Skip to contentSkip to conversation

mlx5 misc fixes 2025-07-10

PATCH Applied Tariq Toukan 3 participants 8 messages Jul 10, 2025 – Jul 11, 2025

Subsystems (5)
lore
Share & tools
lore

Tariq Toukan

author maintainer
#0 lore raw

cover mlx5 misc fixes 2025-07-10

Hi,

This small patchset provides misc bug fixes from the team to the mlx5 core and EN drivers.

Thanks,
Tariq.

Carolina Jubran (2):
net/mlx5: Reset bw_share field when changing a node's parent net/mlx5e: Fix race between DIM disable and net_dim()

Jianbo Liu (1):
net/mlx5e: Add new prio for promiscuous mode

 drivers/net/ethernet/mellanox/mlx5/core/en/fs.h   |  9 +++++++--
 drivers/net/ethernet/mellanox/mlx5/core/en_dim.c  |  4 ++--
 drivers/net/ethernet/mellanox/mlx5/core/en_fs.c   |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c |  1 +
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 13 +++++++++----
 5 files changed, 20 insertions(+), 9 deletions(-)

base-commit: 0fda5ccf5425c7b92eaca868a3fba8a3c9f8b746

Signature · 1 line
2.31.1
author maintainer
#1 lore raw

PATCH 1/3 net/mlx5: Reset bw_share field when changing a node's parent

From: Carolina Jubran <[email protected]>

When changing a node's parent, its scheduling element is destroyed and re-created with bw_share 0. However, the node's bw_share field was not updated accordingly.

Set the node's bw_share to 0 after re-creation to keep the software state in sync with the firmware configuration.

Fixes 9c7bbf4c3304 ("net/mlx5: Add support for setting parent of nodes")
---
 drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c | 1 +
 1 file changed, 1 insertion(+)
drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +1 -0
index b6ae384396b3..ad9f6fca9b6a 100644
@@ -1076,6 +1076,7 @@ static int esw_qos_vports_node_update_parent(struct mlx5_esw_sched_node *node,
 		return err;
 	}
 	esw_qos_node_set_parent(node, parent);
+	node->bw_share = 0;
 
 	return 0;
 }
Signature · 1 line
2.31.1
Reviewed-by on v1 1/3
#2 lore raw

Re: [PATCH net 1/3] net/mlx5: Reset bw_share field when changing a node's parent

Quoted 23 lines · Tariq Toukan

From: Carolina Jubran <[email protected]>

When changing a node's parent, its scheduling element is destroyed and re-created with bw_share 0. However, the node's bw_share field was not updated accordingly.

Set the node's bw_share to 0 after re-creation to keep the software state in sync with the firmware configuration.

Fixes 9c7bbf4c3304 ("net/mlx5: Add support for setting parent of nodes")
---
 drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c | 1 +
 1 file changed, 1 insertion(+)
drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +0 -0
index b6ae384396b3..ad9f6fca9b6a 100644

drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c static int esw_qos_vports_node_update_parent(struct mlx5_esw_sched_node *node, · line 1080

Quoted 5 lines
drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +1 -0
@@ -1076,6 +1076,7 @@ static int esw_qos_vports_node_update_parent(struct mlx5_esw_sched_node *node,
 		return err;
 	}
 	esw_qos_node_set_parent(node, parent);
+	node->bw_share = 0;
 

Makes sense.

Quoted 2 lines
drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +0 -0
 	return 0;
 }
author maintainer
#3 lore raw

PATCH 2/3 net/mlx5e: Fix race between DIM disable and net_dim()

In reply to Tariq Toukan

From: Carolina Jubran <[email protected]>

There's a race between disabling DIM and NAPI callbacks using the dim pointer on the RQ or SQ.

If NAPI checks the DIM state bit and sees it still set, it assumes rq->dim or sq->dim is valid. But if DIM gets disabled right after that check, the pointer might already be set to NULL, leading to a NULL pointer dereference in net_dim().

Fix this by calling synchronize_net() before freeing the DIM context. This ensures all in-progress NAPI callbacks are finished before the pointer is cleared.

Kernel log:

BUG: kernel NULL pointer dereference, address: 0000000000000000
...
RIP: 0010:net_dim+0x23/0x190
...
Call Trace:
 <TASK>
 ? __die+0x20/0x60
 ? page_fault_oops+0x150/0x3e0
 ? common_interrupt+0xf/0xa0
 ? sysvec_call_function_single+0xb/0x90
 ? exc_page_fault+0x74/0x130
 ? asm_exc_page_fault+0x22/0x30
 ? net_dim+0x23/0x190
 ? mlx5e_poll_ico_cq+0x41/0x6f0 [mlx5_core]
 ? sysvec_apic_timer_interrupt+0xb/0x90
 mlx5e_handle_rx_dim+0x92/0xd0 [mlx5_core]
 mlx5e_napi_poll+0x2cd/0xac0 [mlx5_core]
 ? mlx5e_poll_ico_cq+0xe5/0x6f0 [mlx5_core]
 busy_poll_stop+0xa2/0x200
 ? mlx5e_napi_poll+0x1d9/0xac0 [mlx5_core]
 ? mlx5e_trigger_irq+0x130/0x130 [mlx5_core]
 __napi_busy_loop+0x345/0x3b0
 ? sysvec_call_function_single+0xb/0x90
 ? asm_sysvec_call_function_single+0x16/0x20
 ? sysvec_apic_timer_interrupt+0xb/0x90
 ? pcpu_free_area+0x1e4/0x2e0
 napi_busy_loop+0x11/0x20
 xsk_recvmsg+0x10c/0x130
 sock_recvmsg+0x44/0x70
 __sys_recvfrom+0xbc/0x130
 ? __schedule+0x398/0x890
 __x64_sys_recvfrom+0x20/0x30
 do_syscall_64+0x4c/0x100
 entry_SYSCALL_64_after_hwframe+0x4b/0x53
...
---[ end trace 0000000000000000 ]---
...
---[ end Kernel panic - not syncing: Fatal exception in interrupt ]---
Fixes 445a25f6e1a2 ("net/mlx5e: Support updating coalescing configuration without resetting channels")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_dim.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c +2 -2
index 298bb74ec5e9..d1d629697e28 100644
@@ -113,7 +113,7 @@ int mlx5e_dim_rx_change(struct mlx5e_rq *rq, bool enable)
 		__set_bit(MLX5E_RQ_STATE_DIM, &rq->state);
 	} else {
 		__clear_bit(MLX5E_RQ_STATE_DIM, &rq->state);
-
+		synchronize_net();
 		mlx5e_dim_disable(rq->dim);
 		rq->dim = NULL;
 	}
@@ -140,7 +140,7 @@ int mlx5e_dim_tx_change(struct mlx5e_txqsq *sq, bool enable)
 		__set_bit(MLX5E_SQ_STATE_DIM, &sq->state);
 	} else {
 		__clear_bit(MLX5E_SQ_STATE_DIM, &sq->state);
-
+		synchronize_net();
 		mlx5e_dim_disable(sq->dim);
 		sq->dim = NULL;
 	}
Signature · 1 line
2.31.1
Reviewed-by on v1 2/3
#4 lore raw

Re: [PATCH net 2/3] net/mlx5e: Fix race between DIM disable and net_dim()

Quoted 46 lines · Tariq Toukan

From: Carolina Jubran <[email protected]>

There's a race between disabling DIM and NAPI callbacks using the dim pointer on the RQ or SQ.

If NAPI checks the DIM state bit and sees it still set, it assumes rq->dim or sq->dim is valid. But if DIM gets disabled right after that check, the pointer might already be set to NULL, leading to a NULL pointer dereference in net_dim().

Fix this by calling synchronize_net() before freeing the DIM context. This ensures all in-progress NAPI callbacks are finished before the pointer is cleared.

Kernel log:

BUG: kernel NULL pointer dereference, address: 0000000000000000
...
RIP: 0010:net_dim+0x23/0x190
...
Call Trace:
 <TASK>
 ? __die+0x20/0x60
 ? page_fault_oops+0x150/0x3e0
 ? common_interrupt+0xf/0xa0
 ? sysvec_call_function_single+0xb/0x90
 ? exc_page_fault+0x74/0x130
 ? asm_exc_page_fault+0x22/0x30
 ? net_dim+0x23/0x190
 ? mlx5e_poll_ico_cq+0x41/0x6f0 [mlx5_core]
 ? sysvec_apic_timer_interrupt+0xb/0x90
 mlx5e_handle_rx_dim+0x92/0xd0 [mlx5_core]
 mlx5e_napi_poll+0x2cd/0xac0 [mlx5_core]
 ? mlx5e_poll_ico_cq+0xe5/0x6f0 [mlx5_core]
 busy_poll_stop+0xa2/0x200
 ? mlx5e_napi_poll+0x1d9/0xac0 [mlx5_core]
 ? mlx5e_trigger_irq+0x130/0x130 [mlx5_core]
 __napi_busy_loop+0x345/0x3b0
 ? sysvec_call_function_single+0xb/0x90
 ? asm_sysvec_call_function_single+0x16/0x20
 ? sysvec_apic_timer_interrupt+0xb/0x90
 ? pcpu_free_area+0x1e4/0x2e0
 napi_busy_loop+0x11/0x20
 xsk_recvmsg+0x10c/0x130
 sock_recvmsg+0x44/0x70
 __sys_recvfrom+0xbc/0x130
Quoted 9 lines
 ? __schedule+0x398/0x890
 __x64_sys_recvfrom+0x20/0x30
 do_syscall_64+0x4c/0x100
 entry_SYSCALL_64_after_hwframe+0x4b/0x53
...
---[ end trace 0000000000000000 ]---
...
---[ end Kernel panic - not syncing: Fatal exception in interrupt ]---
Quoted 13 lines
Fixes 445a25f6e1a2 ("net/mlx5e: Support updating coalescing configuration without resetting channels")
---
 drivers/net/ethernet/mellanox/mlx5/core/en_dim.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c +0 -0
index 298bb74ec5e9..d1d629697e28 100644

drivers/net/ethernet/mellanox/mlx5/core/en_dim.c int mlx5e_dim_rx_change(struct mlx5e_rq *rq, bool enable) · line 116

Quoted 5 lines
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c +1 -1
@@ -113,7 +113,7 @@ int mlx5e_dim_rx_change(struct mlx5e_rq *rq, bool enable)
 		__set_bit(MLX5E_RQ_STATE_DIM, &rq->state);
 	} else {
 		__clear_bit(MLX5E_RQ_STATE_DIM, &rq->state);
-
+		synchronize_net();

We've requested NAPI be disabled by this point, and just need to guarantee that the already running threads close. Makes sense.

Quoted 12 lines
drivers/net/ethernet/mellanox/mlx5/core/en_dim.c +1 -1
 		mlx5e_dim_disable(rq->dim);
 		rq->dim = NULL;
 	}
@@ -140,7 +140,7 @@ int mlx5e_dim_tx_change(struct mlx5e_txqsq *sq, bool enable)
 		__set_bit(MLX5E_SQ_STATE_DIM, &sq->state);
 	} else {
 		__clear_bit(MLX5E_SQ_STATE_DIM, &sq->state);
-
+		synchronize_net();
 		mlx5e_dim_disable(sq->dim);
 		sq->dim = NULL;
 	}
author maintainer
#5 lore raw

PATCH 3/3 net/mlx5e: Add new prio for promiscuous mode

In reply to Tariq Toukan

From: Jianbo Liu <[email protected]>

An optimization for promiscuous mode adds a high-priority steering table with a single catch-all rule to steer all traffic directly to the TTC table.

However, a gap exists between the creation of this table and the insertion of the catch-all rule. Packets arriving in this brief window would miss as no rule was inserted yet, unnecessarily incrementing the 'rx_steer_missed_packets' counter and dropped.

This patch resolves the issue by introducing a new prio for this table, placing it between MLX5E_TC_PRIO and MLX5E_NIC_PRIO. By doing so, packets arriving during the window now fall through to the next prio (at MLX5E_NIC_PRIO) instead of being dropped.

Fixes 1c46d7409f30 ("net/mlx5e: Optimize promiscuous mode")
---
 drivers/net/ethernet/mellanox/mlx5/core/en/fs.h   |  9 +++++++--
 drivers/net/ethernet/mellanox/mlx5/core/en_fs.c   |  2 +-
 drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 13 +++++++++----
 3 files changed, 17 insertions(+), 7 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en/fs.h +7 -2
index b5c3a2a9d2a5..9560fcba643f 100644
@@ -18,7 +18,8 @@ enum {
 
 enum {
 	MLX5E_TC_PRIO = 0,
-	MLX5E_NIC_PRIO
+	MLX5E_PROMISC_PRIO,
+	MLX5E_NIC_PRIO,
 };
 
 struct mlx5e_flow_table {
@@ -68,9 +69,13 @@ struct mlx5e_l2_table {
 				 MLX5_HASH_FIELD_SEL_DST_IP   |\
 				 MLX5_HASH_FIELD_SEL_IPSEC_SPI)
 
-/* NIC prio FTS */
+/* NIC promisc FT level */
 enum {
 	MLX5E_PROMISC_FT_LEVEL,
+};
+
+/* NIC prio FTS */
+enum {
 	MLX5E_VLAN_FT_LEVEL,
 	MLX5E_L2_FT_LEVEL,
 	MLX5E_TTC_FT_LEVEL,
drivers/net/ethernet/mellanox/mlx5/core/en_fs.c +1 -1
index 04a969128161..265c4ca85f7d 100644
@@ -780,7 +780,7 @@ static int mlx5e_create_promisc_table(struct mlx5e_flow_steering *fs)
 	ft_attr.max_fte = MLX5E_PROMISC_TABLE_SIZE;
 	ft_attr.autogroup.max_num_groups = 1;
 	ft_attr.level = MLX5E_PROMISC_FT_LEVEL;
-	ft_attr.prio = MLX5E_NIC_PRIO;
+	ft_attr.prio = MLX5E_PROMISC_PRIO;
 
 	ft->t = mlx5_create_auto_grouped_flow_table(fs->ns, &ft_attr);
 	if (IS_ERR(ft->t)) {
drivers/net/ethernet/mellanox/mlx5/core/fs_core.c +9 -4
index a8046200d376..3dd9a6f40709 100644
@@ -113,13 +113,16 @@
 #define ETHTOOL_PRIO_NUM_LEVELS 1
 #define ETHTOOL_NUM_PRIOS 11
 #define ETHTOOL_MIN_LEVEL (KERNEL_MIN_LEVEL + ETHTOOL_NUM_PRIOS)
-/* Promiscuous, Vlan, mac, ttc, inner ttc, {UDP/ANY/aRFS/accel/{esp, esp_err}}, IPsec policy,
+/* Vlan, mac, ttc, inner ttc, {UDP/ANY/aRFS/accel/{esp, esp_err}}, IPsec policy,
  * {IPsec RoCE MPV,Alias table},IPsec RoCE policy
  */
-#define KERNEL_NIC_PRIO_NUM_LEVELS 11
+#define KERNEL_NIC_PRIO_NUM_LEVELS 10
 #define KERNEL_NIC_NUM_PRIOS 1
-/* One more level for tc */
-#define KERNEL_MIN_LEVEL (KERNEL_NIC_PRIO_NUM_LEVELS + 1)
+/* One more level for tc, and one more for promisc */
+#define KERNEL_MIN_LEVEL (KERNEL_NIC_PRIO_NUM_LEVELS + 2)
+
+#define KERNEL_NIC_PROMISC_NUM_PRIOS 1
+#define KERNEL_NIC_PROMISC_NUM_LEVELS 1
 
 #define KERNEL_NIC_TC_NUM_PRIOS  1
 #define KERNEL_NIC_TC_NUM_LEVELS 3
@@ -187,6 +190,8 @@ static struct init_tree_node {
 			   ADD_NS(MLX5_FLOW_TABLE_MISS_ACTION_DEF,
 				  ADD_MULTIPLE_PRIO(KERNEL_NIC_TC_NUM_PRIOS,
 						    KERNEL_NIC_TC_NUM_LEVELS),
+				  ADD_MULTIPLE_PRIO(KERNEL_NIC_PROMISC_NUM_PRIOS,
+						    KERNEL_NIC_PROMISC_NUM_LEVELS),
 				  ADD_MULTIPLE_PRIO(KERNEL_NIC_NUM_PRIOS,
 						    KERNEL_NIC_PRIO_NUM_LEVELS))),
 		  ADD_PRIO(0, BY_PASS_MIN_LEVEL, 0, FS_CHAINING_CAPS,
Signature · 1 line
2.31.1
Reviewed-by on v1 3/3
#6 lore raw

Re: [PATCH net 3/3] net/mlx5e: Add new prio for promiscuous mode

Quoted 16 lines · Tariq Toukan

From: Jianbo Liu <[email protected]>

An optimization for promiscuous mode adds a high-priority steering table with a single catch-all rule to steer all traffic directly to the TTC table.

However, a gap exists between the creation of this table and the insertion of the catch-all rule. Packets arriving in this brief window would miss as no rule was inserted yet, unnecessarily incrementing the 'rx_steer_missed_packets' counter and dropped.

This patch resolves the issue by introducing a new prio for this table, placing it between MLX5E_TC_PRIO and MLX5E_NIC_PRIO. By doing so, packets arriving during the window now fall through to the next prio (at MLX5E_NIC_PRIO) instead of being dropped.

Quoted 5 lines
Fixes 1c46d7409f30 ("net/mlx5e: Optimize promiscuous mode")

---

#7 lore raw
In reply to Tariq Toukan
This series was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This series was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:

Quoted 9 lines · On Thu, 10 Jul 2025 16:53:41 +0300 you wrote:

Hi,

This small patchset provides misc bug fixes from the team to the mlx5 core and EN drivers.

Thanks,
Tariq.

[...]

Here is the summary with links:
- [net,1/3] net/mlx5: Reset bw_share field when changing a node's parent

    https://git.kernel.org/netdev/net/c/f7b764668940
  - [net,2/3] net/mlx5e: Fix race between DIM disable and net_dim()
    https://git.kernel.org/netdev/net/c/eb41a264a3a5
  - [net,3/3] net/mlx5e: Add new prio for promiscuous mode
    https://git.kernel.org/netdev/net/c/4c9fce56fa70

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html