lkml.org 
[lkml]   [2021]   [Mar]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH] net/mlx5e: allocate 'indirection_rqt' buffer dynamically
From
Date


On 3/8/2021 5:32 PM, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@arndb.de>
>
> Increasing the size of the indirection_rqt array from 128 to 256 bytes
> pushed the stack usage of the mlx5e_hairpin_fill_rqt_rqns() function
> over the warning limit when building with clang and CONFIG_KASAN:
>
> drivers/net/ethernet/mellanox/mlx5/core/en_tc.c:970:1: error: stack frame size of 1180 bytes in function 'mlx5e_tc_add_nic_flow' [-Werror,-Wframe-larger-than=]
>
> Using dynamic allocation here is safe because the caller does the
> same, and it reduces the stack usage of the function to just a few
> bytes.
>
> Fixes: 1dd55ba2fb70 ("net/mlx5e: Increase indirection RQ table size to 256")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
> drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 16 +++++++++++++---
> 1 file changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> index 0da69b98f38f..66f98618dc13 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
> @@ -445,12 +445,16 @@ static void mlx5e_hairpin_destroy_transport(struct mlx5e_hairpin *hp)
> mlx5_core_dealloc_transport_domain(hp->func_mdev, hp->tdn);
> }
>
> -static void mlx5e_hairpin_fill_rqt_rqns(struct mlx5e_hairpin *hp, void *rqtc)
> +static int mlx5e_hairpin_fill_rqt_rqns(struct mlx5e_hairpin *hp, void *rqtc)
> {
> - u32 indirection_rqt[MLX5E_INDIR_RQT_SIZE], rqn;
> + u32 *indirection_rqt, rqn;
> struct mlx5e_priv *priv = hp->func_priv;
> int i, ix, sz = MLX5E_INDIR_RQT_SIZE;
>
> + indirection_rqt = kzalloc(sz, GFP_KERNEL);
> + if (!indirection_rqt)
> + return -ENOMEM;
> +
> mlx5e_build_default_indir_rqt(indirection_rqt, sz,
> hp->num_channels);
>
> @@ -462,6 +466,9 @@ static void mlx5e_hairpin_fill_rqt_rqns(struct mlx5e_hairpin *hp, void *rqtc)
> rqn = hp->pair->rqn[ix];
> MLX5_SET(rqtc, rqtc, rq_num[i], rqn);
> }
> +
> + kfree(indirection_rqt);
> + return 0;
> }
>
> static int mlx5e_hairpin_create_indirect_rqt(struct mlx5e_hairpin *hp)
> @@ -482,12 +489,15 @@ static int mlx5e_hairpin_create_indirect_rqt(struct mlx5e_hairpin *hp)
> MLX5_SET(rqtc, rqtc, rqt_actual_size, sz);
> MLX5_SET(rqtc, rqtc, rqt_max_size, sz);
>
> - mlx5e_hairpin_fill_rqt_rqns(hp, rqtc);
> + err = mlx5e_hairpin_fill_rqt_rqns(hp, rqtc);
> + if (err)
> + goto out;
>
> err = mlx5_core_create_rqt(mdev, in, inlen, &hp->indir_rqt.rqtn);
> if (!err)
> hp->indir_rqt.enabled = true;
>
> +out:
> kvfree(in);
> return err;
> }
>

Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Thanks for your patch.

Tariq

\
 
 \ /
  Last update: 2021-03-08 17:30    [W:0.072 / U:0.192 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site