[PATCH net 0/5] mlx5 misc fixes 2026-02-24

[PATCH net 0/5] mlx5 misc fixes 2026-02-24

[Tariq Toukan]

Hi,

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

Thanks,
Tariq.

Jianbo Liu (1):
  net/mlx5e: Fix "scheduling while atomic" in IPsec MAC address query

Shay Drory (4):
  net/mlx5: DR, Fix circular locking dependency in dump
  net/mlx5: LAG, disable MPESW in lag_disable_change()
  net/mlx5: E-switch, Clear legacy flag when moving to switchdev
  net/mlx5: Fix missing devlink lock in SRIOV enable error path

 drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c  | 3 +--
 .../net/ethernet/mellanox/mlx5/core/eswitch_offloads.c    | 2 ++
 drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c         | 8 ++++++--
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c       | 8 ++++----
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h       | 5 +++++
 drivers/net/ethernet/mellanox/mlx5/core/sriov.c           | 2 ++
 .../net/ethernet/mellanox/mlx5/core/steering/sws/dr_dbg.c | 4 ++--
 7 files changed, 22 insertions(+), 10 deletions(-)


base-commit: 82aec772fca2223bc5774bd9af486fd95766e578
-- 
2.44.0

[PATCH net 1/5] net/mlx5: DR, Fix circular locking dependency in dump

[Tariq Toukan]

From: Shay Drory <shayd@nvidia.com>

Fix a circular locking dependency between dbg_mutex and the domain
rx/tx mutexes that could lead to a deadlock.

The dump path in dr_dump_domain_all() was acquiring locks in the order:
  dbg_mutex -> rx.mutex -> tx.mutex

While the table/matcher creation paths acquire locks in the order:
  rx.mutex -> tx.mutex -> dbg_mutex

This inverted lock ordering creates a circular dependency. Fix this by
changing dr_dump_domain_all() to acquire the domain lock before
dbg_mutex, matching the order used in mlx5dr_table_create() and
mlx5dr_matcher_create().

Lockdep splat:
 ======================================================
 WARNING: possible circular locking dependency detected
 6.19.0-rc6net_next_e817c4e #1 Not tainted
 ------------------------------------------------------
 sos/30721 is trying to acquire lock:
 ffff888102df5900 (&dmn->info.rx.mutex){+.+.}-{4:4}, at:
dr_dump_start+0x131/0x450 [mlx5_core]

 but task is already holding lock:
 ffff888102df5bc0 (&dmn->dump_info.dbg_mutex){+.+.}-{4:4}, at:
dr_dump_start+0x10b/0x450 [mlx5_core]

 which lock already depends on the new lock.

 the existing dependency chain (in reverse order) is:

 -> #2 (&dmn->dump_info.dbg_mutex){+.+.}-{4:4}:
        __mutex_lock+0x91/0x1060
        mlx5dr_matcher_create+0x377/0x5e0 [mlx5_core]
        mlx5_cmd_dr_create_flow_group+0x62/0xd0 [mlx5_core]
        mlx5_create_flow_group+0x113/0x1c0 [mlx5_core]
        mlx5_chains_create_prio+0x453/0x2290 [mlx5_core]
        mlx5_chains_get_table+0x2e2/0x980 [mlx5_core]
        esw_chains_create+0x1e6/0x3b0 [mlx5_core]
        esw_create_offloads_fdb_tables.cold+0x62/0x63f [mlx5_core]
        esw_offloads_enable+0x76f/0xd20 [mlx5_core]
        mlx5_eswitch_enable_locked+0x35a/0x500 [mlx5_core]
        mlx5_devlink_eswitch_mode_set+0x561/0x950 [mlx5_core]
        devlink_nl_eswitch_set_doit+0x67/0xe0
        genl_family_rcv_msg_doit+0xe0/0x130
        genl_rcv_msg+0x188/0x290
        netlink_rcv_skb+0x4b/0xf0
        genl_rcv+0x24/0x40
        netlink_unicast+0x1ed/0x2c0
        netlink_sendmsg+0x210/0x450
        __sock_sendmsg+0x38/0x60
        __sys_sendto+0x119/0x180
        __x64_sys_sendto+0x20/0x30
        do_syscall_64+0x70/0xd00
        entry_SYSCALL_64_after_hwframe+0x4b/0x53

 -> #1 (&dmn->info.tx.mutex){+.+.}-{4:4}:
        __mutex_lock+0x91/0x1060
        mlx5dr_table_create+0x11d/0x530 [mlx5_core]
        mlx5_cmd_dr_create_flow_table+0x62/0x140 [mlx5_core]
        __mlx5_create_flow_table+0x46f/0x960 [mlx5_core]
        mlx5_create_flow_table+0x16/0x20 [mlx5_core]
        esw_create_offloads_fdb_tables+0x136/0x240 [mlx5_core]
        esw_offloads_enable+0x76f/0xd20 [mlx5_core]
        mlx5_eswitch_enable_locked+0x35a/0x500 [mlx5_core]
        mlx5_devlink_eswitch_mode_set+0x561/0x950 [mlx5_core]
        devlink_nl_eswitch_set_doit+0x67/0xe0
        genl_family_rcv_msg_doit+0xe0/0x130
        genl_rcv_msg+0x188/0x290
        netlink_rcv_skb+0x4b/0xf0
        genl_rcv+0x24/0x40
        netlink_unicast+0x1ed/0x2c0
        netlink_sendmsg+0x210/0x450
        __sock_sendmsg+0x38/0x60
        __sys_sendto+0x119/0x180
        __x64_sys_sendto+0x20/0x30
        do_syscall_64+0x70/0xd00
        entry_SYSCALL_64_after_hwframe+0x4b/0x53

 -> #0 (&dmn->info.rx.mutex){+.+.}-{4:4}:
        __lock_acquire+0x18b6/0x2eb0
        lock_acquire+0xd3/0x2c0
        __mutex_lock+0x91/0x1060
        dr_dump_start+0x131/0x450 [mlx5_core]
        seq_read_iter+0xe3/0x410
        seq_read+0xfb/0x130
        full_proxy_read+0x53/0x80
        vfs_read+0xba/0x330
        ksys_read+0x65/0xe0
        do_syscall_64+0x70/0xd00
        entry_SYSCALL_64_after_hwframe+0x4b/0x53

  Possible unsafe locking scenario:

        CPU0                    CPU1
        ----                    ----
   lock(&dmn->dump_info.dbg_mutex);
                                lock(&dmn->info.tx.mutex);
                                lock(&dmn->dump_info.dbg_mutex);
   lock(&dmn->info.rx.mutex);

                   *** DEADLOCK ***

Fixes: 9222f0b27da2 ("net/mlx5: DR, Add support for dumping steering info")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Reviewed-by: Alex Vesker <valex@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_dbg.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_dbg.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_dbg.c
index 8803fa071c50..18362e9c3314 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_dbg.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/sws/dr_dbg.c
@@ -1051,8 +1051,8 @@ static int dr_dump_domain_all(struct seq_file *file, struct mlx5dr_domain *dmn)
 	struct mlx5dr_table *tbl;
 	int ret;
 
-	mutex_lock(&dmn->dump_info.dbg_mutex);
 	mlx5dr_domain_lock(dmn);
+	mutex_lock(&dmn->dump_info.dbg_mutex);
 
 	ret = dr_dump_domain(file, dmn);
 	if (ret < 0)
@@ -1065,8 +1065,8 @@ static int dr_dump_domain_all(struct seq_file *file, struct mlx5dr_domain *dmn)
 	}
 
 unlock_mutex:
-	mlx5dr_domain_unlock(dmn);
 	mutex_unlock(&dmn->dump_info.dbg_mutex);
+	mlx5dr_domain_unlock(dmn);
 	return ret;
 }
 
-- 
2.44.0

[PATCH net 2/5] net/mlx5: LAG, disable MPESW in lag_disable_change()

[Tariq Toukan]

From: Shay Drory <shayd@nvidia.com>

mlx5_lag_disable_change() unconditionally called mlx5_disable_lag() when
LAG was active, which is incorrect for MLX5_LAG_MODE_MPESW.
Hnece, call mlx5_disable_mpesw() when running in MPESW mode.

Fixes: a32327a3a02c ("net/mlx5: Lag, Control MultiPort E-Switch single FDB mode")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c   | 8 ++++++--
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c | 8 ++++----
 drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h | 5 +++++
 3 files changed, 15 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
index 9fe47c836ebd..859f042caf79 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c
@@ -1869,8 +1869,12 @@ void mlx5_lag_disable_change(struct mlx5_core_dev *dev)
 	mutex_lock(&ldev->lock);
 
 	ldev->mode_changes_in_progress++;
-	if (__mlx5_lag_is_active(ldev))
-		mlx5_disable_lag(ldev);
+	if (__mlx5_lag_is_active(ldev)) {
+		if (ldev->mode == MLX5_LAG_MODE_MPESW)
+			mlx5_lag_disable_mpesw(ldev);
+		else
+			mlx5_disable_lag(ldev);
+	}
 
 	mutex_unlock(&ldev->lock);
 	mlx5_devcom_comp_unlock(dev->priv.hca_devcom_comp);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
index 04762562d7d9..a63d48d18878 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.c
@@ -65,7 +65,7 @@ static int mlx5_mpesw_metadata_set(struct mlx5_lag *ldev)
 	return err;
 }
 
-static int enable_mpesw(struct mlx5_lag *ldev)
+static int mlx5_lag_enable_mpesw(struct mlx5_lag *ldev)
 {
 	struct mlx5_core_dev *dev0;
 	int err;
@@ -126,7 +126,7 @@ static int enable_mpesw(struct mlx5_lag *ldev)
 	return err;
 }
 
-static void disable_mpesw(struct mlx5_lag *ldev)
+void mlx5_lag_disable_mpesw(struct mlx5_lag *ldev)
 {
 	if (ldev->mode == MLX5_LAG_MODE_MPESW) {
 		mlx5_mpesw_metadata_cleanup(ldev);
@@ -152,9 +152,9 @@ static void mlx5_mpesw_work(struct work_struct *work)
 	}
 
 	if (mpesww->op == MLX5_MPESW_OP_ENABLE)
-		mpesww->result = enable_mpesw(ldev);
+		mpesww->result = mlx5_lag_enable_mpesw(ldev);
 	else if (mpesww->op == MLX5_MPESW_OP_DISABLE)
-		disable_mpesw(ldev);
+		mlx5_lag_disable_mpesw(ldev);
 unlock:
 	mutex_unlock(&ldev->lock);
 	mlx5_devcom_comp_unlock(devcom);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
index f5d9b5c97b0d..b767dbb4f457 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/mpesw.h
@@ -31,6 +31,11 @@ int mlx5_lag_mpesw_do_mirred(struct mlx5_core_dev *mdev,
 bool mlx5_lag_is_mpesw(struct mlx5_core_dev *dev);
 void mlx5_lag_mpesw_disable(struct mlx5_core_dev *dev);
 int mlx5_lag_mpesw_enable(struct mlx5_core_dev *dev);
+#ifdef CONFIG_MLX5_ESWITCH
+void mlx5_lag_disable_mpesw(struct mlx5_lag *ldev);
+#else
+static inline void mlx5_lag_disable_mpesw(struct mlx5_lag *ldev) {}
+#endif /* CONFIG_MLX5_ESWITCH */
 
 #ifdef CONFIG_MLX5_ESWITCH
 void mlx5_mpesw_speed_update_work(struct work_struct *work);
-- 
2.44.0

[PATCH net 3/5] net/mlx5: E-switch, Clear legacy flag when moving to switchdev

[Tariq Toukan]

From: Shay Drory <shayd@nvidia.com>

The cited commit introduced MLX5_PRIV_FLAGS_SWITCH_LEGACY to identify
when a transition to legacy mode is requested via devlink.  However, the
logic failed to clear this flag if the mode was subsequently changed
back to MLX5_ESWITCH_OFFLOADS (switchdev).  Consequently, if a user
toggled from legacy to switchdev, the flag remained set, leaving the
driver with wrong state indicating

Fix this by explicitly clearing the MLX5_PRIV_FLAGS_SWITCH_LEGACY bit
when the requested mode is MLX5_ESWITCH_OFFLOADS.

Fixes: 2a4f56fbcc47 ("net/mlx5e: Keep netdev when leave switchdev for devlink set legacy only")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 1b439cef3719..9d51d030596c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -4068,6 +4068,8 @@ int mlx5_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
 
 	if (mlx5_mode == MLX5_ESWITCH_LEGACY)
 		esw->dev->priv.flags |= MLX5_PRIV_FLAGS_SWITCH_LEGACY;
+	if (mlx5_mode == MLX5_ESWITCH_OFFLOADS)
+		esw->dev->priv.flags &= ~MLX5_PRIV_FLAGS_SWITCH_LEGACY;
 	mlx5_eswitch_disable_locked(esw);
 	if (mlx5_mode == MLX5_ESWITCH_OFFLOADS) {
 		if (mlx5_devlink_trap_get_num_active(esw->dev)) {
-- 
2.44.0

[PATCH net 4/5] net/mlx5: Fix missing devlink lock in SRIOV enable error path

[Tariq Toukan]

From: Shay Drory <shayd@nvidia.com>

The cited commit miss to add locking in the error path of
mlx5_sriov_enable(). When pci_enable_sriov() fails,
mlx5_device_disable_sriov() is called to clean up. This cleanup function
now expects to be called with the devlink instance lock held.

Add the missing devl_lock(devlink) and devl_unlock(devlink)

Fixes: 84a433a40d0e ("net/mlx5: Lock mlx5 devlink reload callbacks")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/sriov.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/sriov.c b/drivers/net/ethernet/mellanox/mlx5/core/sriov.c
index a2fc937d5461..172862a70c70 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/sriov.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/sriov.c
@@ -193,7 +193,9 @@ static int mlx5_sriov_enable(struct pci_dev *pdev, int num_vfs)
 	err = pci_enable_sriov(pdev, num_vfs);
 	if (err) {
 		mlx5_core_warn(dev, "pci_enable_sriov failed : %d\n", err);
+		devl_lock(devlink);
 		mlx5_device_disable_sriov(dev, num_vfs, true, true);
+		devl_unlock(devlink);
 	}
 	return err;
 }
-- 
2.44.0

[PATCH net 5/5] net/mlx5e: Fix "scheduling while atomic" in IPsec MAC address query

[Tariq Toukan]

From: Jianbo Liu <jianbol@nvidia.com>

Fix a "scheduling while atomic" bug in mlx5e_ipsec_init_macs() by
replacing mlx5_query_mac_address() with ether_addr_copy() to get the
local MAC address directly from netdev->dev_addr.

The issue occurs because mlx5_query_mac_address() queries the hardware
which involves mlx5_cmd_exec() that can sleep, but it is called from
the mlx5e_ipsec_handle_event workqueue which runs in atomic context.

The MAC address is already available in netdev->dev_addr, so no need
to query hardware. This avoids the sleeping call and resolves the bug.

Call trace:
  BUG: scheduling while atomic: kworker/u112:2/69344/0x00000200
  __schedule+0x7ab/0xa20
  schedule+0x1c/0xb0
  schedule_timeout+0x6e/0xf0
  __wait_for_common+0x91/0x1b0
  cmd_exec+0xa85/0xff0 [mlx5_core]
  mlx5_cmd_exec+0x1f/0x50 [mlx5_core]
  mlx5_query_nic_vport_mac_address+0x7b/0xd0 [mlx5_core]
  mlx5_query_mac_address+0x19/0x30 [mlx5_core]
  mlx5e_ipsec_init_macs+0xc1/0x720 [mlx5_core]
  mlx5e_ipsec_build_accel_xfrm_attrs+0x422/0x670 [mlx5_core]
  mlx5e_ipsec_handle_event+0x2b9/0x460 [mlx5_core]
  process_one_work+0x178/0x2e0
  worker_thread+0x2ea/0x430

Fixes: cee137a63431 ("net/mlx5e: Handle ESN update events")
Signed-off-by: Jianbo Liu <jianbol@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
index 9c7064187ed0..f03507a522b4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec.c
@@ -259,7 +259,6 @@ static void mlx5e_ipsec_init_limits(struct mlx5e_ipsec_sa_entry *sa_entry,
 static void mlx5e_ipsec_init_macs(struct mlx5e_ipsec_sa_entry *sa_entry,
 				  struct mlx5_accel_esp_xfrm_attrs *attrs)
 {
-	struct mlx5_core_dev *mdev = mlx5e_ipsec_sa2dev(sa_entry);
 	struct mlx5e_ipsec_addr *addrs = &attrs->addrs;
 	struct net_device *netdev = sa_entry->dev;
 	struct xfrm_state *x = sa_entry->x;
@@ -276,7 +275,7 @@ static void mlx5e_ipsec_init_macs(struct mlx5e_ipsec_sa_entry *sa_entry,
 	    attrs->type != XFRM_DEV_OFFLOAD_PACKET)
 		return;
 
-	mlx5_query_mac_address(mdev, addr);
+	ether_addr_copy(addr, netdev->dev_addr);
 	switch (attrs->dir) {
 	case XFRM_DEV_OFFLOAD_IN:
 		src = attrs->dmac;
-- 
2.44.0

Re: [PATCH net 0/5] mlx5 misc fixes 2026-02-24

[Simon Horman]

On Tue, Feb 24, 2026 at 01:46:47PM +0200, Tariq Toukan wrote:
> Hi,
> 
> This patchset provides misc bug fixes from the team to the mlx5
> core and Eth drivers.
> 
> Thanks,
> Tariq.
> 
> Jianbo Liu (1):
>   net/mlx5e: Fix "scheduling while atomic" in IPsec MAC address query
> 
> Shay Drory (4):
>   net/mlx5: DR, Fix circular locking dependency in dump
>   net/mlx5: LAG, disable MPESW in lag_disable_change()
>   net/mlx5: E-switch, Clear legacy flag when moving to switchdev
>   net/mlx5: Fix missing devlink lock in SRIOV enable error path

Thanks Tariq, all.

For the series,

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

Re: [PATCH net 0/5] mlx5 misc fixes 2026-02-24

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Tue, 24 Feb 2026 13:46:47 +0200 you wrote:
> Hi,
> 
> This patchset provides misc bug fixes from the team to the mlx5
> core and Eth drivers.
> 
> Thanks,
> Tariq.
> 
> [...]

Here is the summary with links:
  - [net,1/5] net/mlx5: DR, Fix circular locking dependency in dump
    https://git.kernel.org/netdev/net/c/2700b7e603af
  - [net,2/5] net/mlx5: LAG, disable MPESW in lag_disable_change()
    https://git.kernel.org/netdev/net/c/bd7b9f83fb9f
  - [net,3/5] net/mlx5: E-switch, Clear legacy flag when moving to switchdev
    https://git.kernel.org/netdev/net/c/d7073e8b978a
  - [net,4/5] net/mlx5: Fix missing devlink lock in SRIOV enable error path
    https://git.kernel.org/netdev/net/c/60253042c0b8
  - [net,5/5] net/mlx5e: Fix "scheduling while atomic" in IPsec MAC address query
    https://git.kernel.org/netdev/net/c/859380694f43

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