Skip to contentSkip to conversation

mlx5 fixes 2021-01-07

GIT PULL In review Saeed Mahameed 3 participants 21 messages Jan 7, 2021 – Jan 8, 2021

Subsystems (6)
lore
Share & tools
lore
#0 lore raw

From: Saeed Mahameed <[email protected]>

Hi Dave, Jakub,

This series provides some fixes to mlx5 driver. Please pull and let me know if there is any problem.

For -stable v5.2
('net/mlx5: Use port_num 1 instead of 0 when delete a RoCE address')

For -stable v5.5
('net/mlx5e: ethtool, Fix restriction of autoneg with 56G')

For -stable v5.8
('net/mlx5e: In skb build skip setting mark in switchdev mode')

For -stable v5.10
('net/mlx5: Check if lag is supported before creating one') ('net/mlx5e: Fix SWP offsets when vlan inserted by driver')

Thanks,
Saeed.

---
The following changes since commit 5316a7c0130acf09bfc8bb0092407006010fcccc:

tools: selftests: add test for changing routes with PTMU exceptions (2021-01-07 12:03:36 -0800)

are available in the Git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-fixes-2021-01-07

for you to fetch changes up to 5b0bb12c58ac7d22e05b5bfdaa30a116c8c32e32:

net/mlx5e: Fix memleak in mlx5e_create_l2_table_groups (2021-01-07 12:22:51 -0800)

----------------------------------------------------------------
mlx5-fixes-2021-01-07

----------------------------------------------------------------
Alaa Hleihel (1):
net/mlx5: E-Switch, fix changing vf VLANID

Aya Levin (2):
net/mlx5e: Add missing capability check for uplink follow net/mlx5e: ethtool, Fix restriction of autoneg with 56G

Dinghao Liu (2):
net/mlx5e: Fix two double free cases net/mlx5e: Fix memleak in mlx5e_create_l2_table_groups

Leon Romanovsky (1):
net/mlx5: Release devlink object if adev fails

Maor Dickman (1):
net/mlx5e: In skb build skip setting mark in switchdev mode

Mark Zhang (2):
net/mlx5: Check if lag is supported before creating one net/mlx5: Use port_num 1 instead of 0 when delete a RoCE address

Moshe Shemesh (1):
net/mlx5e: Fix SWP offsets when vlan inserted by driver

Oz Shlomo (1):
net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

 .../net/ethernet/mellanox/mlx5/core/en/rep/tc.c    |  5 ++
 drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c | 77 ++++++++++++++--------
 drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h  |  9 +++
 .../mellanox/mlx5/core/en_accel/en_accel.h         |  8 ++-
 .../net/ethernet/mellanox/mlx5/core/en_ethtool.c   | 24 +++++--
 drivers/net/ethernet/mellanox/mlx5/core/en_fs.c    |  3 +
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c  |  3 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_tx.c    |  9 +--
 .../mellanox/mlx5/core/esw/acl/egress_lgcy.c       | 27 ++++----
 drivers/net/ethernet/mellanox/mlx5/core/lag.c      | 11 ++--
 drivers/net/ethernet/mellanox/mlx5/core/main.c     |  7 +-
 drivers/net/ethernet/mellanox/mlx5/core/rdma.c     |  2 +-
 include/linux/mlx5/mlx5_ifc.h                      |  3 +-
 13 files changed, 122 insertions(+), 66 deletions(-)
#1 lore raw

PATCH 1/11 net/mlx5: Check if lag is supported before creating one

From: Mark Zhang <[email protected]>

This patch fixes a memleak issue by preventing to create a lag and add PFs if lag is not supported.

comm “python3”, pid 349349, jiffies 4296985507 (age 1446.976s) hex dump (first 32 bytes):
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ……………. 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 …………….
backtrace:
[<000000005b216ae7>] mlx5_lag_add+0x1d5/0×3f0 [mlx5_core] [<000000000445aa55>] mlx5e_nic_enable+0x66/0×1b0 [mlx5_core] [<00000000c56734c3>] mlx5e_attach_netdev+0x16e/0×200 [mlx5_core] [<0000000030439d1f>] mlx5e_attach+0x5c/0×90 [mlx5_core] [<0000000018fd8615>] mlx5e_add+0x1a4/0×410 [mlx5_core] [<0000000068bc504b>] mlx5_add_device+0x72/0×120 [mlx5_core]

  [<000000009fce51f9>] mlx5_register_device+0x77/0xb0 [mlx5_core]
  [<00000000d0d81ff3>] mlx5_load_one+0xc58/0×1eb0 [mlx5_core]
  [<0000000045077adc>] init_one+0x3ea/0×920 [mlx5_core]
  [<0000000043287674>] pci_device_probe+0xcd/0×150
  [<00000000dafd3279>] really_probe+0x1c9/0×4b0
  [<00000000f06bdd84>] driver_probe_device+0x5d/0×140
  [<00000000e3d508b6>] device_driver_attach+0x4f/0×60
  [<0000000084fba0f0>] bind_store+0xbf/0×120
  [<00000000bf6622b3>] kernfs_fop_write+0x114/0×1b0
Fixes 9b412cc35f00 ("net/mlx5e: Add LAG warning if bond slave is not lag master")
---
 drivers/net/ethernet/mellanox/mlx5/core/lag.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/lag.c +5 -6
index f3d45ef082cd..83a05371e2aa 100644
@@ -564,7 +564,9 @@ void mlx5_lag_add(struct mlx5_core_dev *dev, struct net_device *netdev)
 	struct mlx5_core_dev *tmp_dev;
 	int i, err;
 
-	if (!MLX5_CAP_GEN(dev, vport_group_manager))
+	if (!MLX5_CAP_GEN(dev, vport_group_manager) ||
+	    !MLX5_CAP_GEN(dev, lag_master) ||
+	    MLX5_CAP_GEN(dev, num_lag_ports) != MLX5_MAX_PORTS)
 		return;
 
 	tmp_dev = mlx5_get_next_phys_dev(dev);
@@ -582,12 +584,9 @@ void mlx5_lag_add(struct mlx5_core_dev *dev, struct net_device *netdev)
 	if (mlx5_lag_dev_add_pf(ldev, dev, netdev) < 0)
 		return;
 
-	for (i = 0; i < MLX5_MAX_PORTS; i++) {
-		tmp_dev = ldev->pf[i].dev;
-		if (!tmp_dev || !MLX5_CAP_GEN(tmp_dev, lag_master) ||
-		    MLX5_CAP_GEN(tmp_dev, num_lag_ports) != MLX5_MAX_PORTS)
+	for (i = 0; i < MLX5_MAX_PORTS; i++)
+		if (!ldev->pf[i].dev)
 			break;
-	}
 
 	if (i >= MLX5_MAX_PORTS)
 		ldev->flags |= MLX5_LAG_FLAG_READY;
Signature · 1 line
2.26.2
#2 lore raw

PATCH 2/11 net/mlx5e: Add missing capability check for uplink follow

In reply to Saeed Mahameed

From: Aya Levin <[email protected]>

Expose firmware indication that it supports setting eswitch uplink state to follow (follow the physical link). Condition setting the eswitch uplink admin-state with this capability bit. Older FW may not support the uplink state setting.

Fixes 7d0314b11cdd ("net/mlx5e: Modify uplink state on interface up/down")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 3 ++-
 include/linux/mlx5/mlx5_ifc.h                     | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_main.c +2 -1
index 7a79d330c075..6a852b4901aa 100644
@@ -3161,7 +3161,8 @@ static void mlx5e_modify_admin_state(struct mlx5_core_dev *mdev,
 
 	mlx5_set_port_admin_status(mdev, state);
 
-	if (mlx5_eswitch_mode(mdev) != MLX5_ESWITCH_LEGACY)
+	if (mlx5_eswitch_mode(mdev) == MLX5_ESWITCH_OFFLOADS ||
+	    !MLX5_CAP_GEN(mdev, uplink_follow))
 		return;
 
 	if (state == MLX5_PORT_UP)
include/linux/mlx5/mlx5_ifc.h +2 -1
index 8fbddec26eb8..442c0160caab 100644
@@ -1280,7 +1280,8 @@ struct mlx5_ifc_cmd_hca_cap_bits {
 	u8	   ece_support[0x1];
 	u8	   reserved_at_a4[0x7];
 	u8         log_max_srq[0x5];
-	u8         reserved_at_b0[0x2];
+	u8         reserved_at_b0[0x1];
+	u8         uplink_follow[0x1];
 	u8         ts_cqe_to_dest_cqn[0x1];
 	u8         reserved_at_b3[0xd];
 
Signature · 1 line
2.26.2
#3 lore raw

PATCH 3/11 net/mlx5: Use port_num 1 instead of 0 when delete a RoCE address

In reply to Saeed Mahameed

From: Mark Zhang <[email protected]>

In multi-port mode, FW reports syndrome 0x2ea48 (invalid vhca_port_number) if the port_num is not 1 or 2.

Fixes 80f09dfc237f ("net/mlx5: Eswitch, enable RoCE loopback traffic")
---
 drivers/net/ethernet/mellanox/mlx5/core/rdma.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/rdma.c +1 -1
index 0fc7de4aa572..8e0dddc6383f 100644
@@ -116,7 +116,7 @@ static int mlx5_rdma_enable_roce_steering(struct mlx5_core_dev *dev)
 static void mlx5_rdma_del_roce_addr(struct mlx5_core_dev *dev)
 {
 	mlx5_core_roce_gid_set(dev, 0, 0, 0,
-			       NULL, NULL, false, 0, 0);
+			       NULL, NULL, false, 0, 1);
 }
 
 static void mlx5_rdma_make_default_gid(struct mlx5_core_dev *dev, union ib_gid *gid)
Signature · 1 line
2.26.2
#4 lore raw

PATCH 4/11 net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

In reply to Saeed Mahameed

From: Oz Shlomo <[email protected]>

Connection counters may be shared for both directions when the counter is used for connection aging purposes. However, if TC flow accounting is enabled then a unique counter is required per direction.

Instantiate a unique counter per direction if the conntrack accounting extension is enabled. Use a shared counter when the connection accounting extension is disabled.

Fixes 1edae2335adf ("net/mlx5e: CT: Use the same counter for both directions")
---
 .../ethernet/mellanox/mlx5/core/en/tc_ct.c    | 77 ++++++++++++-------
 1 file changed, 49 insertions(+), 28 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +49 -28
index e521254d886e..072363e73f1c 100644
@@ -118,16 +118,17 @@ struct mlx5_ct_tuple {
 	u16 zone;
 };
 
-struct mlx5_ct_shared_counter {
+struct mlx5_ct_counter {
 	struct mlx5_fc *counter;
 	refcount_t refcount;
+	bool is_shared;
 };
 
 struct mlx5_ct_entry {
 	struct rhash_head node;
 	struct rhash_head tuple_node;
 	struct rhash_head tuple_nat_node;
-	struct mlx5_ct_shared_counter *shared_counter;
+	struct mlx5_ct_counter *counter;
 	unsigned long cookie;
 	unsigned long restore_cookie;
 	struct mlx5_ct_tuple tuple;
@@ -394,13 +395,14 @@ mlx5_tc_ct_set_tuple_match(struct mlx5e_priv *priv, struct mlx5_flow_spec *spec,
 }
 
 static void
-mlx5_tc_ct_shared_counter_put(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_entry *entry)
+mlx5_tc_ct_counter_put(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_ct_entry *entry)
 {
-	if (!refcount_dec_and_test(&entry->shared_counter->refcount))
+	if (entry->counter->is_shared &&
+	    !refcount_dec_and_test(&entry->counter->refcount))
 		return;
 
-	mlx5_fc_destroy(ct_priv->dev, entry->shared_counter->counter);
-	kfree(entry->shared_counter);
+	mlx5_fc_destroy(ct_priv->dev, entry->counter->counter);
+	kfree(entry->counter);
 }
 
 static void
@@ -699,7 +701,7 @@ mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv,
 	attr->dest_ft = ct_priv->post_ct;
 	attr->ft = nat ? ct_priv->ct_nat : ct_priv->ct;
 	attr->outer_match_level = MLX5_MATCH_L4;
-	attr->counter = entry->shared_counter->counter;
+	attr->counter = entry->counter->counter;
 	attr->flags |= MLX5_ESW_ATTR_FLAG_NO_IN_PORT;
 
 	mlx5_tc_ct_set_tuple_match(netdev_priv(ct_priv->netdev), spec, flow_rule);
@@ -732,13 +734,34 @@ mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv,
 	return err;
 }
 
-static struct mlx5_ct_shared_counter *
+static struct mlx5_ct_counter *
+mlx5_tc_ct_counter_create(struct mlx5_tc_ct_priv *ct_priv)
+{
+	struct mlx5_ct_counter *counter;
+	int ret;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	counter->is_shared = false;
+	counter->counter = mlx5_fc_create(ct_priv->dev, true);
+	if (IS_ERR(counter->counter)) {
+		ct_dbg("Failed to create counter for ct entry");
+		ret = PTR_ERR(counter->counter);
+		kfree(counter);
+		return ERR_PTR(ret);
+	}
+
+	return counter;
+}
+
+static struct mlx5_ct_counter *
 mlx5_tc_ct_shared_counter_get(struct mlx5_tc_ct_priv *ct_priv,
 			      struct mlx5_ct_entry *entry)
 {
 	struct mlx5_ct_tuple rev_tuple = entry->tuple;
-	struct mlx5_ct_shared_counter *shared_counter;
-	struct mlx5_core_dev *dev = ct_priv->dev;
+	struct mlx5_ct_counter *shared_counter;
 	struct mlx5_ct_entry *rev_entry;
 	__be16 tmp_port;
 	int ret;
@@ -767,25 +790,20 @@ mlx5_tc_ct_shared_counter_get(struct mlx5_tc_ct_priv *ct_priv,
 	rev_entry = rhashtable_lookup_fast(&ct_priv->ct_tuples_ht, &rev_tuple,
 					   tuples_ht_params);
 	if (rev_entry) {
-		if (refcount_inc_not_zero(&rev_entry->shared_counter->refcount)) {
+		if (refcount_inc_not_zero(&rev_entry->counter->refcount)) {
 			mutex_unlock(&ct_priv->shared_counter_lock);
-			return rev_entry->shared_counter;
+			return rev_entry->counter;
 		}
 	}
 	mutex_unlock(&ct_priv->shared_counter_lock);
 
-	shared_counter = kzalloc(sizeof(*shared_counter), GFP_KERNEL);
-	if (!shared_counter)
-		return ERR_PTR(-ENOMEM);
-
-	shared_counter->counter = mlx5_fc_create(dev, true);
-	if (IS_ERR(shared_counter->counter)) {
-		ct_dbg("Failed to create counter for ct entry");
-		ret = PTR_ERR(shared_counter->counter);
-		kfree(shared_counter);
+	shared_counter = mlx5_tc_ct_counter_create(ct_priv);
+	if (IS_ERR(shared_counter)) {
+		ret = PTR_ERR(shared_counter);
 		return ERR_PTR(ret);
 	}
 
+	shared_counter->is_shared = true;
 	refcount_set(&shared_counter->refcount, 1);
 	return shared_counter;
 }
@@ -798,10 +816,13 @@ mlx5_tc_ct_entry_add_rules(struct mlx5_tc_ct_priv *ct_priv,
 {
 	int err;
 
-	entry->shared_counter = mlx5_tc_ct_shared_counter_get(ct_priv, entry);
-	if (IS_ERR(entry->shared_counter)) {
-		err = PTR_ERR(entry->shared_counter);
-		ct_dbg("Failed to create counter for ct entry");
+	if (nf_ct_acct_enabled(dev_net(ct_priv->netdev)))
+		entry->counter = mlx5_tc_ct_counter_create(ct_priv);
+	else
+		entry->counter = mlx5_tc_ct_shared_counter_get(ct_priv, entry);
+
+	if (IS_ERR(entry->counter)) {
+		err = PTR_ERR(entry->counter);
 		return err;
 	}
 
@@ -820,7 +841,7 @@ mlx5_tc_ct_entry_add_rules(struct mlx5_tc_ct_priv *ct_priv,
 err_nat:
 	mlx5_tc_ct_entry_del_rule(ct_priv, entry, false);
 err_orig:
-	mlx5_tc_ct_shared_counter_put(ct_priv, entry);
+	mlx5_tc_ct_counter_put(ct_priv, entry);
 	return err;
 }
 
@@ -918,7 +939,7 @@ mlx5_tc_ct_del_ft_entry(struct mlx5_tc_ct_priv *ct_priv,
 	rhashtable_remove_fast(&ct_priv->ct_tuples_ht, &entry->tuple_node,
 			       tuples_ht_params);
 	mutex_unlock(&ct_priv->shared_counter_lock);
-	mlx5_tc_ct_shared_counter_put(ct_priv, entry);
+	mlx5_tc_ct_counter_put(ct_priv, entry);
 
 }
 
@@ -956,7 +977,7 @@ mlx5_tc_ct_block_flow_offload_stats(struct mlx5_ct_ft *ft,
 	if (!entry)
 		return -ENOENT;
 
-	mlx5_fc_query_cached(entry->shared_counter->counter, &bytes, &packets, &lastuse);
+	mlx5_fc_query_cached(entry->counter->counter, &bytes, &packets, &lastuse);
 	flow_stats_update(&f->stats, bytes, packets, 0, lastuse,
 			  FLOW_ACTION_HW_STATS_DELAYED);
 
Signature · 1 line
2.26.2
Tested-by on v1 4/11
#5 lore raw

Re: [net 04/11] net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

Quoted 6 lines · Saeed Mahameed

From: Oz Shlomo <[email protected]>

Connection counters may be shared for both directions when the counter is used for connection aging purposes. However, if TC flow accounting is enabled then a unique counter is required per direction.

Quoted 7 lines

Instantiate a unique counter per direction if the conntrack accounting extension is enabled. Use a shared counter when the connection accounting extension is disabled.

Fixes 1edae2335adf ("net/mlx5e: CT: Use the same counter for both directions")

Thanks.

Quoted 3 lines
maintainer
#6 lore raw

Re: [net 04/11] net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

In reply to Saeed Mahameed

drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv, · line 753

Quoted 13 lines · On Thu, 7 Jan 2021 12:28:38 -0800 Saeed Mahameed wrote:
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +13 -0
+	int ret;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	counter->is_shared = false;
+	counter->counter = mlx5_fc_create(ct_priv->dev, true);
+	if (IS_ERR(counter->counter)) {
+		ct_dbg("Failed to create counter for ct entry");
+		ret = PTR_ERR(counter->counter);
+		kfree(counter);
+		return ERR_PTR(ret);

The err ptr -> ret -> err ptr conversion seems entirely pointless, no?

#7 lore raw

Re: [net 04/11] net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

Quoted 17 lines · Jakub Kicinski

drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv, · line 753

Quoted 13 lines · On Thu, 7 Jan 2021 12:28:38 -0800 Saeed Mahameed wrote:
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +13 -0
+	int ret;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	counter->is_shared = false;
+	counter->counter = mlx5_fc_create(ct_priv->dev, true);
+	if (IS_ERR(counter->counter)) {
+		ct_dbg("Failed to create counter for ct entry");
+		ret = PTR_ERR(counter->counter);
+		kfree(counter);
+		return ERR_PTR(ret);

The err ptr -> ret -> err ptr conversion seems entirely pointless, no?

Indeed, will address this in a net-next patch

#8 lore raw

Re: [net 04/11] net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

Quoted 18 lines · Saeed Mahameed
Quoted 17 lines · Jakub Kicinski

drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv, · line 753

Quoted 13 lines · On Thu, 7 Jan 2021 12:28:38 -0800 Saeed Mahameed wrote:
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +13 -0
+	int ret;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	counter->is_shared = false;
+	counter->counter = mlx5_fc_create(ct_priv->dev, true);
+	if (IS_ERR(counter->counter)) {
+		ct_dbg("Failed to create counter for ct entry");
+		ret = PTR_ERR(counter->counter);
+		kfree(counter);
+		return ERR_PTR(ret);

The err ptr -> ret -> err ptr conversion seems entirely pointless, no?

Quoted 2 lines

Indeed, will address this in a net-next patch

Actually no, because counter is being kfreed so we must return ERR_PTR(ret).

maintainer
#9 lore raw

Re: [net 04/11] net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

Quoted 19 lines · On Thu, 07 Jan 2021 20:18:36 -0800 Saeed Mahameed wrote:
Quoted 18 lines · Saeed Mahameed
Quoted 17 lines · Jakub Kicinski

drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv, · line 753

Quoted 13 lines · On Thu, 7 Jan 2021 12:28:38 -0800 Saeed Mahameed wrote:
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +13 -0
+	int ret;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	counter->is_shared = false;
+	counter->counter = mlx5_fc_create(ct_priv->dev, true);
+	if (IS_ERR(counter->counter)) {
+		ct_dbg("Failed to create counter for ct entry");
+		ret = PTR_ERR(counter->counter);
+		kfree(counter);
+		return ERR_PTR(ret);  

The err ptr -> ret -> err ptr conversion seems entirely pointless, no?

Quoted 5 lines
Quoted 2 lines

Indeed, will address this in a net-next patch

Actually no, because counter is being kfreed so we must return ERR_PTR(ret).

Ah, good point, just the other one then:

+ shared_counter = mlx5_tc_ct_counter_create(ct_priv);
+ if (IS_ERR(shared_counter)) {
+ ret = PTR_ERR(shared_counter);
return ERR_PTR(ret);
}

#10 lore raw

Re: [net 04/11] net/mlx5e: CT: Use per flow counter when CT flow accounting is enabled

Quoted 30 lines · Jakub Kicinski
Quoted 21 lines · On Thu, 07 Jan 2021 20:18:36 -0800 Saeed Mahameed wrote:
Quoted 20 lines · Saeed Mahameed
Quoted 19 lines · Jakub Kicinski
Quoted 11 lines · On Thu, 7 Jan 2021 12:28:38 -0800 Saeed Mahameed wrote:
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +10 -0
+	int ret;
+
+	counter = kzalloc(sizeof(*counter), GFP_KERNEL);
+	if (!counter)
+		return ERR_PTR(-ENOMEM);
+
+	counter->is_shared = false;
+	counter->counter = mlx5_fc_create(ct_priv->dev, true);
+	if (IS_ERR(counter->counter)) {
+		ct_dbg("Failed to create counter for ct

entry");

drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv, · line 753

Quoted 3 lines
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +3 -0
+		ret = PTR_ERR(counter->counter);
+		kfree(counter);
+		return ERR_PTR(ret);  

The err ptr -> ret -> err ptr conversion seems entirely pointless,
no?

Quoted 5 lines
Quoted 2 lines

Indeed, will address this in a net-next patch

Actually no, because counter is being kfreed so we must return ERR_PTR(ret).

Ah, good point, just the other one then:

drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c mlx5_tc_ct_shared_counter_get(struct mlx5_tc_ct_priv *ct_priv, · line 804

Quoted 5 lines
drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +3 -0
+	shared_counter = mlx5_tc_ct_counter_create(ct_priv);
+	if (IS_ERR(shared_counter)) {
+		ret = PTR_ERR(shared_counter);
 		return ERR_PTR(ret);
 	}

Done, will send to net-next once this is back merged.

Thanks.

#11 lore raw

PATCH 5/11 net/mlx5e: Fix SWP offsets when vlan inserted by driver

In reply to Saeed Mahameed

From: Moshe Shemesh <[email protected]>

In case WQE includes inline header the vlan is inserted by driver even if vlan offload is set. On geneve over vlan interface where software parser is used the SWP offsets should be updated according to the added vlan.

Fixes e3cfc7e6b7bd ("net/mlx5e: TX, Add geneve tunnel stateless offload support")
---
 drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h        | 9 +++++++++
 .../net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h  | 8 +++++---
 drivers/net/ethernet/mellanox/mlx5/core/en_tx.c          | 9 +++++----
 3 files changed, 19 insertions(+), 7 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h +9 -0
index 7943eb30b837..4880f2179273 100644
@@ -371,6 +371,15 @@ struct mlx5e_swp_spec {
 	u8 tun_l4_proto;
 };
 
+static inline void mlx5e_eseg_swp_offsets_add_vlan(struct mlx5_wqe_eth_seg *eseg)
+{
+	/* SWP offsets are in 2-bytes words */
+	eseg->swp_outer_l3_offset += VLAN_HLEN / 2;
+	eseg->swp_outer_l4_offset += VLAN_HLEN / 2;
+	eseg->swp_inner_l3_offset += VLAN_HLEN / 2;
+	eseg->swp_inner_l4_offset += VLAN_HLEN / 2;
+}
+
 static inline void
 mlx5e_set_eseg_swp(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg,
 		   struct mlx5e_swp_spec *swp_spec)
drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h +5 -3
index 899b98aca0d3..1fae7fab8297 100644
@@ -51,7 +51,7 @@ static inline bool mlx5_geneve_tx_allowed(struct mlx5_core_dev *mdev)
 }
 
 static inline void
-mlx5e_tx_tunnel_accel(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg)
+mlx5e_tx_tunnel_accel(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg, u16 ihs)
 {
 	struct mlx5e_swp_spec swp_spec = {};
 	unsigned int offset = 0;
@@ -85,6 +85,8 @@ mlx5e_tx_tunnel_accel(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg)
 	}
 
 	mlx5e_set_eseg_swp(skb, eseg, &swp_spec);
+	if (skb_vlan_tag_present(skb) &&  ihs)
+		mlx5e_eseg_swp_offsets_add_vlan(eseg);
 }
 
 #else
@@ -163,7 +165,7 @@ static inline unsigned int mlx5e_accel_tx_ids_len(struct mlx5e_txqsq *sq,
 
 static inline bool mlx5e_accel_tx_eseg(struct mlx5e_priv *priv,
 				       struct sk_buff *skb,
-				       struct mlx5_wqe_eth_seg *eseg)
+				       struct mlx5_wqe_eth_seg *eseg, u16 ihs)
 {
 #ifdef CONFIG_MLX5_EN_IPSEC
 	if (xfrm_offload(skb))
@@ -172,7 +174,7 @@ static inline bool mlx5e_accel_tx_eseg(struct mlx5e_priv *priv,
 
 #if IS_ENABLED(CONFIG_GENEVE)
 	if (skb->encapsulation)
-		mlx5e_tx_tunnel_accel(skb, eseg);
+		mlx5e_tx_tunnel_accel(skb, eseg, ihs);
 #endif
 
 	return true;
drivers/net/ethernet/mellanox/mlx5/core/en_tx.c +5 -4
index e47e2a0059d0..61ed671fe741 100644
@@ -682,9 +682,9 @@ void mlx5e_tx_mpwqe_ensure_complete(struct mlx5e_txqsq *sq)
 
 static bool mlx5e_txwqe_build_eseg(struct mlx5e_priv *priv, struct mlx5e_txqsq *sq,
 				   struct sk_buff *skb, struct mlx5e_accel_tx_state *accel,
-				   struct mlx5_wqe_eth_seg *eseg)
+				   struct mlx5_wqe_eth_seg *eseg, u16 ihs)
 {
-	if (unlikely(!mlx5e_accel_tx_eseg(priv, skb, eseg)))
+	if (unlikely(!mlx5e_accel_tx_eseg(priv, skb, eseg, ihs)))
 		return false;
 
 	mlx5e_txwqe_build_eseg_csum(sq, skb, accel, eseg);
@@ -714,7 +714,8 @@ netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
 		if (mlx5e_tx_skb_supports_mpwqe(skb, &attr)) {
 			struct mlx5_wqe_eth_seg eseg = {};
 
-			if (unlikely(!mlx5e_txwqe_build_eseg(priv, sq, skb, &accel, &eseg)))
+			if (unlikely(!mlx5e_txwqe_build_eseg(priv, sq, skb, &accel, &eseg,
+							     attr.ihs)))
 				return NETDEV_TX_OK;
 
 			mlx5e_sq_xmit_mpwqe(sq, skb, &eseg, netdev_xmit_more());
@@ -731,7 +732,7 @@ netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
 	/* May update the WQE, but may not post other WQEs. */
 	mlx5e_accel_tx_finish(sq, wqe, &accel,
 			      (struct mlx5_wqe_inline_seg *)(wqe->data + wqe_attr.ds_cnt_inl));
-	if (unlikely(!mlx5e_txwqe_build_eseg(priv, sq, skb, &accel, &wqe->eth)))
+	if (unlikely(!mlx5e_txwqe_build_eseg(priv, sq, skb, &accel, &wqe->eth, attr.ihs)))
 		return NETDEV_TX_OK;
 
 	mlx5e_sq_xmit_wqe(sq, skb, &attr, &wqe_attr, wqe, pi, netdev_xmit_more());
Signature · 1 line
2.26.2
maintainer
#12 lore raw

Re: [net 05/11] net/mlx5e: Fix SWP offsets when vlan inserted by driver

drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h mlx5e_tx_tunnel_accel(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg) · line 88

Quoted 1 line · On Thu, 7 Jan 2021 12:28:39 -0800 Saeed Mahameed wrote:
drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h +1 -0
+	if (skb_vlan_tag_present(skb) &&  ihs)

Double space.

#13 lore raw

Re: [net 05/11] net/mlx5e: Fix SWP offsets when vlan inserted by driver

Quoted 4 lines · Jakub Kicinski

drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h mlx5e_tx_tunnel_accel(struct sk_buff *skb, struct mlx5_wqe_eth_seg *eseg) · line 88

Quoted 1 line · On Thu, 7 Jan 2021 12:28:39 -0800 Saeed Mahameed wrote:
drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h +1 -0
+	if (skb_vlan_tag_present(skb) &&  ihs)

Double space.

Will address this in a net-next patch.

thanks!

#14 lore raw

PATCH 6/11 net/mlx5: E-Switch, fix changing vf VLANID

In reply to Saeed Mahameed

From: Alaa Hleihel <[email protected]>

Adding vf VLANID for the first time, or after having cleared previously defined VLANID works fine, however, attempting to change an existing vf VLANID clears the rules on the firmware, but does not add new rules for the new vf VLANID.

Fix this by changing the logic in function esw_acl_egress_lgcy_setup() so that it will always configure egress rules.

Fixes ea651a86d468 ("net/mlx5: E-Switch, Refactor eswitch egress acl codes")
---
 .../mellanox/mlx5/core/esw/acl/egress_lgcy.c  | 27 +++++++++----------
 1 file changed, 13 insertions(+), 14 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/esw/acl/egress_lgcy.c +13 -14
index 2b85d4777303..3e19b1721303 100644
@@ -95,22 +95,21 @@ int esw_acl_egress_lgcy_setup(struct mlx5_eswitch *esw,
 		return 0;
 	}
 
-	if (!IS_ERR_OR_NULL(vport->egress.acl))
-		return 0;
-
-	vport->egress.acl = esw_acl_table_create(esw, vport->vport,
-						 MLX5_FLOW_NAMESPACE_ESW_EGRESS,
-						 table_size);
-	if (IS_ERR(vport->egress.acl)) {
-		err = PTR_ERR(vport->egress.acl);
-		vport->egress.acl = NULL;
-		goto out;
+	if (!vport->egress.acl) {
+		vport->egress.acl = esw_acl_table_create(esw, vport->vport,
+							 MLX5_FLOW_NAMESPACE_ESW_EGRESS,
+							 table_size);
+		if (IS_ERR(vport->egress.acl)) {
+			err = PTR_ERR(vport->egress.acl);
+			vport->egress.acl = NULL;
+			goto out;
+		}
+
+		err = esw_acl_egress_lgcy_groups_create(esw, vport);
+		if (err)
+			goto out;
 	}
 
-	err = esw_acl_egress_lgcy_groups_create(esw, vport);
-	if (err)
-		goto out;
-
 	esw_debug(esw->dev,
 		  "vport[%d] configure egress rules, vlan(%d) qos(%d)\n",
 		  vport->vport, vport->info.vlan, vport->info.qos);
Signature · 1 line
2.26.2
#15 lore raw

PATCH 7/11 net/mlx5e: In skb build skip setting mark in switchdev mode

In reply to Saeed Mahameed

From: Maor Dickman <[email protected]>

sop_drop_qpn field in the cqe is used by two features, in SWITCHDEV mode to restore the chain id in case of a miss and in LEGACY mode to support skbedit mark action. In build RX skb, the skb mark field is set regardless of the configured mode which cause a corruption of the mark field in case of switchdev mode.

Fix by overriding the mark value back to 0 in the representor tc update skb flow.

Fixes 8f1e0b97cc70 ("net/mlx5: E-Switch, Mark miss packets with new chain id mapping")
---
 drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/ethernet/mellanox/mlx5/core/en/rep/tc.c +5 -0
index d29af7b9c695..76177f7c5ec2 100644
@@ -626,6 +626,11 @@ bool mlx5e_rep_tc_update_skb(struct mlx5_cqe64 *cqe,
 	if (!reg_c0)
 		return true;
 
+	/* If reg_c0 is not equal to the default flow tag then skb->mark
+	 * is not supported and must be reset back to 0.
+	 */
+	skb->mark = 0;
+
 	priv = netdev_priv(skb->dev);
 	esw = priv->mdev->priv.eswitch;
 
Signature · 1 line
2.26.2
#16 lore raw

PATCH 8/11 net/mlx5e: ethtool, Fix restriction of autoneg with 56G

In reply to Saeed Mahameed

From: Aya Levin <[email protected]>

Prior to this patch, configuring speed to 50G with autoneg off over devices supporting 50G per lane failed.
Support for 50G per lane introduced a new set of link-modes, on which driver always performed a speed validation as if only legacy link-modes were configured. Fix driver speed validation to force setting autoneg over 56G only if in legacy link-mode.

Fixes 3d7cadae51f1 ("net/mlx5e: ethtool, Fix analysis of speed setting")
---
 .../ethernet/mellanox/mlx5/core/en_ethtool.c  | 24 ++++++++++++++-----
 1 file changed, 18 insertions(+), 6 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c +18 -6
index d9076d543104..2d37742a888c 100644
@@ -1010,6 +1010,22 @@ static int mlx5e_get_link_ksettings(struct net_device *netdev,
 	return mlx5e_ethtool_get_link_ksettings(priv, link_ksettings);
 }
 
+static int mlx5e_speed_validate(struct net_device *netdev, bool ext,
+				const unsigned long link_modes, u8 autoneg)
+{
+	/* Extended link-mode has no speed limitations. */
+	if (ext)
+		return 0;
+
+	if ((link_modes & MLX5E_PROT_MASK(MLX5E_56GBASE_R4)) &&
+	    autoneg != AUTONEG_ENABLE) {
+		netdev_err(netdev, "%s: 56G link speed requires autoneg enabled\n",
+			   __func__);
+		return -EINVAL;
+	}
+	return 0;
+}
+
 static u32 mlx5e_ethtool2ptys_adver_link(const unsigned long *link_modes)
 {
 	u32 i, ptys_modes = 0;
@@ -1103,13 +1119,9 @@ int mlx5e_ethtool_set_link_ksettings(struct mlx5e_priv *priv,
 	link_modes = autoneg == AUTONEG_ENABLE ? ethtool2ptys_adver_func(adver) :
 		mlx5e_port_speed2linkmodes(mdev, speed, !ext);
 
-	if ((link_modes & MLX5E_PROT_MASK(MLX5E_56GBASE_R4)) &&
-	    autoneg != AUTONEG_ENABLE) {
-		netdev_err(priv->netdev, "%s: 56G link speed requires autoneg enabled\n",
-			   __func__);
-		err = -EINVAL;
+	err = mlx5e_speed_validate(priv->netdev, ext, link_modes, autoneg);
+	if (err)
 		goto out;
-	}
 
 	link_modes = link_modes & eproto.cap;
 	if (!link_modes) {
Signature · 1 line
2.26.2
#17 lore raw

PATCH 9/11 net/mlx5: Release devlink object if adev fails

In reply to Saeed Mahameed

From: Leon Romanovsky <[email protected]>

Add missed freeing previously allocated devlink object.

Fixes a925b5e309c9 ("net/mlx5: Register mlx5 devices to auxiliary virtual bus")
---
 drivers/net/ethernet/mellanox/mlx5/core/main.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/main.c +5 -2
index c08315b51fd3..ca6f2fc39ea0 100644
@@ -1368,8 +1368,10 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *id)
 			 MLX5_COREDEV_VF : MLX5_COREDEV_PF;
 
 	dev->priv.adev_idx = mlx5_adev_idx_alloc();
-	if (dev->priv.adev_idx < 0)
-		return dev->priv.adev_idx;
+	if (dev->priv.adev_idx < 0) {
+		err = dev->priv.adev_idx;
+		goto adev_init_err;
+	}
 
 	err = mlx5_mdev_init(dev, prof_sel);
 	if (err)
@@ -1403,6 +1405,7 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *id)
 	mlx5_mdev_uninit(dev);
 mdev_init_err:
 	mlx5_adev_idx_free(dev->priv.adev_idx);
+adev_init_err:
 	mlx5_devlink_free(devlink);
 
 	return err;
Signature · 1 line
2.26.2
#18 lore raw

PATCH 10/11 net/mlx5e: Fix two double free cases

In reply to Saeed Mahameed

From: Dinghao Liu <[email protected]>

mlx5e_create_ttc_table_groups() frees ft->g on failure of kvzalloc(), but such failure will be caught by its caller in mlx5e_create_ttc_table() and ft->g will be freed again in mlx5e_destroy_flow_table(). The same issue also occurs in mlx5e_create_ttc_table_groups(). Set ft->g to NULL after kfree() to avoid double free.

Fixes 7b3722fa9ef6 ("net/mlx5e: Support RSS for GRE tunneled packets")
Fixes 33cfaaa8f36f ("net/mlx5e: Split the main flow steering table")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_fs.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/net/ethernet/mellanox/mlx5/core/en_fs.c +2 -0
index fa8149f6eb08..44a2dfbc3853 100644
@@ -942,6 +942,7 @@ static int mlx5e_create_ttc_table_groups(struct mlx5e_ttc_table *ttc,
 	in = kvzalloc(inlen, GFP_KERNEL);
 	if (!in) {
 		kfree(ft->g);
+		ft->g = NULL;
 		return -ENOMEM;
 	}
 
@@ -1087,6 +1088,7 @@ static int mlx5e_create_inner_ttc_table_groups(struct mlx5e_ttc_table *ttc)
 	in = kvzalloc(inlen, GFP_KERNEL);
 	if (!in) {
 		kfree(ft->g);
+		ft->g = NULL;
 		return -ENOMEM;
 	}
 
Signature · 1 line
2.26.2
#19 lore raw

PATCH 11/11 net/mlx5e: Fix memleak in mlx5e_create_l2_table_groups

In reply to Saeed Mahameed

From: Dinghao Liu <[email protected]>

When mlx5_create_flow_group() fails, ft->g should be freed just like when kvzalloc() fails. The caller of mlx5e_create_l2_table_groups() does not catch this issue on failure, which leads to memleak.

Fixes 33cfaaa8f36f ("net/mlx5e: Split the main flow steering table")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_fs.c | 1 +
 1 file changed, 1 insertion(+)
drivers/net/ethernet/mellanox/mlx5/core/en_fs.c +1 -0
index 44a2dfbc3853..e02e5895703d 100644
@@ -1392,6 +1392,7 @@ static int mlx5e_create_l2_table_groups(struct mlx5e_l2_table *l2_table)
 	ft->g[ft->num_groups] = NULL;
 	mlx5e_destroy_groups(ft);
 	kvfree(in);
+	kfree(ft->g);
 
 	return err;
 }
Signature · 1 line
2.26.2