[PATCH net 0/2] mlx5 misc fixes 2025-12-01

[PATCH net 0/2] mlx5 misc fixes 2025-12-01

[Tariq Toukan]

Hi,

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

Thanks,
Tariq.

Cosmin Ratiu (1):
  net/mlx5e: Avoid unregistering PSP twice

Moshe Shemesh (1):
  net/mlx5: make enable_mpesw idempotent

 drivers/net/ethernet/mellanox/mlx5/core/en_main.c   |  1 -
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c | 11 +++++++++--
 2 files changed, 9 insertions(+), 3 deletions(-)


base-commit: e5235eb6cfe02a51256013a78f7b28779a7740d5
-- 
2.31.1

[PATCH net 1/2] net/mlx5: make enable_mpesw idempotent

[Tariq Toukan]

From: Moshe Shemesh <moshe@nvidia.com>

The enable_mpesw() function returns -EINVAL if ldev->mode is not
MLX5_LAG_MODE_NONE. This means attempting to enable MPESW mode when it's
already enabled will fail. In contrast, disable_mpesw() properly checks
if the mode is MLX5_LAG_MODE_MPESW before proceeding, making it
naturally idempotent and safe to call multiple times.

Fix enable_mpesw() to return success if mpesw is already enabled.

Fixes: a32327a3a02c ("net/mlx5: Lag, Control MultiPort E-Switch single FDB mode")
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Shay Drori <shayd@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index aad52d3a90e6..2d86af8f0d9b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -67,12 +67,19 @@ static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
 
 static int enable_mpesw(struct mlx5_lag *ldev)
 {
-	int idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
 	struct mlx5_core_dev *dev0;
 	int err;
+	int idx;
 	int i;
 
-	if (idx < 0 || ldev->mode != MLX5_LAG_MODE_NONE)
+	if (ldev->mode == MLX5_LAG_MODE_MPESW)
+		return 0;
+
+	if (ldev->mode != MLX5_LAG_MODE_NONE)
+		return -EINVAL;
+
+	idx = mlx5_lag_get_dev_index_by_seq(ldev, MLX5_LAG_P1);
+	if (idx < 0)
 		return -EINVAL;
 
 	dev0 = ldev->pf[idx].dev;
-- 
2.31.1

[PATCH net 2/2] net/mlx5e: Avoid unregistering PSP twice

[Tariq Toukan]

From: Cosmin Ratiu <cratiu@nvidia.com>

PSP is unregistered twice in:
_mlx5e_remove -> mlx5e_psp_unregister
mlx5e_nic_cleanup -> mlx5e_psp_unregister

This leads to a refcount underflow in some conditions:
------------[ cut here ]------------
refcount_t: underflow; use-after-free.
WARNING: CPU: 2 PID: 1694 at lib/refcount.c:28 refcount_warn_saturate+0xd8/0xe0
[...]
 mlx5e_psp_unregister+0x26/0x50 [mlx5_core]
 mlx5e_nic_cleanup+0x26/0x90 [mlx5_core]
 mlx5e_remove+0xe6/0x1f0 [mlx5_core]
 auxiliary_bus_remove+0x18/0x30
 device_release_driver_internal+0x194/0x1f0
 bus_remove_device+0xc6/0x130
 device_del+0x159/0x3c0
 mlx5_rescan_drivers_locked+0xbc/0x2a0 [mlx5_core]
[...]

Do not directly remove psp from the _mlx5e_remove path, the PSP cleanup
happens as part of profile cleanup.

Fixes: 89ee2d92f66c ("net/mlx5e: Support PSP offload functionality")
Signed-off-by: Cosmin Ratiu <cratiu@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 5e17eae81f4b..1545f9c008f4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -6805,7 +6805,6 @@ static void _mlx5e_remove(struct auxiliary_device *adev)
 	 * is already unregistered before changing to NIC profile.
 	 */
 	if (priv->netdev->reg_state == NETREG_REGISTERED) {
-		mlx5e_psp_unregister(priv);
 		unregister_netdev(priv->netdev);
 		_mlx5e_suspend(adev, false);
 	} else {
-- 
2.31.1

Re: [PATCH net 1/2] net/mlx5: make enable_mpesw idempotent

[Simon Horman]

On Mon, Dec 01, 2025 at 05:13:27PM +0200, Tariq Toukan wrote:
> From: Moshe Shemesh <moshe@nvidia.com>
> 
> The enable_mpesw() function returns -EINVAL if ldev->mode is not
> MLX5_LAG_MODE_NONE. This means attempting to enable MPESW mode when it's
> already enabled will fail. In contrast, disable_mpesw() properly checks
> if the mode is MLX5_LAG_MODE_MPESW before proceeding, making it
> naturally idempotent and safe to call multiple times.
> 
> Fix enable_mpesw() to return success if mpesw is already enabled.
> 
> Fixes: a32327a3a02c ("net/mlx5: Lag, Control MultiPort E-Switch single FDB mode")
> Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
> Reviewed-by: Shay Drori <shayd@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>

Reviewed-by: Simon Horman <horms@kernel.org>

Re: [PATCH net 2/2] net/mlx5e: Avoid unregistering PSP twice

[Simon Horman]

On Mon, Dec 01, 2025 at 05:13:28PM +0200, Tariq Toukan wrote:
> From: Cosmin Ratiu <cratiu@nvidia.com>
> 
> PSP is unregistered twice in:
> _mlx5e_remove -> mlx5e_psp_unregister
> mlx5e_nic_cleanup -> mlx5e_psp_unregister
> 
> This leads to a refcount underflow in some conditions:
> ------------[ cut here ]------------
> refcount_t: underflow; use-after-free.
> WARNING: CPU: 2 PID: 1694 at lib/refcount.c:28 refcount_warn_saturate+0xd8/0xe0
> [...]
>  mlx5e_psp_unregister+0x26/0x50 [mlx5_core]
>  mlx5e_nic_cleanup+0x26/0x90 [mlx5_core]
>  mlx5e_remove+0xe6/0x1f0 [mlx5_core]
>  auxiliary_bus_remove+0x18/0x30
>  device_release_driver_internal+0x194/0x1f0
>  bus_remove_device+0xc6/0x130
>  device_del+0x159/0x3c0
>  mlx5_rescan_drivers_locked+0xbc/0x2a0 [mlx5_core]
> [...]
> 
> Do not directly remove psp from the _mlx5e_remove path, the PSP cleanup
> happens as part of profile cleanup.
> 
> Fixes: 89ee2d92f66c ("net/mlx5e: Support PSP offload functionality")
> Signed-off-by: Cosmin Ratiu <cratiu@nvidia.com>
> Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>

Reviewed-by: Simon Horman <horms@kernel.org>

Re: [PATCH net 0/2] mlx5 misc fixes 2025-12-01

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net.git (main)
by Paolo Abeni <pabeni@redhat.com>:

On Mon, 1 Dec 2025 17:13:26 +0200 you wrote:
> Hi,
> 
> This small patchset provides misc bug fixes from the team to the mlx5
> core and Eth drivers.
> 
> Thanks,
> Tariq.
> 
> [...]

Here is the summary with links:
  - [net,1/2] net/mlx5: make enable_mpesw idempotent
    https://git.kernel.org/netdev/net/c/cd7671ef4cf2
  - [net,2/2] net/mlx5e: Avoid unregistering PSP twice
    https://git.kernel.org/netdev/net/c/35e93736f699

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