[PATCH net V3] net/mlx5: Fix deadlock between devlink lock and esw->wq

[PATCH net V3] net/mlx5: Fix deadlock between devlink lock and esw->wq

[Tariq Toukan]

From: Cosmin Ratiu <cratiu@nvidia.com>

esw->work_queue executes esw_functions_changed_event_handler ->
esw_vfs_changed_event_handler and acquires the devlink lock.

.eswitch_mode_set (acquires devlink lock in devlink_nl_pre_doit) ->
mlx5_devlink_eswitch_mode_set -> mlx5_eswitch_disable_locked ->
mlx5_eswitch_event_handler_unregister -> flush_workqueue deadlocks
when esw_vfs_changed_event_handler executes.

Fix that by no longer flushing the work to avoid the deadlock, and using
a generation counter to keep track of work relevance. This avoids an old
handler manipulating an esw that has undergone one or more mode changes:
- the counter is incremented in mlx5_eswitch_event_handler_unregister.
- the counter is read and passed to the ephemeral mlx5_host_work struct.
- the work handler takes the devlink lock and bails out if the current
  generation is different than the one it was scheduled to operate on.
- mlx5_eswitch_cleanup does the final draining before destroying the wq.

No longer flushing the workqueue has the side effect of maybe no longer
cancelling pending vport_change_handler work items, but that's ok since
those are disabled elsewhere:
- mlx5_eswitch_disable_locked disables the vport eq notifier.
- mlx5_esw_vport_disable disarms the HW EQ notification and marks
  vport->enabled under state_lock to false to prevent pending vport
  handler from doing anything.
- mlx5_eswitch_cleanup destroys the workqueue and makes sure all events
  are disabled/finished.

Fixes: f1bc646c9a06 ("net/mlx5: Use devl_ API in mlx5_esw_offloads_devlink_port_register")
Signed-off-by: Cosmin Ratiu <cratiu@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/eswitch.c  |  7 ++++---
 .../net/ethernet/mellanox/mlx5/core/eswitch.h  |  2 ++
 .../mellanox/mlx5/core/eswitch_offloads.c      | 18 +++++++++++++-----
 3 files changed, 19 insertions(+), 8 deletions(-)

V3:
- Use another approach to avoid trylock.
- Link to V2: https://lore.kernel.org/all/1769503961-124173-3-git-send-email-tariqt@nvidia.com/

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index d3af87a94a18..123c96716a54 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -1072,10 +1072,11 @@ static void mlx5_eswitch_event_handler_register(struct mlx5_eswitch *esw)
 
 static void mlx5_eswitch_event_handler_unregister(struct mlx5_eswitch *esw)
 {
-	if (esw->mode == MLX5_ESWITCH_OFFLOADS && mlx5_eswitch_is_funcs_handler(esw->dev))
+	if (esw->mode == MLX5_ESWITCH_OFFLOADS &&
+	    mlx5_eswitch_is_funcs_handler(esw->dev)) {
 		mlx5_eq_notifier_unregister(esw->dev, &esw->esw_funcs.nb);
-
-	flush_workqueue(esw->work_queue);
+		atomic_inc(&esw->esw_funcs.generation);
+	}
 }
 
 static void mlx5_eswitch_clear_vf_vports_info(struct mlx5_eswitch *esw)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 6841caef02d1..c2563bee74df 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -335,10 +335,12 @@ struct esw_mc_addr { /* SRIOV only */
 struct mlx5_host_work {
 	struct work_struct	work;
 	struct mlx5_eswitch	*esw;
+	int			work_gen;
 };
 
 struct mlx5_esw_functions {
 	struct mlx5_nb		nb;
+	atomic_t		generation;
 	bool			host_funcs_disabled;
 	u16			num_vfs;
 	u16			num_ec_vfs;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 1366f6e489bd..8c5e48d001be 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -3582,22 +3582,28 @@ static void esw_offloads_steering_cleanup(struct mlx5_eswitch *esw)
 }
 
 static void
-esw_vfs_changed_event_handler(struct mlx5_eswitch *esw, const u32 *out)
+esw_vfs_changed_event_handler(struct mlx5_eswitch *esw, int work_gen,
+			      const u32 *out)
 {
 	struct devlink *devlink;
 	bool host_pf_disabled;
 	u16 new_num_vfs;
 
+	devlink = priv_to_devlink(esw->dev);
+	devl_lock(devlink);
+
+	/* Stale work from one or more mode changes ago. Bail out. */
+	if (work_gen != atomic_read(&esw->esw_funcs.generation))
+		goto unlock;
+
 	new_num_vfs = MLX5_GET(query_esw_functions_out, out,
 			       host_params_context.host_num_of_vfs);
 	host_pf_disabled = MLX5_GET(query_esw_functions_out, out,
 				    host_params_context.host_pf_disabled);
 
 	if (new_num_vfs == esw->esw_funcs.num_vfs || host_pf_disabled)
-		return;
+		goto unlock;
 
-	devlink = priv_to_devlink(esw->dev);
-	devl_lock(devlink);
 	/* Number of VFs can only change from "0 to x" or "x to 0". */
 	if (esw->esw_funcs.num_vfs > 0) {
 		mlx5_eswitch_unload_vf_vports(esw, esw->esw_funcs.num_vfs);
@@ -3612,6 +3618,7 @@ esw_vfs_changed_event_handler(struct mlx5_eswitch *esw, const u32 *out)
 		}
 	}
 	esw->esw_funcs.num_vfs = new_num_vfs;
+unlock:
 	devl_unlock(devlink);
 }
 
@@ -3628,7 +3635,7 @@ static void esw_functions_changed_event_handler(struct work_struct *work)
 	if (IS_ERR(out))
 		goto out;
 
-	esw_vfs_changed_event_handler(esw, out);
+	esw_vfs_changed_event_handler(esw, host_work->work_gen, out);
 	kvfree(out);
 out:
 	kfree(host_work);
@@ -3648,6 +3655,7 @@ int mlx5_esw_funcs_changed_handler(struct notifier_block *nb, unsigned long type
 	esw = container_of(esw_funcs, struct mlx5_eswitch, esw_funcs);
 
 	host_work->esw = esw;
+	host_work->work_gen = atomic_read(&esw_funcs->generation);
 
 	INIT_WORK(&host_work->work, esw_functions_changed_event_handler);
 	queue_work(esw->work_queue, &host_work->work);

base-commit: ae779bcb18cb0ef0da1402b9dd837e2084e23e27
-- 
2.44.0

Re: [PATCH net V3] net/mlx5: Fix deadlock between devlink lock and esw->wq

[patchwork-bot+netdevbpf]

Hello:

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

On Thu, 5 Mar 2026 10:10:19 +0200 you wrote:
> From: Cosmin Ratiu <cratiu@nvidia.com>
> 
> esw->work_queue executes esw_functions_changed_event_handler ->
> esw_vfs_changed_event_handler and acquires the devlink lock.
> 
> .eswitch_mode_set (acquires devlink lock in devlink_nl_pre_doit) ->
> mlx5_devlink_eswitch_mode_set -> mlx5_eswitch_disable_locked ->
> mlx5_eswitch_event_handler_unregister -> flush_workqueue deadlocks
> when esw_vfs_changed_event_handler executes.
> 
> [...]

Here is the summary with links:
  - [net,V3] net/mlx5: Fix deadlock between devlink lock and esw->wq
    https://git.kernel.org/netdev/net/c/aed763abf0e9

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