev ens1f0_1
$ tc filter add dev ens1f0_1 ingress prio 1 chain 1 proto ip flower \
  ct_state +trk+est-inv \
  action mirred egress redirect dev ens1f0_0

Signed-off-by: Ariel Levkovich <lariel@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../ethernet/mellanox/mlx5/core/en/tc_ct.c    | 26 +++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
index df13e5094034..1c44000ad675 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c
@@ -29,6 +29,8 @@
 #define MLX5_CT_STATE_TRK_BIT BIT(2)
 #define MLX5_CT_STATE_NAT_BIT BIT(3)
 #define MLX5_CT_STATE_REPLY_BIT BIT(4)
+#define MLX5_CT_STATE_RELATED_BIT BIT(5)
+#define MLX5_CT_STATE_INVALID_BIT BIT(6)
 
 #define MLX5_FTE_ID_BITS (mlx5e_tc_attr_to_reg_mappings[FTEID_TO_REG].mlen * 8)
 #define MLX5_FTE_ID_MAX GENMASK(MLX5_FTE_ID_BITS - 1, 0)
@@ -1207,8 +1209,8 @@ mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv,
 		     struct mlx5_ct_attr *ct_attr,
 		     struct netlink_ext_ack *extack)
 {
+	bool trk, est, untrk, unest, new, rpl, unrpl, rel, unrel, inv, uninv;
 	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
-	bool trk, est, untrk, unest, new, rpl, unrpl;
 	struct flow_dissector_key_ct *mask, *key;
 	u32 ctstate = 0, ctstate_mask = 0;
 	u16 ct_state_on, ct_state_off;
@@ -1236,7 +1238,9 @@ mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv,
 	if (ct_state_mask & ~(TCA_FLOWER_KEY_CT_FLAGS_TRACKED |
 			      TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED |
 			      TCA_FLOWER_KEY_CT_FLAGS_NEW |
-			      TCA_FLOWER_KEY_CT_FLAGS_REPLY)) {
+			      TCA_FLOWER_KEY_CT_FLAGS_REPLY |
+			      TCA_FLOWER_KEY_CT_FLAGS_RELATED |
+			      TCA_FLOWER_KEY_CT_FLAGS_INVALID)) {
 		NL_SET_ERR_MSG_MOD(extack,
 				   "only ct_state trk, est, new and rpl are supported for offload");
 		return -EOPNOTSUPP;
@@ -1248,9 +1252,13 @@ mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv,
 	new = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_NEW;
 	est = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED;
 	rpl = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_REPLY;
+	rel = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_RELATED;
+	inv = ct_state_on & TCA_FLOWER_KEY_CT_FLAGS_INVALID;
 	untrk = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_TRACKED;
 	unest = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_ESTABLISHED;
 	unrpl = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_REPLY;
+	unrel = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_RELATED;
+	uninv = ct_state_off & TCA_FLOWER_KEY_CT_FLAGS_INVALID;
 
 	ctstate |= trk ? MLX5_CT_STATE_TRK_BIT : 0;
 	ctstate |= est ? MLX5_CT_STATE_ESTABLISHED_BIT : 0;
@@ -1258,6 +1266,20 @@ mlx5_tc_ct_match_add(struct mlx5_tc_ct_priv *priv,
 	ctstate_mask |= (untrk || trk) ? MLX5_CT_STATE_TRK_BIT : 0;
 	ctstate_mask |= (unest || est) ? MLX5_CT_STATE_ESTABLISHED_BIT : 0;
 	ctstate_mask |= (unrpl || rpl) ? MLX5_CT_STATE_REPLY_BIT : 0;
+	ctstate_mask |= unrel ? MLX5_CT_STATE_RELATED_BIT : 0;
+	ctstate_mask |= uninv ? MLX5_CT_STATE_INVALID_BIT : 0;
+
+	if (rel) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "matching on ct_state +rel isn't supported");
+		return -EOPNOTSUPP;
+	}
+
+	if (inv) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "matching on ct_state +inv isn't supported");
+		return -EOPNOTSUPP;
+	}
 
 	if (new) {
 		NL_SET_ERR_MSG_MOD(extack,
-- 
2.30.2

[net-next 02/16] net/mlx5: E-Switch, cut down mlx5_vport_info structure size by 8 bytes

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Structure mlx5_vport_info consumes 40 bytes of space due to a hole
in it. After packing it reduces to 32 bytes.

Currently:
pahole -C mlx5_vport_info drivers/net/ethernet/mellanox/mlx5/core/eswitch.o
struct mlx5_vport_info {
        u8                         mac[6];               /*     0     6 */
        u16                        vlan;                 /*     6     2 */
        u8                         qos;                  /*     8     1 */

        /* XXX 7 bytes hole, try to pack */

        u64                        node_guid;            /*    16     8 */
        int                        link_state;           /*    24     4 */
        u32                        min_rate;             /*    28     4 */
        u32                        max_rate;             /*    32     4 */
        bool                       spoofchk;             /*    36     1 */
        bool                       trusted;              /*    37     1 */

        /* size: 40, cachelines: 1, members: 9 */
        /* sum members: 31, holes: 1, sum holes: 7 */
        /* padding: 2 */
        /* last cacheline: 40 bytes */
};

After packing:

$ pahole -C mlx5_vport_info drivers/net/ethernet/mellanox/mlx5/core/eswitch.o

struct mlx5_vport_info {
        u8                         mac[6];               /*     0     6 */
        u16                        vlan;                 /*     6     2 */
        u64                        node_guid;            /*     8     8 */
        int                        link_state;           /*    16     4 */
        u32                        min_rate;             /*    20     4 */
        u32                        max_rate;             /*    24     4 */
        u8                         qos;                  /*    28     1 */
        u8                         spoofchk:1;           /*    29: 0  1 */
        u8                         trusted:1;            /*    29: 1  1 */

        /* size: 32, cachelines: 1, members: 9 */
        /* padding: 2 */
        /* bit_padding: 6 bits */
        /* last cacheline: 32 bytes */
};

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 56d85cedb9bd..0adffab20244 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -118,13 +118,13 @@ struct mlx5_vport_drop_stats {
 struct mlx5_vport_info {
 	u8                      mac[ETH_ALEN];
 	u16                     vlan;
-	u8                      qos;
 	u64                     node_guid;
 	int                     link_state;
 	u32                     min_rate;
 	u32                     max_rate;
-	bool                    spoofchk;
-	bool                    trusted;
+	u8                      qos;
+	u8                      spoofchk: 1;
+	u8                      trusted: 1;
 };
 
 /* Vport context events */
-- 
2.30.2

[net-next 03/16] net/mlx5: E-Switch, move QoS specific fields to existing qos struct

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Function QoS related fields are already defined in qos related struct.
min and max rate are left out to mlx5_vport_info struct.

Move them to existing qos struct.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/eswitch.c | 26 +++++++++----------
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  4 +--
 2 files changed, 15 insertions(+), 15 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index 6b260dacf853..6cf04a366f99 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -1235,7 +1235,7 @@ static int esw_vport_setup(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
 		return err;
 
 	/* Attach vport to the eswitch rate limiter */
-	esw_vport_enable_qos(esw, vport, vport->info.max_rate, vport->qos.bw_share);
+	esw_vport_enable_qos(esw, vport, vport->qos.max_rate, vport->qos.bw_share);
 
 	if (mlx5_esw_is_manager_vport(esw, vport_num))
 		return 0;
@@ -2078,8 +2078,8 @@ int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw,
 	ivi->qos = evport->info.qos;
 	ivi->spoofchk = evport->info.spoofchk;
 	ivi->trusted = evport->info.trusted;
-	ivi->min_tx_rate = evport->info.min_rate;
-	ivi->max_tx_rate = evport->info.max_rate;
+	ivi->min_tx_rate = evport->qos.min_rate;
+	ivi->max_tx_rate = evport->qos.max_rate;
 	mutex_unlock(&esw->state_lock);
 
 	return 0;
@@ -2319,9 +2319,9 @@ static u32 calculate_vports_min_rate_divider(struct mlx5_eswitch *esw)
 	int i;
 
 	mlx5_esw_for_all_vports(esw, i, evport) {
-		if (!evport->enabled || evport->info.min_rate < max_guarantee)
+		if (!evport->enabled || evport->qos.min_rate < max_guarantee)
 			continue;
-		max_guarantee = evport->info.min_rate;
+		max_guarantee = evport->qos.min_rate;
 	}
 
 	if (max_guarantee)
@@ -2343,8 +2343,8 @@ static int normalize_vports_min_rate(struct mlx5_eswitch *esw)
 	mlx5_esw_for_all_vports(esw, i, evport) {
 		if (!evport->enabled)
 			continue;
-		vport_min_rate = evport->info.min_rate;
-		vport_max_rate = evport->info.max_rate;
+		vport_min_rate = evport->qos.min_rate;
+		vport_max_rate = evport->qos.max_rate;
 		bw_share = 0;
 
 		if (divider)
@@ -2391,24 +2391,24 @@ int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw, u16 vport,
 
 	mutex_lock(&esw->state_lock);
 
-	if (min_rate == evport->info.min_rate)
+	if (min_rate == evport->qos.min_rate)
 		goto set_max_rate;
 
-	previous_min_rate = evport->info.min_rate;
-	evport->info.min_rate = min_rate;
+	previous_min_rate = evport->qos.min_rate;
+	evport->qos.min_rate = min_rate;
 	err = normalize_vports_min_rate(esw);
 	if (err) {
-		evport->info.min_rate = previous_min_rate;
+		evport->qos.min_rate = previous_min_rate;
 		goto unlock;
 	}
 
 set_max_rate:
-	if (max_rate == evport->info.max_rate)
+	if (max_rate == evport->qos.max_rate)
 		goto unlock;
 
 	err = esw_vport_qos_config(esw, evport, max_rate, evport->qos.bw_share);
 	if (!err)
-		evport->info.max_rate = max_rate;
+		evport->qos.max_rate = max_rate;
 
 unlock:
 	mutex_unlock(&esw->state_lock);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 0adffab20244..64db903068c1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -120,8 +120,6 @@ struct mlx5_vport_info {
 	u16                     vlan;
 	u64                     node_guid;
 	int                     link_state;
-	u32                     min_rate;
-	u32                     max_rate;
 	u8                      qos;
 	u8                      spoofchk: 1;
 	u8                      trusted: 1;
@@ -154,6 +152,8 @@ struct mlx5_vport {
 		bool            enabled;
 		u32             esw_tsar_ix;
 		u32             bw_share;
+		u32 min_rate;
+		u32 max_rate;
 	} qos;
 
 	bool                    enabled;
-- 
2.30.2

[net-next 04/16] net/mlx5: Use unsigned int for free_count

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Fix the warning due to missing int.

WARNING: Prefer 'unsigned int' to bare use of 'unsigned'
+       unsigned free_count;

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
index c0656d4782e1..110c0837f95b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c
@@ -61,7 +61,7 @@ struct fw_page {
 	u32			function;
 	unsigned long		bitmask;
 	struct list_head	list;
-	unsigned		free_count;
+	unsigned int free_count;
 };
 
 enum {
-- 
2.30.2

[net-next 05/16] net/mlx5: Pack mlx5_rl_entry structure

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

mlx5_rl_entry structure is not properly packed as shown below. Due to this
an array of size 9144 bytes allocated which is aligned to 16Kbytes.
Hence, pack the structure and avoid the wastage.

This offers 8Kbytes of saving per mlx5_core_dev struct.

pahole -C mlx5_rl_entry  drivers/net/ethernet/mellanox/mlx5/core/en_main.o

Existing layout:

struct mlx5_rl_entry {
        u8                         rl_raw[48];           /*     0    48 */
        u16                        index;                /*    48     2 */

        /* XXX 6 bytes hole, try to pack */

        u64                        refcount;             /*    56     8 */
        /* --- cacheline 1 boundary (64 bytes) --- */
        u16                        uid;                  /*    64     2 */
        u8                         dedicated:1;          /*    66: 0  1 */

        /* size: 72, cachelines: 2, members: 5 */
        /* sum members: 60, holes: 1, sum holes: 6 */
        /* sum bitfield members: 1 bits (0 bytes) */
        /* padding: 5 */
        /* bit_padding: 7 bits */
        /* last cacheline: 8 bytes */
};

After alignment:

struct mlx5_rl_entry {
        u8                         rl_raw[48];           /*     0    48 */
        u64                        refcount;             /*    48     8 */
        u16                        index;                /*    56     2 */
        u16                        uid;                  /*    58     2 */
        u8                         dedicated:1;          /*    60: 0  1 */

        /* size: 64, cachelines: 1, members: 5 */
        /* padding: 3 */
        /* bit_padding: 7 bits */
};

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 include/linux/mlx5/driver.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 23bb01d7c9b9..a9bd7e3bd554 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -517,8 +517,8 @@ struct mlx5_rate_limit {
 
 struct mlx5_rl_entry {
 	u8 rl_raw[MLX5_ST_SZ_BYTES(set_pp_rate_limit_context)];
-	u16 index;
 	u64 refcount;
+	u16 index;
 	u16 uid;
 	u8 dedicated : 1;
 };
-- 
2.30.2

[net-next 06/16] net/mlx5: Do not hold mutex while reading table constants

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Table max_size, min and max rate are constants initialized while table
is created. Reading it doesn't need to hold a table mutex. Hence, read
them without holding table mutex.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/rl.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rl.c b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
index 99039c47ef33..2accc0f324f3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/rl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
@@ -180,16 +180,18 @@ int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 	int err = 0;
 	u32 rate;
 
-	rate = MLX5_GET(set_pp_rate_limit_context, rl_in, rate_limit);
-	mutex_lock(&table->rl_lock);
+	if (!table->max_size)
+		return -EOPNOTSUPP;
 
+	rate = MLX5_GET(set_pp_rate_limit_context, rl_in, rate_limit);
 	if (!rate || !mlx5_rl_is_in_range(dev, rate)) {
 		mlx5_core_err(dev, "Invalid rate: %u, should be %u to %u\n",
 			      rate, table->min_rate, table->max_rate);
-		err = -EINVAL;
-		goto out;
+		return -EINVAL;
 	}
 
+	mutex_lock(&table->rl_lock);
+
 	entry = find_rl_entry(table, rl_in, uid, dedicated_entry);
 	if (!entry) {
 		mlx5_core_err(dev, "Max number of %u rates reached\n",
-- 
2.30.2

[net-next 07/16] net/mlx5: Use helpers to allocate and free rl table entries

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

User helper routines to allocate and free rate limit table entries.
Subsequent patch extends use of these helpers to do allocation
during rate entry allocation callback.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/rl.c | 55 +++++++++++++-------
 1 file changed, 36 insertions(+), 19 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rl.c b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
index 2accc0f324f3..208fd3cad970 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/rl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
@@ -172,6 +172,35 @@ bool mlx5_rl_are_equal(struct mlx5_rate_limit *rl_0,
 }
 EXPORT_SYMBOL(mlx5_rl_are_equal);
 
+static int mlx5_rl_table_alloc(struct mlx5_rl_table *table)
+{
+	int i;
+
+	table->rl_entry = kcalloc(table->max_size, sizeof(struct mlx5_rl_entry),
+				  GFP_KERNEL);
+	if (!table->rl_entry)
+		return -ENOMEM;
+
+	/* The index represents the index in HW rate limit table
+	 * Index 0 is reserved for unlimited rate
+	 */
+	for (i = 0; i < table->max_size; i++)
+		table->rl_entry[i].index = i + 1;
+
+	return 0;
+}
+
+static void mlx5_rl_table_free(struct mlx5_core_dev *dev, struct mlx5_rl_table *table)
+{
+	int i;
+
+	/* Clear all configured rates */
+	for (i = 0; i < table->max_size; i++)
+		if (table->rl_entry[i].refcount)
+			mlx5_set_pp_rate_limit_cmd(dev, &table->rl_entry[i], false);
+	kfree(table->rl_entry);
+}
+
 int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 			 bool dedicated_entry, u16 *index)
 {
@@ -302,7 +331,7 @@ EXPORT_SYMBOL(mlx5_rl_remove_rate);
 int mlx5_init_rl_table(struct mlx5_core_dev *dev)
 {
 	struct mlx5_rl_table *table = &dev->priv.rl_table;
-	int i;
+	int err;
 
 	mutex_init(&table->rl_lock);
 	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, packet_pacing)) {
@@ -315,18 +344,10 @@ int mlx5_init_rl_table(struct mlx5_core_dev *dev)
 	table->max_rate = MLX5_CAP_QOS(dev, packet_pacing_max_rate);
 	table->min_rate = MLX5_CAP_QOS(dev, packet_pacing_min_rate);
 
-	table->rl_entry = kcalloc(table->max_size, sizeof(struct mlx5_rl_entry),
-				  GFP_KERNEL);
-	if (!table->rl_entry)
-		return -ENOMEM;
-
-	/* The index represents the index in HW rate limit table
-	 * Index 0 is reserved for unlimited rate
-	 */
-	for (i = 0; i < table->max_size; i++)
-		table->rl_entry[i].index = i + 1;
+	err = mlx5_rl_table_alloc(table);
+	if (err)
+		return err;
 
-	/* Index 0 is reserved */
 	mlx5_core_info(dev, "Rate limit: %u rates are supported, range: %uMbps to %uMbps\n",
 		       table->max_size,
 		       table->min_rate >> 10,
@@ -338,13 +359,9 @@ int mlx5_init_rl_table(struct mlx5_core_dev *dev)
 void mlx5_cleanup_rl_table(struct mlx5_core_dev *dev)
 {
 	struct mlx5_rl_table *table = &dev->priv.rl_table;
-	int i;
 
-	/* Clear all configured rates */
-	for (i = 0; i < table->max_size; i++)
-		if (table->rl_entry[i].refcount)
-			mlx5_set_pp_rate_limit_cmd(dev, &table->rl_entry[i],
-						   false);
+	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, packet_pacing))
+		return;
 
-	kfree(dev->priv.rl_table.rl_entry);
+	mlx5_rl_table_free(dev, table);
 }
-- 
2.30.2

[net-next 08/16] net/mlx5: Use helper to increment, decrement rate entry refcount

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Rate limit entry refcount can be incremented uniformly when it is newly
allocated or reused.
So simplify the code to increment refcount at one place.

Use decrement refcount helper in two routines.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/rl.c | 34 +++++++++++---------
 1 file changed, 18 insertions(+), 16 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rl.c b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
index 208fd3cad970..08792fe701e3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/rl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
@@ -201,6 +201,19 @@ static void mlx5_rl_table_free(struct mlx5_core_dev *dev, struct mlx5_rl_table *
 	kfree(table->rl_entry);
 }
 
+static void mlx5_rl_entry_get(struct mlx5_rl_entry *entry)
+{
+	entry->refcount++;
+}
+
+static void
+mlx5_rl_entry_put(struct mlx5_core_dev *dev, struct mlx5_rl_entry *entry)
+{
+	entry->refcount--;
+	if (!entry->refcount)
+		mlx5_set_pp_rate_limit_cmd(dev, entry, false);
+}
+
 int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 			 bool dedicated_entry, u16 *index)
 {
@@ -228,13 +241,10 @@ int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 		err = -ENOSPC;
 		goto out;
 	}
-	if (entry->refcount) {
-		/* rate already configured */
-		entry->refcount++;
-	} else {
+	if (!entry->refcount) {
+		/* new rate limit */
 		memcpy(entry->rl_raw, rl_in, sizeof(entry->rl_raw));
 		entry->uid = uid;
-		/* new rate limit */
 		err = mlx5_set_pp_rate_limit_cmd(dev, entry, true);
 		if (err) {
 			mlx5_core_err(
@@ -248,9 +258,9 @@ int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 			goto out;
 		}
 
-		entry->refcount = 1;
 		entry->dedicated = dedicated_entry;
 	}
+	mlx5_rl_entry_get(entry);
 	*index = entry->index;
 
 out:
@@ -266,10 +276,7 @@ void mlx5_rl_remove_rate_raw(struct mlx5_core_dev *dev, u16 index)
 
 	mutex_lock(&table->rl_lock);
 	entry = &table->rl_entry[index - 1];
-	entry->refcount--;
-	if (!entry->refcount)
-		/* need to remove rate */
-		mlx5_set_pp_rate_limit_cmd(dev, entry, false);
+	mlx5_rl_entry_put(dev, entry);
 	mutex_unlock(&table->rl_lock);
 }
 EXPORT_SYMBOL(mlx5_rl_remove_rate_raw);
@@ -317,12 +324,7 @@ void mlx5_rl_remove_rate(struct mlx5_core_dev *dev, struct mlx5_rate_limit *rl)
 			       rl->rate, rl->max_burst_sz, rl->typical_pkt_sz);
 		goto out;
 	}
-
-	entry->refcount--;
-	if (!entry->refcount)
-		/* need to remove rate */
-		mlx5_set_pp_rate_limit_cmd(dev, entry, false);
-
+	mlx5_rl_entry_put(dev, entry);
 out:
 	mutex_unlock(&table->rl_lock);
 }
-- 
2.30.2

[net-next 09/16] net/mlx5: Allocate rate limit table when rate is configured

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

A device supports 128 rate limiters. A static table allocation consumes
8KB of memory even when rate is not configured.

Instead, allocate the table when at least one rate is configured.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/rl.c | 46 ++++++++++++++++----
 include/linux/mlx5/driver.h                  |  1 +
 2 files changed, 38 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rl.c b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
index 08792fe701e3..0526e3798c09 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/rl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
@@ -117,6 +117,9 @@ static struct mlx5_rl_entry *find_rl_entry(struct mlx5_rl_table *table,
 	bool empty_found = false;
 	int i;
 
+	lockdep_assert_held(&table->rl_lock);
+	WARN_ON(!table->rl_entry);
+
 	for (i = 0; i < table->max_size; i++) {
 		if (dedicated) {
 			if (!table->rl_entry[i].refcount)
@@ -172,10 +175,17 @@ bool mlx5_rl_are_equal(struct mlx5_rate_limit *rl_0,
 }
 EXPORT_SYMBOL(mlx5_rl_are_equal);
 
-static int mlx5_rl_table_alloc(struct mlx5_rl_table *table)
+static int mlx5_rl_table_get(struct mlx5_rl_table *table)
 {
 	int i;
 
+	lockdep_assert_held(&table->rl_lock);
+
+	if (table->rl_entry) {
+		table->refcount++;
+		return 0;
+	}
+
 	table->rl_entry = kcalloc(table->max_size, sizeof(struct mlx5_rl_entry),
 				  GFP_KERNEL);
 	if (!table->rl_entry)
@@ -187,13 +197,27 @@ static int mlx5_rl_table_alloc(struct mlx5_rl_table *table)
 	for (i = 0; i < table->max_size; i++)
 		table->rl_entry[i].index = i + 1;
 
+	table->refcount++;
 	return 0;
 }
 
+static void mlx5_rl_table_put(struct mlx5_rl_table *table)
+{
+	lockdep_assert_held(&table->rl_lock);
+	if (--table->refcount)
+		return;
+
+	kfree(table->rl_entry);
+	table->rl_entry = NULL;
+}
+
 static void mlx5_rl_table_free(struct mlx5_core_dev *dev, struct mlx5_rl_table *table)
 {
 	int i;
 
+	if (!table->rl_entry)
+		return;
+
 	/* Clear all configured rates */
 	for (i = 0; i < table->max_size; i++)
 		if (table->rl_entry[i].refcount)
@@ -219,8 +243,8 @@ int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 {
 	struct mlx5_rl_table *table = &dev->priv.rl_table;
 	struct mlx5_rl_entry *entry;
-	int err = 0;
 	u32 rate;
+	int err;
 
 	if (!table->max_size)
 		return -EOPNOTSUPP;
@@ -233,13 +257,16 @@ int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 	}
 
 	mutex_lock(&table->rl_lock);
+	err = mlx5_rl_table_get(table);
+	if (err)
+		goto out;
 
 	entry = find_rl_entry(table, rl_in, uid, dedicated_entry);
 	if (!entry) {
 		mlx5_core_err(dev, "Max number of %u rates reached\n",
 			      table->max_size);
 		err = -ENOSPC;
-		goto out;
+		goto rl_err;
 	}
 	if (!entry->refcount) {
 		/* new rate limit */
@@ -255,14 +282,18 @@ int mlx5_rl_add_rate_raw(struct mlx5_core_dev *dev, void *rl_in, u16 uid,
 					 burst_upper_bound),
 				MLX5_GET(set_pp_rate_limit_context, rl_in,
 					 typical_packet_size));
-			goto out;
+			goto rl_err;
 		}
 
 		entry->dedicated = dedicated_entry;
 	}
 	mlx5_rl_entry_get(entry);
 	*index = entry->index;
+	mutex_unlock(&table->rl_lock);
+	return 0;
 
+rl_err:
+	mlx5_rl_table_put(table);
 out:
 	mutex_unlock(&table->rl_lock);
 	return err;
@@ -277,6 +308,7 @@ void mlx5_rl_remove_rate_raw(struct mlx5_core_dev *dev, u16 index)
 	mutex_lock(&table->rl_lock);
 	entry = &table->rl_entry[index - 1];
 	mlx5_rl_entry_put(dev, entry);
+	mlx5_rl_table_put(table);
 	mutex_unlock(&table->rl_lock);
 }
 EXPORT_SYMBOL(mlx5_rl_remove_rate_raw);
@@ -325,6 +357,7 @@ void mlx5_rl_remove_rate(struct mlx5_core_dev *dev, struct mlx5_rate_limit *rl)
 		goto out;
 	}
 	mlx5_rl_entry_put(dev, entry);
+	mlx5_rl_table_put(table);
 out:
 	mutex_unlock(&table->rl_lock);
 }
@@ -333,7 +366,6 @@ EXPORT_SYMBOL(mlx5_rl_remove_rate);
 int mlx5_init_rl_table(struct mlx5_core_dev *dev)
 {
 	struct mlx5_rl_table *table = &dev->priv.rl_table;
-	int err;
 
 	mutex_init(&table->rl_lock);
 	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, packet_pacing)) {
@@ -346,10 +378,6 @@ int mlx5_init_rl_table(struct mlx5_core_dev *dev)
 	table->max_rate = MLX5_CAP_QOS(dev, packet_pacing_max_rate);
 	table->min_rate = MLX5_CAP_QOS(dev, packet_pacing_min_rate);
 
-	err = mlx5_rl_table_alloc(table);
-	if (err)
-		return err;
-
 	mlx5_core_info(dev, "Rate limit: %u rates are supported, range: %uMbps to %uMbps\n",
 		       table->max_size,
 		       table->min_rate >> 10,
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index a9bd7e3bd554..baf38b5a2a8c 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -530,6 +530,7 @@ struct mlx5_rl_table {
 	u32                     max_rate;
 	u32                     min_rate;
 	struct mlx5_rl_entry   *rl_entry;
+	u64 refcount;
 };
 
 struct mlx5_core_roce {
-- 
2.30.2

[net-next 10/16] net/mlx5: Pair mutex_destory with mutex_init for rate limit table

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Add missing mutex_destroy() to pair with mutex_init().

This should be done only when table is initialized, hence perform
mutex_init() only when table is initialized.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/rl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/rl.c b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
index 0526e3798c09..7161220afe30 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/rl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/rl.c
@@ -367,12 +367,13 @@ int mlx5_init_rl_table(struct mlx5_core_dev *dev)
 {
 	struct mlx5_rl_table *table = &dev->priv.rl_table;
 
-	mutex_init(&table->rl_lock);
 	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, packet_pacing)) {
 		table->max_size = 0;
 		return 0;
 	}
 
+	mutex_init(&table->rl_lock);
+
 	/* First entry is reserved for unlimited rate */
 	table->max_size = MLX5_CAP_QOS(dev, packet_pacing_rate_table_size) - 1;
 	table->max_rate = MLX5_CAP_QOS(dev, packet_pacing_max_rate);
@@ -394,4 +395,5 @@ void mlx5_cleanup_rl_table(struct mlx5_core_dev *dev)
 		return;
 
 	mlx5_rl_table_free(dev, table);
+	mutex_destroy(&table->rl_lock);
 }
-- 
2.30.2

[net-next 11/16] net/mlx5: E-Switch, cut down mlx5_vport_info structure size by 8 bytes

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Structure mlx5_vport_info consumes 40 bytes of space due to a hole
in it. After packing it reduces to 32 bytes.

Currently:
pahole -C mlx5_vport_info drivers/net/ethernet/mellanox/mlx5/core/eswitch.o
struct mlx5_vport_info {
        u8                         mac[6];               /*     0     6 */
        u16                        vlan;                 /*     6     2 */
        u8                         qos;                  /*     8     1 */

        /* XXX 7 bytes hole, try to pack */

        u64                        node_guid;            /*    16     8 */
        int                        link_state;           /*    24     4 */
        u32                        min_rate;             /*    28     4 */
        u32                        max_rate;             /*    32     4 */
        bool                       spoofchk;             /*    36     1 */
        bool                       trusted;              /*    37     1 */

        /* size: 40, cachelines: 1, members: 9 */
        /* sum members: 31, holes: 1, sum holes: 7 */
        /* padding: 2 */
        /* last cacheline: 40 bytes */
};

After packing:

$ pahole -C mlx5_vport_info drivers/net/ethernet/mellanox/mlx5/core/eswitch.o

struct mlx5_vport_info {
        u8                         mac[6];               /*     0     6 */
        u16                        vlan;                 /*     6     2 */
        u64                        node_guid;            /*     8     8 */
        int                        link_state;           /*    16     4 */
        u32                        min_rate;             /*    20     4 */
        u32                        max_rate;             /*    24     4 */
        u8                         qos;                  /*    28     1 */
        u8                         spoofchk:1;           /*    29: 0  1 */
        u8                         trusted:1;            /*    29: 1  1 */

        /* size: 32, cachelines: 1, members: 9 */
        /* padding: 2 */
        /* bit_padding: 6 bits */
        /* last cacheline: 32 bytes */
};

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 64db903068c1..5d50b127f81a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -120,6 +120,8 @@ struct mlx5_vport_info {
 	u16                     vlan;
 	u64                     node_guid;
 	int                     link_state;
+	u32                     min_rate;
+	u32                     max_rate;
 	u8                      qos;
 	u8                      spoofchk: 1;
 	u8                      trusted: 1;
-- 
2.30.2

[net-next 12/16] net/mlx5: E-Switch, move QoS specific fields to existing qos struct

[Saeed Mahameed]

From: Parav Pandit <parav@nvidia.com>

Function QoS related fields are already defined in qos related struct.
min and max rate are left out to mlx5_vport_info struct.

Move them to existing qos struct.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/eswitch.h | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 5d50b127f81a..64db903068c1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -120,8 +120,6 @@ struct mlx5_vport_info {
 	u16                     vlan;
 	u64                     node_guid;
 	int                     link_state;
-	u32                     min_rate;
-	u32                     max_rate;
 	u8                      qos;
 	u8                      spoofchk: 1;
 	u8                      trusted: 1;
-- 
2.30.2

[net-next 13/16] net/mlx5: Use ida_alloc_range() instead of ida_simple_alloc()

[Saeed Mahameed]

From: Roi Dayan <roid@nvidia.com>

ida_simple_alloc() and remove functions are deprecated.
Related change:
commit 3264ceec8f17 ("lib/idr.c: document that ida_simple_{get,remove}() are deprecated")

Signed-off-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c |  6 +++---
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c    |  6 +++---
 drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c    | 10 +++++-----
 3 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c
index 22bee4990232..d43a05e77f67 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/ipsec.c
@@ -707,7 +707,7 @@ static void *mlx5_fpga_ipsec_create_sa_ctx(struct mlx5_core_dev *mdev,
 	}
 
 	if (accel_xfrm->attrs.action == MLX5_ACCEL_ESP_ACTION_DECRYPT) {
-		err = ida_simple_get(&fipsec->halloc, 1, 0, GFP_KERNEL);
+		err = ida_alloc_min(&fipsec->halloc, 1, GFP_KERNEL);
 		if (err < 0) {
 			context = ERR_PTR(err);
 			goto exists;
@@ -758,7 +758,7 @@ static void *mlx5_fpga_ipsec_create_sa_ctx(struct mlx5_core_dev *mdev,
 unlock_hash:
 	mutex_unlock(&fipsec->sa_hash_lock);
 	if (accel_xfrm->attrs.action == MLX5_ACCEL_ESP_ACTION_DECRYPT)
-		ida_simple_remove(&fipsec->halloc, sa_ctx->sa_handle);
+		ida_free(&fipsec->halloc, sa_ctx->sa_handle);
 exists:
 	mutex_unlock(&fpga_xfrm->lock);
 	kfree(sa_ctx);
@@ -852,7 +852,7 @@ mlx5_fpga_ipsec_release_sa_ctx(struct mlx5_fpga_ipsec_sa_ctx *sa_ctx)
 
 	if (sa_ctx->fpga_xfrm->accel_xfrm.attrs.action &
 	    MLX5_ACCEL_ESP_ACTION_DECRYPT)
-		ida_simple_remove(&fipsec->halloc, sa_ctx->sa_handle);
+		ida_free(&fipsec->halloc, sa_ctx->sa_handle);
 
 	mutex_lock(&fipsec->sa_hash_lock);
 	WARN_ON(rhashtable_remove_fast(&fipsec->sa_hash, &sa_ctx->hash,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
index dbd910656574..0216bd63a42d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c
@@ -590,7 +590,7 @@ static void del_sw_fte(struct fs_node *node)
 				     &fte->hash,
 				     rhash_fte);
 	WARN_ON(err);
-	ida_simple_remove(&fg->fte_allocator, fte->index - fg->start_index);
+	ida_free(&fg->fte_allocator, fte->index - fg->start_index);
 	kmem_cache_free(steering->ftes_cache, fte);
 }
 
@@ -640,7 +640,7 @@ static int insert_fte(struct mlx5_flow_group *fg, struct fs_fte *fte)
 	int index;
 	int ret;
 
-	index = ida_simple_get(&fg->fte_allocator, 0, fg->max_ftes, GFP_KERNEL);
+	index = ida_alloc_max(&fg->fte_allocator, fg->max_ftes - 1, GFP_KERNEL);
 	if (index < 0)
 		return index;
 
@@ -656,7 +656,7 @@ static int insert_fte(struct mlx5_flow_group *fg, struct fs_fte *fte)
 	return 0;
 
 err_ida_remove:
-	ida_simple_remove(&fg->fte_allocator, index);
+	ida_free(&fg->fte_allocator, index);
 	return ret;
 }
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
index 97324d9d4f2a..3f9869c7e326 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/gid.c
@@ -88,12 +88,12 @@ void mlx5_core_unreserve_gids(struct mlx5_core_dev *dev, unsigned int count)
 int mlx5_core_reserved_gid_alloc(struct mlx5_core_dev *dev, int *gid_index)
 {
 	int end = dev->roce.reserved_gids.start +
-		  dev->roce.reserved_gids.count;
+		  dev->roce.reserved_gids.count - 1;
 	int index = 0;
 
-	index = ida_simple_get(&dev->roce.reserved_gids.ida,
-			       dev->roce.reserved_gids.start, end,
-			       GFP_KERNEL);
+	index = ida_alloc_range(&dev->roce.reserved_gids.ida,
+				dev->roce.reserved_gids.start, end,
+				GFP_KERNEL);
 	if (index < 0)
 		return index;
 
@@ -105,7 +105,7 @@ int mlx5_core_reserved_gid_alloc(struct mlx5_core_dev *dev, int *gid_index)
 void mlx5_core_reserved_gid_free(struct mlx5_core_dev *dev, int gid_index)
 {
 	mlx5_core_dbg(dev, "Freeing reserved GID %u\n", gid_index);
-	ida_simple_remove(&dev->roce.reserved_gids.ida, gid_index);
+	ida_free(&dev->roce.reserved_gids.ida, gid_index);
 }
 
 unsigned int mlx5_core_reserved_gids_count(struct mlx5_core_dev *dev)
-- 
2.30.2

[net-next 14/16] net/mlx5e: Reject tc rules which redirect from a VF to itself

[Saeed Mahameed]

From: Ariel Levkovich <lariel@nvidia.com>

Since there are self loopback prevention mechanisms at the
VF level, offloading such rules which redirect from a VF
to itself in the eswitch will break the datapath since the
packets will be dropped once they go back to the vport they
came from.

Therefore, offloading such rules will be rejected and left to
be handled by SW.

Signed-off-by: Ariel Levkovich <lariel@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index dbc06c71c170..a4a4cdecbdea 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -3111,6 +3111,13 @@ static bool same_hw_devs(struct mlx5e_priv *priv, struct mlx5e_priv *peer_priv)
 	return (fsystem_guid == psystem_guid);
 }
 
+static bool same_vf_reps(struct mlx5e_priv *priv,
+			 struct net_device *out_dev)
+{
+	return mlx5e_eswitch_vf_rep(priv->netdev) &&
+	       priv->netdev == out_dev;
+}
+
 static int add_vlan_rewrite_action(struct mlx5e_priv *priv, int namespace,
 				   const struct flow_action_entry *act,
 				   struct mlx5e_tc_flow_parse_attr *parse_attr,
@@ -3796,6 +3803,12 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,
 					return -EOPNOTSUPP;
 				}
 
+				if (same_vf_reps(priv, out_dev)) {
+					NL_SET_ERR_MSG_MOD(extack,
+							   "can't forward from a VF to itself");
+					return -EOPNOTSUPP;
+				}
+
 				out_priv = netdev_priv(out_dev);
 				rpriv = out_priv->ppriv;
 				esw_attr->dests[esw_attr->out_count].rep = rpriv->rep;
-- 
2.30.2

[net-next 15/16] net/mlx5e: Dynamic alloc arfs table for netdev when needed

[Saeed Mahameed]

From: Vu Pham <vuhuong@nvidia.com>

Dynamic allocate arfs table in mlx5e_priv for EN netdev
when needed. Don't allocate it for representor netdev.

Signed-off-by: Vu Pham <vuhuong@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/en/fs.h   | 28 +-----
 .../net/ethernet/mellanox/mlx5/core/en_arfs.c | 95 +++++++++++++------
 2 files changed, 69 insertions(+), 54 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
index d53fb1e31b05..cbfe31fb2b12 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
@@ -200,31 +200,7 @@ static inline int mlx5e_ethtool_get_rxnfc(struct net_device *dev,
 #endif /* CONFIG_MLX5_EN_RXNFC */
 
 #ifdef CONFIG_MLX5_EN_ARFS
-#define ARFS_HASH_SHIFT BITS_PER_BYTE
-#define ARFS_HASH_SIZE BIT(BITS_PER_BYTE)
-
-struct arfs_table {
-	struct mlx5e_flow_table  ft;
-	struct mlx5_flow_handle	 *default_rule;
-	struct hlist_head	 rules_hash[ARFS_HASH_SIZE];
-};
-
-enum  arfs_type {
-	ARFS_IPV4_TCP,
-	ARFS_IPV6_TCP,
-	ARFS_IPV4_UDP,
-	ARFS_IPV6_UDP,
-	ARFS_NUM_TYPES,
-};
-
-struct mlx5e_arfs_tables {
-	struct arfs_table arfs_tables[ARFS_NUM_TYPES];
-	/* Protect aRFS rules list */
-	spinlock_t                     arfs_lock;
-	struct list_head               rules;
-	int                            last_filter_id;
-	struct workqueue_struct        *wq;
-};
+struct mlx5e_arfs_tables;
 
 int mlx5e_arfs_create_tables(struct mlx5e_priv *priv);
 void mlx5e_arfs_destroy_tables(struct mlx5e_priv *priv);
@@ -260,7 +236,7 @@ struct mlx5e_flow_steering {
 	struct mlx5e_ttc_table          ttc;
 	struct mlx5e_ttc_table          inner_ttc;
 #ifdef CONFIG_MLX5_EN_ARFS
-	struct mlx5e_arfs_tables        arfs;
+	struct mlx5e_arfs_tables       *arfs;
 #endif
 #ifdef CONFIG_MLX5_EN_TLS
 	struct mlx5e_accel_fs_tcp      *accel_tcp;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
index 39475f6565c7..d5b1eb74d5e5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c
@@ -36,6 +36,32 @@
 #include <linux/ipv6.h>
 #include "en.h"
 
+#define ARFS_HASH_SHIFT BITS_PER_BYTE
+#define ARFS_HASH_SIZE BIT(BITS_PER_BYTE)
+
+struct arfs_table {
+	struct mlx5e_flow_table  ft;
+	struct mlx5_flow_handle	 *default_rule;
+	struct hlist_head	 rules_hash[ARFS_HASH_SIZE];
+};
+
+enum arfs_type {
+	ARFS_IPV4_TCP,
+	ARFS_IPV6_TCP,
+	ARFS_IPV4_UDP,
+	ARFS_IPV6_UDP,
+	ARFS_NUM_TYPES,
+};
+
+struct mlx5e_arfs_tables {
+	struct arfs_table arfs_tables[ARFS_NUM_TYPES];
+	/* Protect aRFS rules list */
+	spinlock_t                     arfs_lock;
+	struct list_head               rules;
+	int                            last_filter_id;
+	struct workqueue_struct        *wq;
+};
+
 struct arfs_tuple {
 	__be16 etype;
 	u8     ip_proto;
@@ -121,7 +147,7 @@ int mlx5e_arfs_enable(struct mlx5e_priv *priv)
 
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
 	for (i = 0; i < ARFS_NUM_TYPES; i++) {
-		dest.ft = priv->fs.arfs.arfs_tables[i].ft.t;
+		dest.ft = priv->fs.arfs->arfs_tables[i].ft.t;
 		/* Modify ttc rules destination to point on the aRFS FTs */
 		err = mlx5e_ttc_fwd_dest(priv, arfs_get_tt(i), &dest);
 		if (err) {
@@ -141,25 +167,31 @@ static void arfs_destroy_table(struct arfs_table *arfs_t)
 	mlx5e_destroy_flow_table(&arfs_t->ft);
 }
 
-void mlx5e_arfs_destroy_tables(struct mlx5e_priv *priv)
+static void _mlx5e_cleanup_tables(struct mlx5e_priv *priv)
 {
 	int i;
 
-	if (!(priv->netdev->hw_features & NETIF_F_NTUPLE))
-		return;
-
 	arfs_del_rules(priv);
-	destroy_workqueue(priv->fs.arfs.wq);
+	destroy_workqueue(priv->fs.arfs->wq);
 	for (i = 0; i < ARFS_NUM_TYPES; i++) {
-		if (!IS_ERR_OR_NULL(priv->fs.arfs.arfs_tables[i].ft.t))
-			arfs_destroy_table(&priv->fs.arfs.arfs_tables[i]);
+		if (!IS_ERR_OR_NULL(priv->fs.arfs->arfs_tables[i].ft.t))
+			arfs_destroy_table(&priv->fs.arfs->arfs_tables[i]);
 	}
 }
 
+void mlx5e_arfs_destroy_tables(struct mlx5e_priv *priv)
+{
+	if (!(priv->netdev->hw_features & NETIF_F_NTUPLE))
+		return;
+
+	_mlx5e_cleanup_tables(priv);
+	kvfree(priv->fs.arfs);
+}
+
 static int arfs_add_default_rule(struct mlx5e_priv *priv,
 				 enum arfs_type type)
 {
-	struct arfs_table *arfs_t = &priv->fs.arfs.arfs_tables[type];
+	struct arfs_table *arfs_t = &priv->fs.arfs->arfs_tables[type];
 	struct mlx5e_tir *tir = priv->indir_tir;
 	struct mlx5_flow_destination dest = {};
 	MLX5_DECLARE_FLOW_ACT(flow_act);
@@ -290,7 +322,7 @@ static int arfs_create_groups(struct mlx5e_flow_table *ft,
 static int arfs_create_table(struct mlx5e_priv *priv,
 			     enum arfs_type type)
 {
-	struct mlx5e_arfs_tables *arfs = &priv->fs.arfs;
+	struct mlx5e_arfs_tables *arfs = priv->fs.arfs;
 	struct mlx5e_flow_table *ft = &arfs->arfs_tables[type].ft;
 	struct mlx5_flow_table_attr ft_attr = {};
 	int err;
@@ -330,20 +362,27 @@ int mlx5e_arfs_create_tables(struct mlx5e_priv *priv)
 	if (!(priv->netdev->hw_features & NETIF_F_NTUPLE))
 		return 0;
 
-	spin_lock_init(&priv->fs.arfs.arfs_lock);
-	INIT_LIST_HEAD(&priv->fs.arfs.rules);
-	priv->fs.arfs.wq = create_singlethread_workqueue("mlx5e_arfs");
-	if (!priv->fs.arfs.wq)
+	priv->fs.arfs = kvzalloc(sizeof(*priv->fs.arfs), GFP_KERNEL);
+	if (!priv->fs.arfs)
 		return -ENOMEM;
 
+	spin_lock_init(&priv->fs.arfs->arfs_lock);
+	INIT_LIST_HEAD(&priv->fs.arfs->rules);
+	priv->fs.arfs->wq = create_singlethread_workqueue("mlx5e_arfs");
+	if (!priv->fs.arfs->wq)
+		goto err;
+
 	for (i = 0; i < ARFS_NUM_TYPES; i++) {
 		err = arfs_create_table(priv, i);
 		if (err)
-			goto err;
+			goto err_des;
 	}
 	return 0;
+
+err_des:
+	_mlx5e_cleanup_tables(priv);
 err:
-	mlx5e_arfs_destroy_tables(priv);
+	kvfree(priv->fs.arfs);
 	return err;
 }
 
@@ -358,8 +397,8 @@ static void arfs_may_expire_flow(struct mlx5e_priv *priv)
 	int j;
 
 	HLIST_HEAD(del_list);
-	spin_lock_bh(&priv->fs.arfs.arfs_lock);
-	mlx5e_for_each_arfs_rule(arfs_rule, htmp, priv->fs.arfs.arfs_tables, i, j) {
+	spin_lock_bh(&priv->fs.arfs->arfs_lock);
+	mlx5e_for_each_arfs_rule(arfs_rule, htmp, priv->fs.arfs->arfs_tables, i, j) {
 		if (!work_pending(&arfs_rule->arfs_work) &&
 		    rps_may_expire_flow(priv->netdev,
 					arfs_rule->rxq, arfs_rule->flow_id,
@@ -370,7 +409,7 @@ static void arfs_may_expire_flow(struct mlx5e_priv *priv)
 				break;
 		}
 	}
-	spin_unlock_bh(&priv->fs.arfs.arfs_lock);
+	spin_unlock_bh(&priv->fs.arfs->arfs_lock);
 	hlist_for_each_entry_safe(arfs_rule, htmp, &del_list, hlist) {
 		if (arfs_rule->rule)
 			mlx5_del_flow_rules(arfs_rule->rule);
@@ -387,12 +426,12 @@ static void arfs_del_rules(struct mlx5e_priv *priv)
 	int j;
 
 	HLIST_HEAD(del_list);
-	spin_lock_bh(&priv->fs.arfs.arfs_lock);
-	mlx5e_for_each_arfs_rule(rule, htmp, priv->fs.arfs.arfs_tables, i, j) {
+	spin_lock_bh(&priv->fs.arfs->arfs_lock);
+	mlx5e_for_each_arfs_rule(rule, htmp, priv->fs.arfs->arfs_tables, i, j) {
 		hlist_del_init(&rule->hlist);
 		hlist_add_head(&rule->hlist, &del_list);
 	}
-	spin_unlock_bh(&priv->fs.arfs.arfs_lock);
+	spin_unlock_bh(&priv->fs.arfs->arfs_lock);
 
 	hlist_for_each_entry_safe(rule, htmp, &del_list, hlist) {
 		cancel_work_sync(&rule->arfs_work);
@@ -436,7 +475,7 @@ static struct arfs_table *arfs_get_table(struct mlx5e_arfs_tables *arfs,
 static struct mlx5_flow_handle *arfs_add_rule(struct mlx5e_priv *priv,
 					      struct arfs_rule *arfs_rule)
 {
-	struct mlx5e_arfs_tables *arfs = &priv->fs.arfs;
+	struct mlx5e_arfs_tables *arfs = priv->fs.arfs;
 	struct arfs_tuple *tuple = &arfs_rule->tuple;
 	struct mlx5_flow_handle *rule = NULL;
 	struct mlx5_flow_destination dest = {};
@@ -554,9 +593,9 @@ static void arfs_handle_work(struct work_struct *work)
 
 	mutex_lock(&priv->state_lock);
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
-		spin_lock_bh(&priv->fs.arfs.arfs_lock);
+		spin_lock_bh(&priv->fs.arfs->arfs_lock);
 		hlist_del(&arfs_rule->hlist);
-		spin_unlock_bh(&priv->fs.arfs.arfs_lock);
+		spin_unlock_bh(&priv->fs.arfs->arfs_lock);
 
 		mutex_unlock(&priv->state_lock);
 		kfree(arfs_rule);
@@ -609,7 +648,7 @@ static struct arfs_rule *arfs_alloc_rule(struct mlx5e_priv *priv,
 	tuple->dst_port = fk->ports.dst;
 
 	rule->flow_id = flow_id;
-	rule->filter_id = priv->fs.arfs.last_filter_id++ % RPS_NO_FILTER;
+	rule->filter_id = priv->fs.arfs->last_filter_id++ % RPS_NO_FILTER;
 
 	hlist_add_head(&rule->hlist,
 		       arfs_hash_bucket(arfs_t, tuple->src_port,
@@ -653,7 +692,7 @@ int mlx5e_rx_flow_steer(struct net_device *dev, const struct sk_buff *skb,
 			u16 rxq_index, u32 flow_id)
 {
 	struct mlx5e_priv *priv = netdev_priv(dev);
-	struct mlx5e_arfs_tables *arfs = &priv->fs.arfs;
+	struct mlx5e_arfs_tables *arfs = priv->fs.arfs;
 	struct arfs_table *arfs_t;
 	struct arfs_rule *arfs_rule;
 	struct flow_keys fk;
@@ -687,7 +726,7 @@ int mlx5e_rx_flow_steer(struct net_device *dev, const struct sk_buff *skb,
 			return -ENOMEM;
 		}
 	}
-	queue_work(priv->fs.arfs.wq, &arfs_rule->arfs_work);
+	queue_work(priv->fs.arfs->wq, &arfs_rule->arfs_work);
 	spin_unlock_bh(&arfs->arfs_lock);
 	return arfs_rule->filter_id;
 }
-- 
2.30.2

[net-next 16/16] net/mlx5e: Dynamic alloc vlan table for netdev when needed

[Saeed Mahameed]

From: Vu Pham <vuhuong@nvidia.com>

Dynamic allocate vlan table in mlx5e_priv for EN netdev
when needed. Don't allocate it for representor netdev.

Signed-off-by: Vu Pham <vuhuong@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/en/fs.h   |  18 +--
 .../net/ethernet/mellanox/mlx5/core/en_fs.c   | 132 +++++++++++-------
 .../net/ethernet/mellanox/mlx5/core/en_main.c |   3 +-
 .../net/ethernet/mellanox/mlx5/core/en_tc.c   |   4 +-
 4 files changed, 90 insertions(+), 67 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
index cbfe31fb2b12..373bd89484cb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h
@@ -49,18 +49,10 @@ struct mlx5e_promisc_table {
 	struct mlx5_flow_handle	*rule;
 };
 
-struct mlx5e_vlan_table {
-	struct mlx5e_flow_table		ft;
-	DECLARE_BITMAP(active_cvlans, VLAN_N_VID);
-	DECLARE_BITMAP(active_svlans, VLAN_N_VID);
-	struct mlx5_flow_handle	*active_cvlans_rule[VLAN_N_VID];
-	struct mlx5_flow_handle	*active_svlans_rule[VLAN_N_VID];
-	struct mlx5_flow_handle	*untagged_rule;
-	struct mlx5_flow_handle	*any_cvlan_rule;
-	struct mlx5_flow_handle	*any_svlan_rule;
-	struct mlx5_flow_handle	*trap_rule;
-	bool			cvlan_filter_disabled;
-};
+/* Forward declaration and APIs to get private fields of vlan_table */
+struct mlx5e_vlan_table;
+unsigned long *mlx5e_vlan_get_active_svlans(struct mlx5e_vlan_table *vlan);
+struct mlx5_flow_table *mlx5e_vlan_get_flowtable(struct mlx5e_vlan_table *vlan);
 
 struct mlx5e_l2_table {
 	struct mlx5e_flow_table    ft;
@@ -231,7 +223,7 @@ struct mlx5e_flow_steering {
 #endif
 	struct mlx5e_tc_table           tc;
 	struct mlx5e_promisc_table      promisc;
-	struct mlx5e_vlan_table         vlan;
+	struct mlx5e_vlan_table         *vlan;
 	struct mlx5e_l2_table           l2;
 	struct mlx5e_ttc_table          ttc;
 	struct mlx5e_ttc_table          inner_ttc;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c b/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
index 98f0b857947e..0d571a0c76d9 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_fs.c
@@ -108,6 +108,29 @@ static void mlx5e_del_l2_from_hash(struct mlx5e_l2_hash_node *hn)
 	kfree(hn);
 }
 
+struct mlx5e_vlan_table {
+	struct mlx5e_flow_table		ft;
+	DECLARE_BITMAP(active_cvlans, VLAN_N_VID);
+	DECLARE_BITMAP(active_svlans, VLAN_N_VID);
+	struct mlx5_flow_handle	*active_cvlans_rule[VLAN_N_VID];
+	struct mlx5_flow_handle	*active_svlans_rule[VLAN_N_VID];
+	struct mlx5_flow_handle	*untagged_rule;
+	struct mlx5_flow_handle	*any_cvlan_rule;
+	struct mlx5_flow_handle	*any_svlan_rule;
+	struct mlx5_flow_handle	*trap_rule;
+	bool			cvlan_filter_disabled;
+};
+
+unsigned long *mlx5e_vlan_get_active_svlans(struct mlx5e_vlan_table *vlan)
+{
+	return vlan->active_svlans;
+}
+
+struct mlx5_flow_table *mlx5e_vlan_get_flowtable(struct mlx5e_vlan_table *vlan)
+{
+	return vlan->ft.t;
+}
+
 static int mlx5e_vport_context_update_vlans(struct mlx5e_priv *priv)
 {
 	struct net_device *ndev = priv->netdev;
@@ -119,7 +142,7 @@ static int mlx5e_vport_context_update_vlans(struct mlx5e_priv *priv)
 	int i;
 
 	list_size = 0;
-	for_each_set_bit(vlan, priv->fs.vlan.active_cvlans, VLAN_N_VID)
+	for_each_set_bit(vlan, priv->fs.vlan->active_cvlans, VLAN_N_VID)
 		list_size++;
 
 	max_list_size = 1 << MLX5_CAP_GEN(priv->mdev, log_max_vlan_list);
@@ -136,7 +159,7 @@ static int mlx5e_vport_context_update_vlans(struct mlx5e_priv *priv)
 		return -ENOMEM;
 
 	i = 0;
-	for_each_set_bit(vlan, priv->fs.vlan.active_cvlans, VLAN_N_VID) {
+	for_each_set_bit(vlan, priv->fs.vlan->active_cvlans, VLAN_N_VID) {
 		if (i >= list_size)
 			break;
 		vlans[i++] = vlan;
@@ -163,7 +186,7 @@ static int __mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
 				 enum mlx5e_vlan_rule_type rule_type,
 				 u16 vid, struct mlx5_flow_spec *spec)
 {
-	struct mlx5_flow_table *ft = priv->fs.vlan.ft.t;
+	struct mlx5_flow_table *ft = priv->fs.vlan->ft.t;
 	struct mlx5_flow_destination dest = {};
 	struct mlx5_flow_handle **rule_p;
 	MLX5_DECLARE_FLOW_ACT(flow_act);
@@ -180,24 +203,24 @@ static int __mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
 		 * disabled in match value means both S & C tags
 		 * don't exist (untagged of both)
 		 */
-		rule_p = &priv->fs.vlan.untagged_rule;
+		rule_p = &priv->fs.vlan->untagged_rule;
 		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.cvlan_tag);
 		break;
 	case MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID:
-		rule_p = &priv->fs.vlan.any_cvlan_rule;
+		rule_p = &priv->fs.vlan->any_cvlan_rule;
 		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.cvlan_tag);
 		MLX5_SET(fte_match_param, spec->match_value, outer_headers.cvlan_tag, 1);
 		break;
 	case MLX5E_VLAN_RULE_TYPE_ANY_STAG_VID:
-		rule_p = &priv->fs.vlan.any_svlan_rule;
+		rule_p = &priv->fs.vlan->any_svlan_rule;
 		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.svlan_tag);
 		MLX5_SET(fte_match_param, spec->match_value, outer_headers.svlan_tag, 1);
 		break;
 	case MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID:
-		rule_p = &priv->fs.vlan.active_svlans_rule[vid];
+		rule_p = &priv->fs.vlan->active_svlans_rule[vid];
 		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.svlan_tag);
 		MLX5_SET(fte_match_param, spec->match_value, outer_headers.svlan_tag, 1);
@@ -207,7 +230,7 @@ static int __mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
 			 vid);
 		break;
 	default: /* MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID */
-		rule_p = &priv->fs.vlan.active_cvlans_rule[vid];
+		rule_p = &priv->fs.vlan->active_cvlans_rule[vid];
 		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.cvlan_tag);
 		MLX5_SET(fte_match_param, spec->match_value, outer_headers.cvlan_tag, 1);
@@ -257,33 +280,33 @@ static void mlx5e_del_vlan_rule(struct mlx5e_priv *priv,
 {
 	switch (rule_type) {
 	case MLX5E_VLAN_RULE_TYPE_UNTAGGED:
-		if (priv->fs.vlan.untagged_rule) {
-			mlx5_del_flow_rules(priv->fs.vlan.untagged_rule);
-			priv->fs.vlan.untagged_rule = NULL;
+		if (priv->fs.vlan->untagged_rule) {
+			mlx5_del_flow_rules(priv->fs.vlan->untagged_rule);
+			priv->fs.vlan->untagged_rule = NULL;
 		}
 		break;
 	case MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID:
-		if (priv->fs.vlan.any_cvlan_rule) {
-			mlx5_del_flow_rules(priv->fs.vlan.any_cvlan_rule);
-			priv->fs.vlan.any_cvlan_rule = NULL;
+		if (priv->fs.vlan->any_cvlan_rule) {
+			mlx5_del_flow_rules(priv->fs.vlan->any_cvlan_rule);
+			priv->fs.vlan->any_cvlan_rule = NULL;
 		}
 		break;
 	case MLX5E_VLAN_RULE_TYPE_ANY_STAG_VID:
-		if (priv->fs.vlan.any_svlan_rule) {
-			mlx5_del_flow_rules(priv->fs.vlan.any_svlan_rule);
-			priv->fs.vlan.any_svlan_rule = NULL;
+		if (priv->fs.vlan->any_svlan_rule) {
+			mlx5_del_flow_rules(priv->fs.vlan->any_svlan_rule);
+			priv->fs.vlan->any_svlan_rule = NULL;
 		}
 		break;
 	case MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID:
-		if (priv->fs.vlan.active_svlans_rule[vid]) {
-			mlx5_del_flow_rules(priv->fs.vlan.active_svlans_rule[vid]);
-			priv->fs.vlan.active_svlans_rule[vid] = NULL;
+		if (priv->fs.vlan->active_svlans_rule[vid]) {
+			mlx5_del_flow_rules(priv->fs.vlan->active_svlans_rule[vid]);
+			priv->fs.vlan->active_svlans_rule[vid] = NULL;
 		}
 		break;
 	case MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID:
-		if (priv->fs.vlan.active_cvlans_rule[vid]) {
-			mlx5_del_flow_rules(priv->fs.vlan.active_cvlans_rule[vid]);
-			priv->fs.vlan.active_cvlans_rule[vid] = NULL;
+		if (priv->fs.vlan->active_cvlans_rule[vid]) {
+			mlx5_del_flow_rules(priv->fs.vlan->active_cvlans_rule[vid]);
+			priv->fs.vlan->active_cvlans_rule[vid] = NULL;
 		}
 		mlx5e_vport_context_update_vlans(priv);
 		break;
@@ -330,27 +353,27 @@ mlx5e_add_trap_rule(struct mlx5_flow_table *ft, int trap_id, int tir_num)
 
 int mlx5e_add_vlan_trap(struct mlx5e_priv *priv, int trap_id, int tir_num)
 {
-	struct mlx5_flow_table *ft = priv->fs.vlan.ft.t;
+	struct mlx5_flow_table *ft = priv->fs.vlan->ft.t;
 	struct mlx5_flow_handle *rule;
 	int err;
 
 	rule = mlx5e_add_trap_rule(ft, trap_id, tir_num);
 	if (IS_ERR(rule)) {
 		err = PTR_ERR(rule);
-		priv->fs.vlan.trap_rule = NULL;
+		priv->fs.vlan->trap_rule = NULL;
 		netdev_err(priv->netdev, "%s: add VLAN trap rule failed, err %d\n",
 			   __func__, err);
 		return err;
 	}
-	priv->fs.vlan.trap_rule = rule;
+	priv->fs.vlan->trap_rule = rule;
 	return 0;
 }
 
 void mlx5e_remove_vlan_trap(struct mlx5e_priv *priv)
 {
-	if (priv->fs.vlan.trap_rule) {
-		mlx5_del_flow_rules(priv->fs.vlan.trap_rule);
-		priv->fs.vlan.trap_rule = NULL;
+	if (priv->fs.vlan->trap_rule) {
+		mlx5_del_flow_rules(priv->fs.vlan->trap_rule);
+		priv->fs.vlan->trap_rule = NULL;
 	}
 }
 
@@ -382,10 +405,10 @@ void mlx5e_remove_mac_trap(struct mlx5e_priv *priv)
 
 void mlx5e_enable_cvlan_filter(struct mlx5e_priv *priv)
 {
-	if (!priv->fs.vlan.cvlan_filter_disabled)
+	if (!priv->fs.vlan->cvlan_filter_disabled)
 		return;
 
-	priv->fs.vlan.cvlan_filter_disabled = false;
+	priv->fs.vlan->cvlan_filter_disabled = false;
 	if (priv->netdev->flags & IFF_PROMISC)
 		return;
 	mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID, 0);
@@ -393,10 +416,10 @@ void mlx5e_enable_cvlan_filter(struct mlx5e_priv *priv)
 
 void mlx5e_disable_cvlan_filter(struct mlx5e_priv *priv)
 {
-	if (priv->fs.vlan.cvlan_filter_disabled)
+	if (priv->fs.vlan->cvlan_filter_disabled)
 		return;
 
-	priv->fs.vlan.cvlan_filter_disabled = true;
+	priv->fs.vlan->cvlan_filter_disabled = true;
 	if (priv->netdev->flags & IFF_PROMISC)
 		return;
 	mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_ANY_CTAG_VID, 0);
@@ -406,11 +429,11 @@ static int mlx5e_vlan_rx_add_cvid(struct mlx5e_priv *priv, u16 vid)
 {
 	int err;
 
-	set_bit(vid, priv->fs.vlan.active_cvlans);
+	set_bit(vid, priv->fs.vlan->active_cvlans);
 
 	err = mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, vid);
 	if (err)
-		clear_bit(vid, priv->fs.vlan.active_cvlans);
+		clear_bit(vid, priv->fs.vlan->active_cvlans);
 
 	return err;
 }
@@ -420,11 +443,11 @@ static int mlx5e_vlan_rx_add_svid(struct mlx5e_priv *priv, u16 vid)
 	struct net_device *netdev = priv->netdev;
 	int err;
 
-	set_bit(vid, priv->fs.vlan.active_svlans);
+	set_bit(vid, priv->fs.vlan->active_svlans);
 
 	err = mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, vid);
 	if (err) {
-		clear_bit(vid, priv->fs.vlan.active_svlans);
+		clear_bit(vid, priv->fs.vlan->active_svlans);
 		return err;
 	}
 
@@ -456,10 +479,10 @@ int mlx5e_vlan_rx_kill_vid(struct net_device *dev, __be16 proto, u16 vid)
 		return 0; /* no vlan table for uplink rep */
 
 	if (be16_to_cpu(proto) == ETH_P_8021Q) {
-		clear_bit(vid, priv->fs.vlan.active_cvlans);
+		clear_bit(vid, priv->fs.vlan->active_cvlans);
 		mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, vid);
 	} else if (be16_to_cpu(proto) == ETH_P_8021AD) {
-		clear_bit(vid, priv->fs.vlan.active_svlans);
+		clear_bit(vid, priv->fs.vlan->active_svlans);
 		mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, vid);
 		netdev_update_features(dev);
 	}
@@ -473,14 +496,14 @@ static void mlx5e_add_vlan_rules(struct mlx5e_priv *priv)
 
 	mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_UNTAGGED, 0);
 
-	for_each_set_bit(i, priv->fs.vlan.active_cvlans, VLAN_N_VID) {
+	for_each_set_bit(i, priv->fs.vlan->active_cvlans, VLAN_N_VID) {
 		mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, i);
 	}
 
-	for_each_set_bit(i, priv->fs.vlan.active_svlans, VLAN_N_VID)
+	for_each_set_bit(i, priv->fs.vlan->active_svlans, VLAN_N_VID)
 		mlx5e_add_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, i);
 
-	if (priv->fs.vlan.cvlan_filter_disabled)
+	if (priv->fs.vlan->cvlan_filter_disabled)
 		mlx5e_add_any_vid_rules(priv);
 }
 
@@ -490,11 +513,11 @@ static void mlx5e_del_vlan_rules(struct mlx5e_priv *priv)
 
 	mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_UNTAGGED, 0);
 
-	for_each_set_bit(i, priv->fs.vlan.active_cvlans, VLAN_N_VID) {
+	for_each_set_bit(i, priv->fs.vlan->active_cvlans, VLAN_N_VID) {
 		mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_CTAG_VID, i);
 	}
 
-	for_each_set_bit(i, priv->fs.vlan.active_svlans, VLAN_N_VID)
+	for_each_set_bit(i, priv->fs.vlan->active_svlans, VLAN_N_VID)
 		mlx5e_del_vlan_rule(priv, MLX5E_VLAN_RULE_TYPE_MATCH_STAG_VID, i);
 
 	WARN_ON_ONCE(!(test_bit(MLX5E_STATE_DESTROYING, &priv->state)));
@@ -504,7 +527,7 @@ static void mlx5e_del_vlan_rules(struct mlx5e_priv *priv)
 	/* must be called after DESTROY bit is set and
 	 * set_rx_mode is called and flushed
 	 */
-	if (priv->fs.vlan.cvlan_filter_disabled)
+	if (priv->fs.vlan->cvlan_filter_disabled)
 		mlx5e_del_any_vid_rules(priv);
 }
 
@@ -1692,10 +1715,15 @@ static int mlx5e_create_vlan_table_groups(struct mlx5e_flow_table *ft)
 
 static int mlx5e_create_vlan_table(struct mlx5e_priv *priv)
 {
-	struct mlx5e_flow_table *ft = &priv->fs.vlan.ft;
 	struct mlx5_flow_table_attr ft_attr = {};
+	struct mlx5e_flow_table *ft;
 	int err;
 
+	priv->fs.vlan = kvzalloc(sizeof(*priv->fs.vlan), GFP_KERNEL);
+	if (!priv->fs.vlan)
+		return -ENOMEM;
+
+	ft = &priv->fs.vlan->ft;
 	ft->num_groups = 0;
 
 	ft_attr.max_fte = MLX5E_VLAN_TABLE_SIZE;
@@ -1703,12 +1731,11 @@ static int mlx5e_create_vlan_table(struct mlx5e_priv *priv)
 	ft_attr.prio = MLX5E_NIC_PRIO;
 
 	ft->t = mlx5_create_flow_table(priv->fs.ns, &ft_attr);
-
 	if (IS_ERR(ft->t)) {
 		err = PTR_ERR(ft->t);
-		ft->t = NULL;
-		return err;
+		goto err_free_t;
 	}
+
 	ft->g = kcalloc(MLX5E_NUM_VLAN_GROUPS, sizeof(*ft->g), GFP_KERNEL);
 	if (!ft->g) {
 		err = -ENOMEM;
@@ -1727,7 +1754,9 @@ static int mlx5e_create_vlan_table(struct mlx5e_priv *priv)
 	kfree(ft->g);
 err_destroy_vlan_table:
 	mlx5_destroy_flow_table(ft->t);
-	ft->t = NULL;
+err_free_t:
+	kvfree(priv->fs.vlan);
+	priv->fs.vlan = NULL;
 
 	return err;
 }
@@ -1735,7 +1764,8 @@ static int mlx5e_create_vlan_table(struct mlx5e_priv *priv)
 static void mlx5e_destroy_vlan_table(struct mlx5e_priv *priv)
 {
 	mlx5e_del_vlan_rules(priv);
-	mlx5e_destroy_flow_table(&priv->fs.vlan.ft);
+	mlx5e_destroy_flow_table(&priv->fs.vlan->ft);
+	kvfree(priv->fs.vlan);
 }
 
 int mlx5e_create_flow_steering(struct mlx5e_priv *priv)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index db2942b61fd5..773449c1424b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -3823,7 +3823,8 @@ static netdev_features_t mlx5e_fix_features(struct net_device *netdev,
 
 	mutex_lock(&priv->state_lock);
 	params = &priv->channels.params;
-	if (!bitmap_empty(priv->fs.vlan.active_svlans, VLAN_N_VID)) {
+	if (!priv->fs.vlan ||
+	    !bitmap_empty(mlx5e_vlan_get_active_svlans(priv->fs.vlan), VLAN_N_VID)) {
 		/* HW strips the outer C-tag header, this is a problem
 		 * for S-tag traffic.
 		 */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index a4a4cdecbdea..bb1e0d442b5c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -906,7 +906,7 @@ mlx5e_add_offloaded_nic_rule(struct mlx5e_priv *priv,
 			if (IS_ERR(dest[dest_ix].ft))
 				return ERR_CAST(dest[dest_ix].ft);
 		} else {
-			dest[dest_ix].ft = priv->fs.vlan.ft.t;
+			dest[dest_ix].ft = mlx5e_vlan_get_flowtable(priv->fs.vlan);
 		}
 		dest_ix++;
 	}
@@ -4753,7 +4753,7 @@ int mlx5e_tc_nic_init(struct mlx5e_priv *priv)
 	attr.ns = MLX5_FLOW_NAMESPACE_KERNEL;
 	attr.max_ft_sz = mlx5e_tc_nic_get_ft_size(dev);
 	attr.max_grp_num = MLX5E_TC_TABLE_NUM_GROUPS;
-	attr.default_ft = priv->fs.vlan.ft.t;
+	attr.default_ft = mlx5e_vlan_get_flowtable(priv->fs.vlan);
 
 	tc->chains = mlx5_chains_create(dev, &attr);
 	if (IS_ERR(tc->chains)) {
-- 
2.30.2

Re: [net-next 01/16] net/mlx5: CT: Add support for matching on ct_state inv and rel flags

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (refs/heads/master):

On Sat,  3 Apr 2021 21:19:39 -0700 you wrote:
> From: Ariel Levkovich <lariel@nvidia.com>
> 
> Add support for matching on ct_state inv and rel flags.
> 
> Currently the support is only for match on -inv and -rel.
> Matching on +inv and +rel will be rejected.
> 
> [...]

Here is the summary with links:
  - [net-next,01/16] net/mlx5: CT: Add support for matching on ct_state inv and rel flags
    https://git.kernel.org/netdev/net-next/c/116c76c51035
  - [net-next,02/16] net/mlx5: E-Switch, cut down mlx5_vport_info structure size by 8 bytes
    https://git.kernel.org/netdev/net-next/c/cadb129ffdfe
  - [net-next,03/16] net/mlx5: E-Switch, move QoS specific fields to existing qos struct
    https://git.kernel.org/netdev/net-next/c/e591605f801e
  - [net-next,04/16] net/mlx5: Use unsigned int for free_count
    https://git.kernel.org/netdev/net-next/c/c6baac47d9e6
  - [net-next,05/16] net/mlx5: Pack mlx5_rl_entry structure
    https://git.kernel.org/netdev/net-next/c/4c4c0a89abd5
  - [net-next,06/16] net/mlx5: Do not hold mutex while reading table constants
    https://git.kernel.org/netdev/net-next/c/16e74672a21b
  - [net-next,07/16] net/mlx5: Use helpers to allocate and free rl table entries
    https://git.kernel.org/netdev/net-next/c/51ccc9f5f106
  - [net-next,08/16] net/mlx5: Use helper to increment, decrement rate entry refcount
    https://git.kernel.org/netdev/net-next/c/97d85aba2543
  - [net-next,09/16] net/mlx5: Allocate rate limit table when rate is configured
    https://git.kernel.org/netdev/net-next/c/6b30b6d4d36c
  - [net-next,10/16] net/mlx5: Pair mutex_destory with mutex_init for rate limit table
    https://git.kernel.org/netdev/net-next/c/19779f28c96d
  - [net-next,11/16] net/mlx5: E-Switch, cut down mlx5_vport_info structure size by 8 bytes
    https://git.kernel.org/netdev/net-next/c/b47e1056257c
  - [net-next,12/16] net/mlx5: E-Switch, move QoS specific fields to existing qos struct
    https://git.kernel.org/netdev/net-next/c/233dd7d6565e
  - [net-next,13/16] net/mlx5: Use ida_alloc_range() instead of ida_simple_alloc()
    https://git.kernel.org/netdev/net-next/c/8802b8a44ef8
  - [net-next,14/16] net/mlx5e: Reject tc rules which redirect from a VF to itself
    https://git.kernel.org/netdev/net-next/c/bb5696570b0b
  - [net-next,15/16] net/mlx5e: Dynamic alloc arfs table for netdev when needed
    https://git.kernel.org/netdev/net-next/c/f6755b80d693
  - [net-next,16/16] net/mlx5e: Dynamic alloc vlan table for netdev when needed
    https://git.kernel.org/netdev/net-next/c/6783f0a21a3c

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html