2020-03-18 16:45:34

by Niklas Schnelle

Subject: [RFC 1/1] net/mlx5: Fix failing fw tracer allocation on s390

On s390 FORCE_MAX_ZONEORDER is 9 instead of 11, thus a larger kzalloc()
allocation as done for the firmware tracer will always fail.

Looking at mlx5_fw_tracer_save_trace(), it is actually the driver itself
that copies the debug data into the trace array and there is no need for
the allocation to be contiguous in physical memory. We can therefor use
kvzalloc() instead of kzalloc() and get rid of the large contiguous
allcoation.

Signed-off-by: Niklas Schnelle <[email protected]>
---
drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
index 94d7b69a95c7..eb2e57ff08a6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
@@ -935,7 +935,7 @@ struct mlx5_fw_tracer *mlx5_fw_tracer_create(struct mlx5_core_dev *dev)
return NULL;
}

- tracer = kzalloc(sizeof(*tracer), GFP_KERNEL);
+ tracer = kvzalloc(sizeof(*tracer), GFP_KERNEL);
if (!tracer)
return ERR_PTR(-ENOMEM);

@@ -982,7 +982,7 @@ struct mlx5_fw_tracer *mlx5_fw_tracer_create(struct mlx5_core_dev *dev)
tracer->dev = NULL;
destroy_workqueue(tracer->work_queue);
free_tracer:
- kfree(tracer);
+ kvfree(tracer);
return ERR_PTR(err);
}

@@ -1061,7 +1061,7 @@ void mlx5_fw_tracer_destroy(struct mlx5_fw_tracer *tracer)
mlx5_fw_tracer_destroy_log_buf(tracer);
flush_workqueue(tracer->work_queue);
destroy_workqueue(tracer->work_queue);
- kfree(tracer);
+ kvfree(tracer);
}

static int fw_tracer_event(struct notifier_block *nb, unsigned long action, void *data)
--
2.17.1

2020-03-18 17:15:35

by Leon Romanovsky

Subject: Re: [RFC 1/1] net/mlx5: Fix failing fw tracer allocation on s390

On Wed, Mar 18, 2020 at 05:44:31PM +0100, Niklas Schnelle wrote:
> On s390 FORCE_MAX_ZONEORDER is 9 instead of 11, thus a larger kzalloc()
> allocation as done for the firmware tracer will always fail.
>
> Looking at mlx5_fw_tracer_save_trace(), it is actually the driver itself
> that copies the debug data into the trace array and there is no need for
> the allocation to be contiguous in physical memory. We can therefor use
> kvzalloc() instead of kzalloc() and get rid of the large contiguous
> allcoation.
>
> Signed-off-by: Niklas Schnelle <[email protected]>
> ---
> drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
> index 94d7b69a95c7..eb2e57ff08a6 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
> @@ -935,7 +935,7 @@ struct mlx5_fw_tracer *mlx5_fw_tracer_create(struct mlx5_core_dev *dev)
> return NULL;
> }
>
> - tracer = kzalloc(sizeof(*tracer), GFP_KERNEL);
> + tracer = kvzalloc(sizeof(*tracer), GFP_KERNEL);

Saeed, Moshe

Can we use vzalloc() instead of kvzalloc() here?

Thanks